Proceedings of Réanimation 2020, the French Intensive Care Society International Congress

Acknowledgements

This abstract book was edited and corrected by the members of the Congress Committee of the French Intensive Care Society:

• Frédéric Pène (Paris)

• Hafid Ait-Oufella (Paris)

• Pierre Asfar (Angers)

• Cécile Aubron (Brest)

• Emmanuel Canet (Nantes)

• Guillaume Carteaux (Créteil)

• Stephan Ehrmann (Tours)

• Jean-Pierre Frat (Poitiers)

• Guillaume Geri (Boulogne-Billancourt)

• Julie Helms (Strasbourg)

• Alexandre Demoule (Paris)

• Saad Nseir (Lille)

• Mehdi Oualha (Paris)

Oral Communications

COK-1 Expiratory muscle weakness quantified by maximal expiratory pressure may be insufficient in predicting critical outcomes in mechanically ventilated patients

Yann Combret¹, Guillaume Prieur¹, Francis-Edouard Gravier², Tristan Bonnevie², Olivier Contal³, Bouchra Lamia¹, Clement Medrinal¹

¹Groupe Hospitalier du Havre, Le Havre, France; ²ADIR Association, Rouen, Fs; ³Haute Ecole de Santé Vaud, Lausanne, Switzerland

Correspondence: Yann Combret (yann.combret@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):COK-1

Rationale: Expiratory muscles has recently been stated as the «neglected component» in mechanically ventilated patient. Several authors stated these muscles importance in cough capacity, contractile efficiency of the diaphragm or reduction of hyperinflation. However, few studies reported potential factors leading to expiratory muscle weakness and its importance on weaning success or survival after mechanical ventilation.

Patients and methods: This study is a secondary analysis of our previously described cohort of 124 patients ventilated for at least 24 h assessed for respiratory muscles function. Maximal expiratory pressure (MEP) measurement was carried out during spontaneous breathing trial using a manometer with an unidirectional valve. MEP diagnostic accuracy to predict ICU-AW (ICU acquired weakness), weaning success and sursvival within 30 days were assessed using expiratory muscle strength as absolute values (cmH₂O), as %predicted values and as %lower limit of normal.

Results: Due to the paucity of data reporting threshold value for expiratory muscle weakness, we considered our median value (47 cmH₂O (IQR 44)) as the threshold value for expiratory muscle weakness group (MEP ≤ 47 cmH₂O) and normal expiratory muscle group (MEP > 47 cmH₂O). Patients with low MEP received more catecholamines (p = 0.04) and a higher duration of mechanical ventilation (p = 0.001). Inversely, higher body mass index was associated with higher MEP. Patients with low MEP presented more ICU-AW compared to normal MEP patients (64% vs. 35%; p = 0.003). No other outcomes were different between groups. MEP was statistically able to predict ICU-AW but area under (AUC) receiving operating curves showed weak predictive ability (AUC: 0.66 (95% IC 0.55–0.77; p < 0.01) for a threshold value ≤ 49 cmH₂O. Expiratory muscle weakness was unable to predict critical outcomes when adjusting MEP to the %predicted or lower limit of normal.

Discussion: Possible explanation is that contrary to inspiratory muscle weakness, cough inefficacy after weaning from mechanical ventilation could be managed with cough supplementation techniques (i.e. mechanical in-exsufflation).

Conclusion: In our cohort, MEP was not associated with mechanical ventilation weaning or death. Despite our results, different clinical techniques for quantifying expiratory muscle weakness may provide more beneficial results.

Compliance with ethics regulations: Yes

Table 1 Patients’ outcomes

COK-2 Consequences of ICU acquired weakness: a systematic review and meta-analysis

Clement Medrinal¹, Yann Combret², Roger Hilfiker³, Nadine Aroichane⁴, Tristan Bonnevie⁵, Francis-Edouard Gravier⁵, Guillaume Prieur¹, Olivier Contal⁶, Bouchra Lamia⁷

¹Le Havre Hospital, ICU Department. Normandie University, UNIROUEN, UPRES EA3830-GRHV, Institute for Research and Innovation in Biomedicine (IRIB), Le Havre, France; ²Le Havre Hospital, ICU Department. Research and Clinical Experimentation Institute (IREC), Pulmonology, ORL and Dermatology, Louvain Catholic University, Brussels, Le Havre, France; ³University of Applied Sciences and Arts Western Switzerland Valais (HES-SO Valais-Wallis), School of Health Sciences, Leukerbad, Leukerbad, Switzerland; ⁴IFMK de Rouen, Rouen, France; ⁵ADIR Association Normandie University, UNIROUEN, UPRES EA3830-GRHV, Institute for Research and Innovation in Biomedicine (IRIB), Bois Guillaume, France; ⁶University of Applied Sciences and Arts Western Switzerland (HES-SO), Lausanne, Switzerland; ⁷Le Havre Hospital, Pulmonology Department. Normandie University, UNIROUEN, UPRES EA3830-GRHV, Institute for Research and Innovation in Biomedicine (IRIB), Le Havre, France

Correspondence: Clement Medrinal (medrinal.clement.mk@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):COK-2

Rationale: It is increasingly recognized that identification of muscle weakness in ICU should be as early as possible. Unfortunately, volitional tests can be assessed in a minority of patients who are awake and cooperative. Some studies proposed alternative assessments conductable without patient cooperation such as magnetic phrenic nerve stimulation or muscular ultrasonography. However, clinical relevance of these tests are not determined and further correlation with muscle strength and functional outcomes is needed in a large representative population of critically ill patients. We undertook a systematic review and meta-analysis to estimate the clinical relevance and association between the different ICU-Aw diagnostic tests and critical outcomes such as mortality or weaning failure.

Patients and methods: We built our systematic review using the Preferred Reported Items for Systematic Review. Literature research was conducted in five databases (PubMed, EMBASE, CINAHL, Cochrane library, Science Direct). Search terms combined keywords relative to three domains: ICU-Aw, diagnostic test of ICU-Aw and outcomes. All observational studies published between January 2000 and December 2018 were included. Randomised controlled trial were excluded. Reviewers abstracted study data by using a standardized form and assessed quality by using Quality in Prognosis Studies tool and the Quality Assessment of Diagnostic Accuracy Studies criteria.

Results: 60 studies were analyzed including 4382 patients. 23 studies (1390 patients) performed diaphragm ultrasound, 6 (292 patients) performed transdiaphragmatic twitch pressure and 13 (709 patients) performed maximal inspiratory pressure (MIP). 14 studies (1656 patients) performed MRC score and 4 (335 patients) performed Handgrip test. Overall, ICU-Aw prevalence was 47%. ICU-Aw significantly increased the odds of global mortality (OR to 3.2 95% IC (2.3–4.4); p < 0.0001). The best predictive capacity of global mortality was obtained using the transdiaphragmatic twich pressure with a pooled sensitivity of 0.87 95% IC (0.76–0.93) and a pooled specificity of 0.36 95% IC (0.27–0.43) and an AUC of 0.74 95% IC (0.70–0.78). ICU-Aw was also associated with weaning failure (OR to 3.06 95% IC (2.57–3.63)s; p < 0.001). Diaphragm thickening fraction showed the best predictive capacity for weaning failure with a pooled sensitivity to 0.76 95% IC (0.67–0.83) and a pooled specificity to 0.86 95% IC (0.78–0.92) and an AUC to 0.86 95% IC (0.83–0.89).

Conclusion: ICU-Aw is highly prevalent and is a serious problem associated with a higher mortality and a higher mechanical ventilation weaning failure. Respiratory muscle evaluation should be the most important muscular assessment in ICU patients.

Compliance with ethics regulations: Yes

COK-3 Accuracy of diaphragm ultrasound to predict weaning outcome during spontaneous breathing trial: a meta-analysis

Aymeric Le Neindre¹, Francois Philippart¹, Marta Luperto², Johan Wormser¹, Johanna Morel³, Serge Aho², Silvia Mongodi⁴, Francesco Mojoli⁴, Belaid Bouhemad²

¹Groupe Hospitalier Paris Saint-Joseph, Paris, France; ²CHU de Dijon, Dijon, France; ³CHU Henri Mondor, Créteil, France; ⁴Fondazione IRCCS Policlinico San Matteo, Pavia, Italy

Correspondence: Aymeric Le Neindre (aymeric.leneindre@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):COK-3

Rationale: Many studies have demonstrated the potential of diaphragm ultrasound (DUS) to predict weaning outcome but no comprehensive conclusion has been reached. The aim of this meta-analysis was to assess diagnostic accuracy of DUS for the prediction of weaning failure in critically ill patients.

Patients and methods: PubMed, Science direct, Cochrane Library, CENTRAL and MEDLINE were searched. Three investigators independently selected studies meeting inclusion criteria, extracted data and performed bias analysis using the Quality Assessment of Diagnostic Accuracy Studies-2 instrument. The bivariate model was used to estimate the pooled results of sensitivity, specificity and diagnostic odds ratio. Sources of heterogeneity were explored and Bayesian analysis was used to confirm the estimates.

Results: Twenty-three studies with a total were selected and included in the meta-analysis. The pooled sensitivity, specificity and Diagnostic Odd ratio (DOR) of Diaphragm Thickening Fraction were 0.72, 0.88 and 18.82. The Diaphragm Excursion results was lower with pooled sensitivity, specificity and DOR of 0.73, 0.81 and 12.0, respectively. Important heterogeneity was observed among the studies. High risk of bias in selection of patients and conditions of measure was highlighted. A Bayesian analysis confirms the global estimation of diagnostic performance.

Conclusion: Diaphragm ultrasound provides good diagnostic performance to predict weaning outcome. Nevertheless, heterogeneity in studies and recent results suggest that DUS may currently be used to explain weaning failure rather than predict weaning outcome.

Compliance with ethics regulations: NA

CO-01 Risk factors for candidemia: a prospective matched case–control study

Julien Poissy¹, Lauro Damonti², Anne Bignon³, Nina Khanna⁴, Matthias Von Kietzell⁵, Katia Boggian⁵, Dionysios Neofytos⁶, Fanny Vuotto⁷, Valerie Coiteux⁸, Florent Artru⁹, Stefan Zimmerli², Jean-Luc Pagani¹⁰, Thierry Calandra¹¹, Boualem Sendid¹², Daniel Poulain¹², Christian Van Delden⁶, Frédéric Lamoth¹³, Oscar Marchetti¹¹, Pierre-Yves Bochud¹¹

¹Intensive Care Department, University hospital of Lille, Lille, France; ²Department of Infectious Diseases, Inselspital, Bern University Hospital, University of Bern, Bern, Switzerland; ³Surgical Intensive Care Unit. University hospital of Lille, Lille, France; ⁴Division of Infectious Diseases and Hospital Epidemiology, University Hospital of Basel, Basel, Switzerland; ⁵Department of Infectious Diseases, Cantonal Hospital of Saint Gallen, Saint Gallen, Switzerland; ⁶Infectious diseases, University Hospital of Geneva, Geneva, Switzerland; ⁷Infectious Diseases Department, University hospital of Lille, Lille, France; ⁸Hematological Disorders Department, University Hospital of Lille, Lille, France; ⁹Digestive Intensive Care Department, University Hospital of Lille, Lille, France; ¹⁰Intensive Care Department, University hospital of Lausanne, Lausanne, Switzerland; ¹¹Infectious Diseases Department, University Hospital of Lausanne, Lausanne, Switzerland; ¹²Laboratory of Mycology and Parasitology, University Hospital of Lille, Lille, France; ¹³Microbiology Institute, University Hospital of Lausanne, Lausanne, Switzerland

Correspondence: Julien Poissy (julien_poissy@hotmail.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-01

Rationale: Candidemia is an opportunistic infection associated with high morbidity and mortality in hospitalized patients, both inside and outside intensive care units (ICUs). Identification of patients at risk for preemptive approach and early detection is crucial. Prospective control-matched studies and comparison between ICU and non-ICU patients are lacking in this field. We aim to identify and compare specific risk factors in ICU and non-ICU patients.

Patients and methods: This was a prospective multicenter matched case–control study assessing risk factors for candidemia and death in candidemic patients, both outside and inside ICUs, from 6 teaching hospitals. Controls were matched to cases based on age, hospitalization ward, hospitalization duration and, when applicable, type of surgery. Risk factors were analyzed by univariate and multivariate conditional regression models, as a basis for a new scoring system to predict candidemia.

Results: The study included 192 cases and 411 matched controls. 44% were hospitalized inside ICU and 56% outside. Independent risk factors for candidemia within the ICU population included total parenteral nutrition (TPN) (OR = 6.75, p < 0.001), acute kidney injury (OR = 4.77, p < 0.001), heart disease (OR = 3.78, p = 0.006), previous septic shock (OR = 2.39, p = 0.02) and exposure to aminoglycosides (OR = 2.28, p = 0.05). Independent risk factors for candidemia within the non-ICU population included central venous catheter (CVC) (OR = 9.77, p < 0.001), TPN (OR = 3.29, p = 0.003), exposure to glycopeptides (OR = 3.31, p = 0.04), and to nitroimidazoles (OR = 3.12, p = 0.04).

The weighted scores and their ROC curves are presented in Fig. 1. The weighted ICU-score was as follows: TPN, + 2.5; AKI, + 1.5; heart disease, + 1.5; previous septic shock, + 1.0; aminoglycosides, + 1.0. AUC of the ROC curve was 0.768. The optimal cut-off was ≥ 4 (sensitivity = 69%, specificity = 70%). The best cut-off to optimize specificity was ≥ 5 (sensitivity = 43%, specificity = 88%). The weighted non-ICU score was as follows: CVC, + 2.5; nitroimidazole: + 1.0; TPN, + 1.0; Glycopeptide: + 1.0. AUC of the ROC curve was 0.717. The optimal cut-off was ≥ 2 (sensitivity = 83%, specificity = 50%). The best cut-off to optimize specificity was ≥ 4 (sensitivity = 51%, specificity = 81%).

Independent factors for death in candidemic ICU patients were septic shock (OR = 4.09, p = 0.003), acute kidney injury (OR = 3.45, p = 0.02), the number of antibiotics to which patients were exposed before candidemia (OR = 1.37 per unit, p = 0.02). For non-ICU patients, acute kidney injury (OR = 11.9, p = 0.002) and septic shock (OR = 8.70, p = 0.002) were the only variables significantly associated with death.

Conclusion: Risk factors for candidemia differ between ICU and non-ICU settings, including different patterns of antibiotic exposure, leading to different weighted scores predictive of candidemia, with better performances for ICU patients.

Compliance with ethics regulations: Yes

Fig. 1

Risk scores for candidemia

CO-02 A stronger signal of ART-123 efficacy in patients with sepsis associated coagulopathy enrolled into the SCARLET Trial in France

Bruno François¹, Maud Fiancette², Julie Helms³, Emmanuelle Mercier⁴, Jean-Baptiste Lascarrou⁵, Xavier Wittebole⁶, Amanda Radford⁷, Toshihiko Kayanoki⁷, Kosuke Tanaka⁷, David Fineberg⁷, Jean-Louis Vincent⁸

¹Réanimation polyvalente/Inserm CIC 1435/Inserm UMR 1092, CHU Dupuytren, Limoges, France; ²Médecine Intensive Réanimation, Centre Hospitalier Départemental Vendée, La Roche-Sur-Yon, France; ³Université de Strasbourg (UNISTRA), Faculté de Médecine; Hôpitaux universitaires de Strasbourg, Médecine Intensive-Réanimation, Nouvel Hôpital Civil, Strasbourg, France; ⁴Médecine Intensive Réanimation/CRICS-TRIGGERSEP, CHRU de Tours, Tours, France; ⁵Service de Médecine Intensive Réanimation, CHU de Nantes, Nantes, France; ⁶Department of Critical Care Medicine, Clinique universitaire St Luc, Université Catholique de Louvain, Brussels, Belgium; ⁷Asahi Kasei Pharma America, Waltham, UNITED STATES; ⁸Department of Intensive Care, Erasme Hospital, Universite Libre de Bruxelles, Brussels, Belgium

Correspondence: Bruno François (b.francois@unilim.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-02

Rationale: To assess the effect of recombinant human soluble thrombomodulin (ART-123) treatment on 28-day all-cause mortality in patients with sepsis associated coagulopathy (SAC) enrolled in France, the highest enrolling country, as compared to the global population of patients in the randomized, double-blinded, placebo-controlled, SCARLET phase 3 study (NCT01598831), and especially to evaluate factors that may have contributed to greater mortality reduction in the French patients.

Patients and methods: Among the global patient population with SAC defined as an international normalized ratio (INR) > 1.40, platelet count > 30– < 150 × 109 or > 30% decrease in 24 h, and with concomitant cardiovascular and/or respiratory failure, patients in France were analyzed. All subjects were treated with 0.06 mg/kg/day of intravenous ART-123 (n = 75/395) or placebo (n = 74/405) for up to 6 days in addition to standard of care.

Results: In the SCARLET trial, ART-123 did not significantly reduce 28-day all-cause mortality (800 patients from 149 sites in 26 countries) with an absolute risk reduction (ARR) of 2.55% (P = 0.32) in ART-123 treated patients with SAC, and 5.4% in those patients who maintained protocol specified coagulopathy (INR > 1.40, platelet count > 30 × 109) at baseline prior to study drug dosing. In France, there were 149/800 (18.6%) patients enrolled at 16/149 (10.7%) sites. Considering all randomized and dosed patients in France as compared to those patients that maintained protocol specified coagulopathy, the 28-day ARR for ART-123 was higher in France than for the global population. Patients enrolled in France were more likely to have the protocol specified coagulopathy criteria at baseline than in other countries. They also had overall more severe baseline disease as measured by the number of patients with at least 3 organ dysfunctions (cardiovascular, respiratory, renal, and hepatic) and a higher APACHE II score than the other patients. However, they were less likely to be treated by renal replacement therapy (RRT) or to receive heparin at baseline. In France there was a greater proportion of enrollment from individual sites enrolling 6 patients or more.

Conclusion: The trend towards a higher difference of mortality benefit in patients treated with ART-123 and enrolled in France may be attributed to better patient selection, and a higher rate of patients meeting the protocol specified coagulopathy criteria at baseline by primarily high enrolling research sites.

Compliance with ethics regulations: Yes

CO-03 Use and impact of aminoglycoside empirical therapy in extended spectrum beta-lactamase enterobacteriaceae bloodstream infections in intensive care unit

Lucie Benetazzo, Pierre-Yves Delannoy, Olivier Leroy, Olivier Robineau, Agnes Meybeck

CH Tourcoing, Tourcoing, France

Correspondence: Lucie Benetazzo (lucie.benetazzo@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-03

Rationale: Aminoglycosides are prescribed in severe infections for bactericidal activity and broadening of the spectrum. As the prevalence of extended-spectrum beta-lactamase-producing enterobacteriaceae (ESBLE) increases, aminoglycosides may be interesting for their treatment. However, their use is limited by their toxicity, especially renal. Our study evaluated the impact of an aminoglycoside in empirical treatment of ESBLE bloodstream infection in intensive care unit (ICU).

Patients and methods: Between January 2011 and September 2017, patients treated for ESBLE bacteraemia in the ICU of 5 French hospitals from Hauts de France were included in a retrospective observational cohort study. In order to evaluate the impact of the empirical prescription of an aminoglycoside, a bivariate and multivariate analysis were performed. The primary endpoint was mortality on day 30. Secondary endpoints were empirical antibiotic therapy adequacy and renal failure rates.

Results: Three hundred and seven patients were included, 169 received an aminoglycoside as initial treatment. The death rate at day 30 was 40%. We did not find any difference in mortality between aminoglycoside and non-aminoglycoside group (43.4% vs. 39.3%, p = 0.545). Renal impairment occurred in aminoglycoside and non-aminoglycoside groups in 20.7% and 23.9%, respectively (p = 0.59). The adequacy rate of empirical antibiotic therapy was higher in the aminoglycoside group (91.7% vs. 77%, p = 0.001). An age greater than 70 years, a history of transplantation, or the nosocomial origin of bacteraemia were associated with mortality at day 30. Maintenance of amines more than 48 h after bacteraemia, occurrence of ARDS or acute renal failure also increased mortality on day 30.

Conclusion: Our study did not show any impact of aminoglycoside empirical prescription for the treatment of ESBLE bacteraemia even if it led to an increase in the adequacy rate of empirical therapy. We did not find any renal toxicity caused by aminoglycosides.

Compliance with ethics regulations: Yes

CO-04 Impact of a restrictive antibiotic policy on the emergence of extended-spectrum ß-lactamase producing Enterobacteriaceae (ESBL-E) in the ICU. A quasi-experimental observational study

Christophe Le Terrier¹, Marco Vinetti¹, Régine Richard¹, Bruno Jarrige², Sébastien Breurec³, Michel Carles¹, Guillaume Thiery¹

¹CHU Guadeloupe Intensive Care Unit, Pointe-À-Pitre, France; ²CHU Guadeloupe Hospital Infection Control Department, Pointe-À-Pitre, France; ³CHU Guadeloupe Microbiology unit, Pointe-À-Pitre, France

Correspondence: Christophe Le Terrier (christophe.clt@orange.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-04

Rationale: Massive consumption of antibiotics in the intensive care unit (ICU) is a major determinant of extended-spectrum beta-lactamase–producing Enterobacteriaceae (ESBL-E) spreading. We evaluated whether a stewardship program including restrictive antibiotic policy in the ICU would reduce ESBL-E emergence without worsening patients’ outcomes.

Patients and methods: We conducted an observational quasi-experimental pre-post intervention study of all consecutive patients with length of stay (LOS) superior to 48 h in the medical-surgical ICU of University Hospital of Guadeloupe. From January 1, 2014 to December 31, 2014, a liberal strategy was used including a broad-spectrum antibiotic as initial empirical treatment in case of sepsis or suspected infection, followed by de-escalation after 48–72 h. From January 1, 2015 to December 31, 2015, a restrictive strategy was adopted which consisted of limitation of broad-spectrum antibiotics, avoidance of antibiotics targeting anaerobic microbiota and shortening of antibiotic duration. In addition, antibiotic therapy was initiated only after microbiological identification, except in cases of septic shock, acute respiratory distress syndrome and meningitis, where an empiric therapy was started immediately after the microbiological samples were taken. Our primary outcome was the incidence of ICU-acquired ESBL-E and the main secondary outcome were all-cause ICU mortality and the rate of ESBL-E infections.

Results: 1009 and 1067 patients were admitted to ICU during the liberal and the restrictive strategy period of the study respectively. Among them, 767 and 826 patients were hospitalized > 48 h and were enrolled in the study (Table 1). During the restrictive strategy period, less patients were treated with antibiotic therapy (41 vs 52%; p < 0.001), treatment duration was shorter (5 vs 6 days; p = 0.01) and antibiotics targeting anaerobic pathogens were significantly less administrated (87.1% vs 37.5%; p < 0.0001). The rate of ICU-acquired ESBL-E carriage was significantly lower during the restrictive strategy period (18.9% vs 11.1%; p < 0.0001). Similarly, ICU-acquired ESBL-E infection rate and ICU mortality were lower during the restrictive strategy period. In multivariate analysis, the length of stay in the ICU, the number of antibiotic adminsitered and the restrictive strategy period were independently associated with a lower rate of ESBL-E acquisition.

Conclusion: In a large cohort of consecutive ICU patients, a stewardship program including a restrictive antibiotic strategy has proven effective in terms of reduction of antibiotic consumption, especially broad spectrum antibiotics and those targeting anaerobic microbiota. This strategy was associated with a lower rate of ESBL-E acquisition without worsening patients’ outcomes.

Compliance with ethics regulations: Yes.

Table 1 Demographic characteristics, outcomes and antibiotics used

Abbreviations: SAPS II: Simplified Acute Physiology Score II; ICU: Intensive care unit; ESBL-E: extended-spectrum beta-lactamase-producing Enterobacteriaceae; SD: standard deviation

C1G: 1^st generation cephalosporin; C2G: 2^nd generation cephalosporin; C3G: 3^rd generation cephalosporin

^*Amoxicillin/Clavulanic acid, Piperacillin/Tazobactam, Carbapenem, Cefoxitin, Clindamycin, Metronidazole

# when not specified, results are n (%). In bold, p values < 0.05.

CO-05 Awake venoarterial extracorporeal membrane oxygenation in cardiogenic shock: a propensity score matched analysis

Santiago Montero¹, Florent Huang², Juliette Chommeloux², Nicolas Brechot², Pierre Demondion³, Guillaume Franchineau², Guillaume Hekimian², Romain Persichini⁴, Charles-Edouard Luyt², Guillaume Lebreton³, Alain Combes², Matthieu Schmidt²

¹Hospital Germans Trias i Pujol, Acute Cardiovascular Care Unit, Cardiology, Universitat Autònoma de Barcelona, Barcelona, Spain; ²Assistance Publique–Hôpitaux de Paris, Pitié–Salpêtrière Hospital, Medical Intensive Care Unit, 75651 Paris Cedex 13, France; ³Assistance Publique–Hôpitaux de Paris, Pitié–Salpêtrière Hospital, Thoracic and Cardiovascular department, 75651 Paris Cedex 13, France; ⁴Medical–Surgical Intensive Care Unit, CHU de La Réunion, Felix-Guyon Hospital, Saint Denis, La Réunion, France

Correspondence: Santiago Montero (monteroaradas@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-05

Rationale: Venoarterial extracorporeal membrane oxygenation (VA-ECMO) is the first-line therapy for refractory cardiogenic shock (CS), but its applicability is undermined by the high morbidity associated with its complications, especially those related to mechanical ventilation (MV). We aimed at assessing the impact on survival of keeping patients awake during the VA-ECMO run in the context of refractory CS.

Patients and methods: A 7-year database of CS-patients supported with peripheral VA-ECMO was used to perform a propensity score (PS) matched analysis in order to balance their clinical profile. Patients were classified as “awake and partially awake” or “non-awake” if mechanical ventilation was present ≤ 50% or > 50% of the ECMO run. Primary outcomes were 60-day and 1-year mortality, and secondary outcomes included rates of ventilator-associated pneumonia (VAP) and ECMO-related complications. A multivariate logistic regression analysis was performed to identify if respiratory status at cannulation was independently associated with 60-day mortality.

Results: Out of 231 patients included, 91 (39%) were “awake and partially awake” and 140 (61%) “non-awake”. After PS matching adjustment, the “awake and partially awake” group had significantly better 60-day (19% vs 46%, p < 0.006; 95% CI OR 0.36 [0.17–0.79], p = 0.01) and 1-year survival (32% vs 57%, p < 0.018; 95% CI OR 0.43 [0.22–0.82], p = 0.01) compared to the “non-awake” group, as well as reduced rates of VAP (34% vs 64%, p = 0.004) and less antibiotic and sedative drugs consumption. However, MV at ECMO cannulation was not independently related to 60-day mortality.

Discussion: Previous clinical series have suggested better survival and less VAP rates in ECLS-supported patients, but the small number of patients and the inclusion of non-arousable patients with likely neurological impairment hampered the interpretation and validity of this strategy. Although the likelihood of remaining awake seem to increase in vigil patients at cannulation, we did not find an independent association between this fact and 60-day mortality. Therefore, it is likely that the earliest possible extubation represents the main message in this setting, allowing to longer time on spontaneous breathing whilst on VA-ECMO support. Beyond better outcome, physical rehabilitation, communication with relatives and, specifically, interactive information of the medical decisions, are key potential benefits from a non-intubation or an early-extubation management in these patients. Indeed, the avoidance of relevant complications (mainly VAP) may additionally contribute to these advantages.

Conclusion: An “awake and partially awake” VA-ECMO strategy in CS is safe and is associated with improved short- and long-term survival compared to mechanically ventilated patients.

Compliance with ethics regulations: Yes.

Fig. 1

Kaplan-Meier curves at 60-day and 1-year according to mechanical ventilation strategy

CO-06 Impact of pulmonary hypertension on post heart transplant outcome

Cyrielle Desnos¹, Guillaume Coutance¹, Mathieu Kerneis¹, Amandine Baptiste², Guillaume Lebreton³, Charles-Edouard Luyt⁴, Alain Combes⁴, Shaida Varnous¹, Nicolas Brechot⁴

¹Cardiology department, La Pitié Salpêtrière hospital, Paris, France; ²Biostatistics and public health department, La Pitié Salpêtrière hospital, Paris, France; ³Cardio-thoracic surgery department, La Pitié Salpêtrière hospital, Paris, France; ⁴Intensive care unit, Cardiology institute, La Pitié Salpêtrière hospital, Paris, France

Correspondence: Cyrielle Desnos (cyrielle.desnos@wanadoo.f)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-06

Rationale: Pulmonary hypertension (PH) is associated with a higher risk of early right ventricular graft failure after heart transplantation, and poorer short and long-term outcomes. However, improvements in short term extra corporeal life support in the past decades allows an easier supply of early graft failure. Thereby, we aimed to compare the outcome of patients undergoing heart transplantation with or without elevated pulmonary vascular resistances (PVR), in an expert center for mechanical circulatory support.

Patients and methods: We conducted a retrospective monocentric cohort study including all consecutive patients receiving a heart transplant in our center from 2011 to 2017, with an assessment by right heart catheterization in the year before transplantation.

Results: Among the 304 patients included, 129 (42%) had low PVR (PVR ≤ 2.5 WU), and 175 (58%) had high PVR (PVR > 2.5 WU) before transplant. 82% of patients with high PVR and 81% of patients with low PVR were alive at 1 year follow-up (p = 0.6). Patients with high PVR were more likely to require immediate post-transplant circulatory support by veno-arterial extracorporeal membrane oxygenation (VA-ECMO) (58% vs 44%, p = 0.02), as well as a pulmonary vasodilator treatment by Sildenafil (14% vs 7%, p = 0.045). PVR were not associated with 1-year post-transplant mortality in multivariate analysis. At 1-year follow-up, right and left ventricular graft function were preserved for all survivors, and did not differ between high and low PVR groups. Even in the subgroup of the 24 patients transplanted with PVR > 5WU, unresponsive to vasodilator challenge, 1-year survival was 92%, with preserved right and left ventricular function.

Conclusion: PH was not associated with a poor outcome in our cohort, even when it was severe and unresponsive to vasodilator challenge. However, the rate of VA-ECMO support immediately after heart transplant was higher in patients with high PVR. Cardiac transplantation as first line strategy may be a valuable option in patients with elevated pre-transplant PVR.

Compliance with ethics regulations: Yes.

CO-07 Hemodynamic and microcirculation evaluation of Vasopressin versus Norepinephrine in a porcine model of refractory cardiac arrest resuscitated by venous-arterial ECMO

Thomas Klein¹, Caroline Fritz¹, Daniel Grandmougin², Yihua Liu², Sophie Orlowski³, Tran N’guyen⁴, Eliane Albuisson⁵, Bruno Lévy¹

¹Service de Réanimation Médicale Brabois, CHRU Nancy, Pôle Cardio-Médico-Chirurgical, INSERM U1116 Vandoeuvre-Les-Nancy, France; ²Service de Chirurgie Cardiaque Brabois, CHRU Nancy, Pôle Cardio-Médico-Chirurgicale, INSERM U1116, Vandoeuvre-Les-Nancy, France; ³Service de Biochimie, Pôle Laboratoires, CHRU Nancy Central, Nancy, France; ⁴Ecole de chirurgie, Faculte de Médecine, Université de Lorraine, Vandoeuvre-Les-Nancy, France; ⁵Plateforme d’aide à la recherche clinique (PARC-UMDS), Hôpital de Brabois, CHRU Nancy, Vandoeuvre-Les-Nancy, France

Correspondence: Thomas Klein (thomas090@outlook.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-07

Rationale: Venoarterial extracorporeal membrane oxygenation (VA-ECMO) is used to support tissue perfusion during extracorporeal cardiopulmonary resuscitation (e-CPR). Shock, resuscitation and the extracorporeal circuit may trigger a capillary leakage and a vasoplegic shock. Currently, in these situations, high doses of Norepinephrine (NE) are required. Because high NE doses may have significant cardiovascular side effects, alternative options to support arterial blood pressure are needed. In recent years, several approaches to decrease the administration of high NE doses have been tested, one of them is the administration of Vasopressin (AVP). Randomized trials have shown that AVP infusion increases arterial pressure and systemic vascular resistance, decreases catecholamine requirements in patients with or at high risk of vasoplegic syndrome and attenuates vascular dysfunction. Currently, no data are available for the study of the effects of AVP in shock state in post refractory cardiac arrest.

Patients and methods: 20 pigs were randomized into two groups, in order to receive AVP or NE. A refractory cardiac arrest of ischemic origin was surgically created and VA-ECMO was started after a 30 min period of cardio-pulmonary resuscitation. Then, resuscitation lasted 6 h in each randomization group. The evolution of the consequences of the shock was evaluated by lactatemia and microcirculation (SDF and NIRS) at baseline hour, H0 (when ECMO starts), H3 and H6. Renal and hepatic functions were assessed.

Results: Experimental conditions were met for 16 animals (AVP, n = 8; NE, n = 8). The groups were comparable on the shock impact and its severity. No significant differences were found between populations for ECMO flow and MAP. There was a significant difference on fluid volume resuscitation amount (14000 [11.250–15.250] mL in the NE group versus 3500 [1750–4000] mL in the AVP group, p < 0.05) (Fig. 1). No significant difference between the NE and AVP groups for lactate clearance between H0 and H6 (25.6 [− 7.31 to 35.34]% vs 47.84 [13.42–82.73]%, p = 0.686). We did not find any significant for sublingual microcirculation indices and NIRS values. Renal and liver function evolution were similar in the two groups during the protocol.

Conclusion: AVP administration in refractory cardiac arrest resuscitated by VA-ECMO when compared to NE is associated with less fluid volume for similar global and regional hemodynamic effects.

Compliance with ethics regulations: Yes.

Fig. 1

Fluid volume resuscitation in Vasopressin group compared to Norepinephrine group

CO-08 Veno-arterial Extracorporeal Membrane Oxygenation flow or Dobutamine to increase microcirculation for Refractory Cardiogenic Shock

Juliette Chommeloux¹, Santiago Montero², Guillaume Franchineau¹, Alain Combes¹, Matthieu Schmidt¹

¹Assistance Publique–Hôpitaux de Paris, Pitié–Salpêtrière Hospital, Medical Intensive Care Unit, 75651 Paris Cedex 13, France; ²Acute and intensive cardiovascular care unit, department of cardiology, hospital de la Santa Creu i Sant Pau, Barcelona, Spain

Correspondence: Juliette Chommeloux (juliette.chommeloux@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-08

Rationale: Venoarterial extracorporeal membrane oxygenation (VA-ECMO) is an effective technique to support refractory cardiogenic shock (CS) and increase macro- and microcirculation. Given that major and persistent microcirculation alterations are associated with worse outcome, we investigated the respective impact of an increase of either the VA-ECMO flow and or dobutamine dose on microcirculation in stabilized patients with refractory CS on VA-ECMO.

Patients and methods: Prospective study in academic medical intensive unit. Consecutive patients with ECMO-supported CS instability, who were able to tolerate a stepwise increment of the dobutamine dose and the ECMO flow. Baseline was defined by the lowest VA-ECMO flow and dobutamine 5 µg/kg/min for PAM ≥ 65 mmHg. Starting from the baseline, VA-ECMO flow was progressively increased by 25% (ECMO125%, ECMO150%, ECMO175%, ECMO200%). Back to baseline, a stepwise increase of the dobutamine to 10, 15 and 20 µg/kg/min (DOBU10, DOBU15, DOBU20) was performed. Macro- and microcirculatory evaluations were made after 30 min in each condition.

Results: Fourteen patients (median age 52 [40–61] years; SAPS II 68 [52–76]) were included. Acute myocardial infarction was the main cause of cardiogenic shock (64%). Macro- and microcirculation were assessed 2 [2–5] days after ECMO start. The increment of the dobutamine dose did not modify microcirculation parameters. De Backer score tended to be reduced (p = 0.08), with a significant mean arterial pressure (MAP) increase during the ECMO flow increment. These findings were not different between patients successfully weaned-off ECMO (n = 6) and those who did not.

Conclusion: When macro and microcirculation are already restored on ECMO-supported refractory CS, increasing dobutamine (above 5 µg/kg/min) or ECMO flow did not further improve microcirculation. For now, ECMO flow and dobutamine should be set as the minimum flow to get a MAP ≥ 65 mmHg and lactate < 2.5 mmol/L, and the minimal dose to maintain aortic valve opened, respectively.

Compliance with ethics regulations: Yes.

Fig. 1

Bar plots of macro- and microcirculation-parameters’ trends during a progressive increment of either the dobutamine dose or the ECMO flow. The point corresponds to the median, the T-bars represent the 25th and 75th percentile, p shows the effect of a progressive increment of either the dobutamine or the ECMO flow

CO-09 Weaning from noninvasive ventilation and high flow nasal cannula in patients with severe bronchiolitis

Julie Cassibba¹, Anne Ego², Isabelle Pin¹, Guillaume Mortamet³

Service de pédiatrie, CHU de Grenoble, Grenoble, France; ²Département de santé publique, CHU de Grenoble, Grenoble, France; ³Service de réanimation pédiatrique, CHU de Grenoble, Grenoble, France

Correspondence: Julie Cassibba (jcassibba@chu-grenoble.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-09

Rationale: Non Invasive Ventilation (NIV) and High Flow Nasal Cannula (HFNC) are the first-line therapies for the most severe patients with acute bronchiolitis. Unlike invasive mechanical ventilation, there is no consensus with regards to weaning from NIV/HFNC. The main objective of this study is to describe the weaning practices from NIV/HFNC in patients with acute bronchiolitis.

Patients and methods: A single-center prospective study. Patients younger than 6 months with severe bronchiolitis and supported by NIV or HFNC were included. NIV/HFNC was discontinued according to the local practices and no protocol existed. Exceptt the principal investigator, the attending team was blinded to the study. Weaning failure was defined as the need to reinstate NIV/HFNC in the 48 h after discontinuation. Ethical approval was not necessary for this study in accordance with the French data protection autority methodology reference number MR-004.

Results: A total of 95 patients (median age 47 days, 53 (56%) males) were included. Respectively, 72 (76%) and 23 patients (24%) were supported by NIV and HFNC at admission (Fig. 1). Regarding the mode of NIV, a bilevel mode was used in 46 patients (48%) (Fig. 1). In patients supported by HFNC, the ventilatory support was discontinued progressively by decreasing air flow in 9 patients (39%) while it was stopped abruptly in 5 (22%). In patients supported by NIV, the respiratory support was stopped abruptly in 5 (19%) of them while HFNC was used as a weaning method for 17 (65%) patients. A total of 22 (23%) patients experienced a weaning failure. Patients supported by NIV/HFNC who experienced a prompt weaning had a lower Pediatric Intensive Care Unit (PICU) length of stay as compared to patients in whom HFNC was used as a weaning method (78 ± 27 h versus 112 ± 112 h, p = 0.01). However, the hospital length of stay was similar according to the weaning method (6 ± 3 days versus 7 ± 3 days for prompt and progressive methods respectively, p = 0.07). The duration of the weaning process did not differ according to the bed-availability in PICU.

Conclusion: In patients with severe bronchiolitis, a prompt weaning from NIV/HFNC was associated with a lower length of stay in PICU. However, the hospital length of stay was similar according to the weaning method. We suggest that a prompt weaning should be preferred in order to reduce the risk of PICU related complications.

Compliance with ethics regulations: Yes.

Fig. 1

Flow chart of the study

CO-10 Influenza-associated encephalitis: a French multicentric retrospective study in pediatric intensive care units

Pierre CLEUZIOU¹, Florence RENALDO¹, Sylvain RENOLLEAU², Isabelle DESGUERRE², Etienne JAVOUHEY³, Pierre TISSIERES⁴, Pierre-Louis LEGER⁵, Stéphane DAUGER¹, Michael LEVY¹

¹Hôpital Robert Debré-APHP, Paris, France; ²Hôpital Necker Enfants Malades-APHP, Paris, France; ³Hôpitaux de Lyon, Lyon, France; ⁴Hôpital du Kremlin-Bicêtre-APHP, Le Kremlin Bicêtre, France; ⁵Chef de service de la réanimaton pédiatrique de l’hôpital Trousseau, Paris, France

Correspondence: Pierre Cleuziou (pierre.cleu@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-10

Rationale: Influenza-associated encephalitis (IAE) is a very rare disease with poorly understood pathophysiology. It has mainly been described in Japan, where the death rate is around 8% and where one-third of the survivors suffer from neurological sequelae. The objective of this study was to describe severe forms of the disease among children hospitalized in pediatric intensive care unit (PICU) and to estimate the death rate in this population.

Patients and methods: By using consensus definition criteria, we retrospectively identified children hospitalized between 2010 and 2018 in 12 french PICU for an encephalitis associated with a laboratory-proven acute infection to Influenzae virus. Patients with preexisting neurological chronic disorder or presenting a co-infection potentially responsible of the disease were excluded. We collected data describing clinical presentation, cerebro-spinal fluid (CSF) results, electroencephalographic and MRI findings, therapeutics used in PICU and outcome at discharge.

Results: 41 patients were included with 4.7 years old as median age (range 0.8–15.4 years old). Most of the patients were admitted in ICU less than 48 h after the first symptoms (62%, n = 25). The main clinical features were fever (93%, n = 38), vomiting (44%, n = 18), altered consciousness (100%, n = 41), epileptic seizures (88%, n = 36), status epilepticus (54%, n = 22) and motor weakness or pyramidal signs (71%, n = 29). 48% of patients had meningitis (n = 16) and the virus was never found in the CSF (n = 0/13). One-third of children (n = 13) presented MRI lesions compatible with acute necrotizing encephalitis. Regarding therapeutics, 80% of patients required mechanical ventilation, especially for neurologic dysfunction, and the use of specific treatments was very heterogeneous: 68% had oseltamivir, 49% boluses of corticosteroids, 24% intravenous immunoglobulins and 10% plasma exchanges. The median length of stay (LOS) in PICU was 7 days (range 1–87 days) and there were 7 fatalities (17%). Among survivors, 35% had severe neurological sequelae at discharge from the hospital (n = 11). A predisposing mutation in the RANBP2 gene was rarely sought (15%, n = 6) but was positive in one out of two patients.

Conclusion: Patients requiring PICU for an IAE still have an extremely severe prognosis with a high mortality rate and frequent neurological sequelae. It appears that patients’ therapeutic management is still heterogeneous because of the lack of consensual guidelines. The research of a predisposing genetic mutation in RANBP2 is not yet part of the systematic etiologic assessment although it is known to be an important risk factor for the severe form of the disease.

Compliance with ethics regulations: Yes.

CO-11 Population pharmacokinetics of Cefazolin in critically ill children infected with methicillin-sensitive Staphylococcus aureus

Elodie Salvador, Mehdi Oualha, Emmanuelle Bille, Olivier Bustarret, Agathe Béranger, Guillaume Geslain, Florence Moulin, Julie Toubiana, Sihem Benaboud, Sylvain Renolleau, Jean-Marc Tréluyer, Déborah Hirt

Necker Hospital, Paris, France

Correspondence: Elodie Salvador (elodie.salvador@aphp.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-11

Rationale: Cefazolin is one of curative treatments for infections with methicillin-sensitive Staphylococcus aureus (MSSA) which may occur in critically ill children. Both growth and critical illness may impact of pharmacokinetics (PK) in this vulnerable population. We aim to establish a PK model for cefazolin, using a population approach, and in turn to optimize individual dosing regimens.

Patients and methods: We included all children (age < 18 years, bodyweight (BW) > 2.5 kg) receiving cefazolin and infected with MSSA. Cefazolin serum total concentrations were quantified by high-performance liquid chromatography. Data modelling process has been done with a non-linear mixed-effect modeling software MONOLIX. Monte Carlo simulations were used to optimize individual dosing regimens in order to attain the target of 100% [fT (4xMIC)].

Results: Thirty-nine patients with a median age of 7 years (0.1–17), a body weight (BW) of 21 kg (2.8–79) and an estimated glomerular filtration rate (eGFR) of 189 mL/min/1.73 m² (66–486) were included. The PK was ascribed a one-compartment model with first-order elimination, where clearance and volume of distribution estimated were 1.4 L/h and 3.3 L respectively, normalized to a median subject of 21 kg and eGFR of 189 mL/min/1.73 m². BW, according to the allometric rules, and eGFR were the significant covariates. Under simulations, continuous infusion with a dose of 100 mg/kg/day was the best scheme to reach the target of 100% [fT (4xMIC)] (Fig. 1). A dose of 150 mg/kg per day by continuous infusion is more appropriate for children with BW < 10 kg or eGFR > 200 mL/min, while also limiting side effects.

Conclusion: In critically ill children infected with MSSA, BW with allometric scaling and eGFR were the main influential covariates on cefazolin PK parameters. Current and recommended dosing regimens of cefazolin may be not sufficient to reach the target of 100% [fT (4xMIC)] in all children. Continuous infusion with a dosing of 100 to 150 mg/kg/day seems to be the best scheme to achieve the target of 100% [fT (4xMIC)] in children with normal and augmented renal function, respectively.

Compliance with ethics regulations: Yes.

Fig. 1

Probability of target attainment obtained on 39,000 simulations from study population for a target defined as 100% 〖fT〗_ (4× MIC.) BW, body weight (kg); eGFR, estimated glomerular filtration (mL/min); PTA, probability of target attainment (%); MIC, minimum inhibitory concentration (mg/L); q6h, administration every 6 h; q8h, administration every 8 h; CI, continuous infusion

CO-12 Factors influencing unplanned extubations in a pediatric intensive care unit: a 9-year prospective study

Guillaume Geslain, Jeanne Mayeur, Michael Levy, Géraldine Poncelet, Fleur Le Bourgeois, Arielle Maroni, Camille Dollat, Jérôme Naudin, Stéphane Dauger, Maryline Chomton, Anna Deho

Pediatric Intensive Care Unit, Assistance Publique-Hôpitaux de Paris, Robert Debre University Hospital, Université de Paris, Paris, France

Correspondence: Guillaume Geslain (guillaume.geslain@aphp.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-12

Rationale: Unplanned extubation (UE) is a potentially serious complication that can contribute to increase patient morbidity, mortality and length of stay in pediatric intensive care unit (PICU). The most frequent risk factors reported in literature are younger age, patient agitation, inadequate tube fixation, copious secretions, procedures performed during medical or paramedical care and nursing workload. The aim of our study is to determine incidence and contributing factors of UE in our PICU.

Patients and methods: We conducted a prospective monocentric study from January 1, 2010, to December 31, 2018 in a tertiary PICU. All mechanically ventilated children less than 18 years-old admitted to our PICU were included except for patients with tracheotomy. We monitored all cases of UE occurred during PICU stay. Demographic and clinical data were collected using an event report form including patient characteristics, a description of the extubation circumstances and the outcomes.

Results: We reported 102 UE in 9 years for 14,661 days of mechanical ventilation. The overall UE incidence was 6.96 per 1000 ventilation days. Characteristics of patients and UE circumstances are reported in Table 1. Some data on events after UE are missing. Thirteen patients on 82 (15.9%) needed non-invasive ventilation after UE, 4 of them needed reintubation (30.8%). Sixty patients on 97 (61.9%) needed reintubation, and 47 of them (78.3%) required immediate reintubation. Over the study period, the number of UE per year was stable. Presence of physical restraints, nasally inserted endotracheal tubes and cuffed tubes seemed not to prevent patients from self-extubating.

Discussion: In North America, the nurse-to-patient ratio in PICU is 1:1, in our unit this ratio is at 1:2.29. A reorganization of the PICU staff for intubated patients should be redesigned in our unit. It would also be interesting to study the kinetic of sedation decreasing and to collect the presence or absence of a withdrawal syndrome for its potential implication in UE.

Conclusion: Areas for potential improvement seem to lie in implementation of monitoring and staff education programs. Iterative intubations can cause laryngeal and tracheal trauma and prolong length of mechanical ventilation and PICU stay. Our aim is to implement strategies to reduce the occurrence of UE.

Compliance with ethics regulations: Yes.

Table 1 Characteristics of patients and circumstances of unplanned extubation

CO-13 Score for the risk of acquisition of pressure ulcer in the ICU: Data from the Pressure Study

Philippe Michel¹, Gwenaëlle Jacq¹, Gregoire Muller¹, Guillaume Decormeille¹, Atika Youssoufa¹, Laurent Poiroux¹, Brigitte Barrois², Nadia Aissaoui¹, Saber Davide Barbar¹, Florence Boissier¹, David Grimaldi¹, Sami Hraiech¹, Nicholas Heming¹, Bertrand Hermann¹, Jean-FrançOis Llitjos¹, Lamia Besbes¹, Jean-Baptiste Lascarrou¹, Gaël Piton¹

¹CERC, Paris, France; ²Association PERSE, Paris, France

Correspondence: Philippe Michel (philippe@docteur-michel.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-13

Rationale: Critically ill patients are at risk of developing pressure ulcers (PUs) during their ICU stay. Existing scores of PU have not been validated in the context of ICU. We aimed to create a score evaluating the risk of acquisition of PU being specific of the ICU.

Patients and methods: Data from a one-day point prevalence study performed in June 2017 in 1228 patients in 86 ICUs in France (The Pressure Study). On the same day, the presence or absence of PUs in all hospitalized patients of participating ICUs, data on the ICUs, and the characteristics of patients and of PUs had been evaluated. Factors having been significantly associated with acquisition of PU in the ICU by univariate analysis were selected. Quantitative variables were dichotomized at a threshold identified by the ROC curve analysis. The variables were included in a multiple logistic regression analysis for the acquisition of PU in the ICU. The independent variables being the most clinically relevant were included in a score, with a weighting of 1 point for each variable.

Results: After exclusion of the IGS2 score, 6 variables were independently associated with acquisition of PU in the ICU: having been confined to bed before ICU admission, presence of motor neurological disorder, body weight ≥ 90 kg, use of high-dose steroids, length of ICU stay > 10 days and need for artificial nutrition (AUC = 0.78). The AUC of a model limited to the four first described variables was correct (AUC = 0.73). A “BCD Weight” score (Bed, Corticosteroids, motor Deficit, body WEIGHT) ranging from 0 to 4 was created. The prevalence of acquired PU was 4.6%, 12.9%, 26.0%, 48.8%, and 100%, among patients presenting with a score 0, 1, 2, 3, and 4, respectively (Fig. 1). Comparatively, the prevalence of acquired PU was 0%, 3.7%, 9.3% and 26.9% among patients classified as no risk, low risk, moderate risk, and high risk with the usual scale evaluating the risk of PU.

Conclusion: We identified an easy to remember score addressing the risk of PU in the context of ICU. This “BCD Weight” score, calculable from ICU admission and throughout the stay, might help clinicians to evaluate the risk of acquisition of PU in their patients.

Compliance with ethics regulations: Yes.

Fig. 1

Comparison of a score specific of the ICU with usual scale for the acquisition of PU

CO-14 Effect of short incitement to follow social media to improve knowledge of scientific literature by intensive care trainees

Jean-Baptiste Lascarrou¹, Stephan Ehrmann², Maelle Martin¹, Pierre Potier¹, Jean-Marie Castillo³, Jean Reignier¹, Emmanuel Canet¹

¹Médecine Intensive Réanimation, Nantes, France; ²Médecine Intensive Réanimation, Tours, France; ³General Practionner, Nantes, France

Correspondence: Jean-Baptiste Lascarrou (jeanbaptiste.lascarrou@chu-nantes.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-14

Rationale: Improving knowledge of recent medical literature by intensive care trainees is a major target for medical education and the first step to improve patient’s care. We test the hypothesis that following major Twitter feeds dedicated to intensive care will improve knowledge of recent studies published in medical journals by intensive care trainees.

Patients and methods: We included French intensive care trainees (ICT) between March 2019 and October 2019 in an educational randomized trial. All ICT were informed and they consent to the trial. After initial online interrogation, ICT were separated between those already on Twitter© or not (Group 1: ICT already on Twitter©). All ICT not already on Twitter were randomized on 2 groups: Group 2 (short information and incitation to open a Twitter account and to follow critical care journal feeds) or group 3 (control group). ICT were interrogated on their recent medical literature knowledge at 3 and 6 month on 5 trials published in pre-selected journals.

Results: During the study period, on the 969 French ICT contacted, 77 agree to participate: 16 were already on Twitter, 31 were randomized to Twitter incitation and 30 to control group. At 3 month, there were 62 who answered electronic questionnaire. Self-declaration of article knowledge was not different between 3 groups (P = 0.85). Knowledge of primary outcome of each trial was not significantly better in 3 groups (P = 0.09). In per-protocol analysis of ICT on Twitter or not, knowledge of article and primary outcome were also not significantly different (respectively P = 0.57 and P = 0.17).

Conclusion: Short incitation to open a Twitter account and follow major medical journals with specific focus on cardiac arrest did not improve knowledge of medical literature by intensive care trainees at 3 month. Further trials are needed to better imply intensive care trainees in scientific medical literature.

Compliance with ethics regulations: Yes.

Table 1 Results at 3 month

CO-15 Construction of a stress scale specific to intensive care units: the PS-ICU scale

Alicia Fournier¹, Florent Lheureux², Maria Cruz Martin Delgado³, Maria Grazia Bocci⁴, Alessia Prestifilippo⁵, Amélie Anota⁶, Pierre Aslanian⁷, Guillaume Besch⁸, Jean-Michel Constantin⁹, Jean-Pierre Quenot¹⁰, Belaid Bouhemad¹¹, Gilles Capellier¹², Alexandra Laurent¹

¹Department of psychology, University of Burgundy, Dijon, France; ²Department of psychology, University of Franche-Comté, Besançon, France; ³Hospital Universitario Torrejón en Torrejón de Ardoz, Madrid, Spain; ⁴Department of anesthesiology and intensive care medicine, Fondazione Policlinico Universitario Agostino Gemelli—IRCCS, Roma, Italy; ⁵Psychologist, Private practice, Roma, Italy; ⁶Methodology and quality of life unit in oncology, Centre hospitalier régional universitaire de Besançon, Besançon, France; ⁷Department of intensive care medicine, CHUM, Montréal, Canada; ⁸Surgical intensive care unit, Centre hospitalier régional universitaire de Besançon, Besançon, France; ⁹Anesthesiology critical care and perioperatrive medicine, Hôpital La Pitié Salpêtrière, Paris, France; ¹⁰Department of intensive care medicine, Centre hospitalier régional universitaire de Dijon, Dijon, France; ¹¹Surgical intensive care unit, Centre hospitalier régional universitaire de Dijon, Dijon, France; ¹²Department of intensive care medicine–samu 25, Centre hospitalier régional universitaire de Besançon, Besançon, France

Correspondence: Alicia Fournier (alicia.fournier@u-bourgogne.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-15

Rationale: The intensive care units (ICU) are fertile ground for the emergence of professional stressors. Assessed by individuals as situations that weaken or are beyond their resources, work-related stressors impact the mental and physical health of workers and the quality and safety of care. Currently, many tools are used to assess caregiver stress in ICU, but do not consider the specificity of this work. The objective of this international and multicentric study was to develop a perceived stress scale specific to ICU.

Patients and methods: Interviews were conducted with 166 caregivers (84 nurses and 81 physicians) in four countries (France, Italy, Spain, Canada). These interviews were recorded, transcribed and then a thematic analysis was carried out to identify stress factors. A first version of the scale was pre-tested with 70 caregivers (30 physicians and 40 nurses) in the same countries. Finally, we carried out qualitative and quantitative analyses select the most relevant items.

Results: We identified 99 stressors specific to the ICU that were grouped into eight main themes: stress in relation to 1) the patient, 2) the task to be performed, 3) the institutional context, 4) the team, 5) the organization of the unit, 6) the personal dimensions, 7) the patient’s family, 8) the working conditions. Following the pre-test, 50 items were selected to constitute the PS-ICU scale.

Conclusion: Our results highlight specific items related to vital risk/emergency management and ethically and morally problematic situations. These dimensions will be discussed and compared against existing scales (e.g., JCQ). The PS-ICU scale will allow to better identify and measure stressors in ICU. This scale will contribute to the development of targeted actions in terms of prevention, training and support for professionals. The creation of an internationally validated tool will make it possible to develop comparative studies on cultural and organizational factors.

Compliance with ethics regulations: Yes.

CO-16 Are Intensive Care Residents more exposed to Anxiety/Depression?

Mehdi Marzouk¹, Manel Lahmar², Zeineb Hammouda², Islem Ouanes², Fahmi Dachraoui², Lamia Besbes², Fekri Abroug²

¹Service de Réanimation Centre Hospitalier de Béthune, Beuvry, France; ²Service de Réanimation Polyvalente. CHU F.Bourguiba, Monastir, Tunisia

Correspondence: Mehdi Marzouk (docmehdi.marzouk@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-16

Rationale: When compared to general population, moods disorders are more prevalent among health care workers and especially young doctors. Whether certain specialties are more exposed than others given the burden of workload and specific aspects is not known. The aim of this study is to assess the prevalence of anxiety and depressive symptoms among Tunisian young residents and verify whether they are more frequent in specialties with high workload such as Intensive Care Medicine.

Patients and methods: We conducted a cross-sectional survey in all Tunisian medical residents brought together between 14 and 22 December 2015 to choose their next 6-month rotation. The items of the Hospital Anxiety and Depression (HAD) questionnaire were employed to capture the prevalence of anxiety and/or depression among the residents. The statistical relationships between anxiety and depression (HAD score) and work-related data were explored by Poisson regression. In particular we compared a group of specialties including Intensive care, Anesthesiology, and Emergency medicine (Acute care group), to the rest of specialties.

Results: 1700 out of 2200 (77%) medical residents answered the questionnaire. Among these, residents who started the first semester of a new curriculum (n = 320) were not included. 243 (17.6%) were in the acute care group.

Overall, 73.5% of the participating residents had either definite (44.2%) or probable (29.3%) anxiety, while 65% had definite (33.5%) or probable (31.5%) depression symptoms. In the acute care group, these proportions were not substantially higher: 48.6% and 27.6% for definite and probable anxiety, respectively; and 37% and 31.7% for definite and probable depression, respectively. Total HAD score was significantly associated with the resident’s age (OR = 1.01, 95% CI 1.004 to 1.02, p = 0.001); female gender; and the heavy burden of work imposed on a weekly or monthly basis, as reflected by the number of hours worked per week (0.3% increase per worked hour per week), and the number of night shifts per month (1.5% increase per night shift).

Conclusion: Anxiety/Depression symptoms are not more frequent in Intensive Care, Anesthesiology, or Emergency medicine residents. Rather, these symptoms are related to the socio-demographic situation of residents, and the workload characteristics in general.

Compliance with ethics regulations: NA.

CO-17 Outcome of adult sickle cell patients admitted in ICU: national retrospective study in French ICUs

Maïté Agbakou¹, Noelle Brule², Morgane Pere³, Emmanuel Canet², Jean Reignier², Jean-Baptiste Lascarrou²

¹Service de médecine intensive réanimation, CHU Nantes, Nantes, France; ²CHU Nantes, Service De Médecine Intensive Réanimation, Chu Nantes, France; ³Plateforme de Méthodologie et Bistastistiques, direction de la recherche, CHU Nantes, Nantes, France

Correspondence: Maïté Agbakou (maite.agbakou@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-17

Rationale: Sickle cell disease (SCD) is associated with high morbidity and mortality, and most of deaths in sickle cell patients occur in the ICU. The purpose of this styudy was to describe clinical characteristics of adult sickle cell patients requiring ICU admission, and identify prognostic factors associated with an adverse outcome defined by death in the ICU or need for vital support.

Patients and methods: This multicentric observational retrospective study included all adult patients with SCD, admitted in 16 French university hospital ICUs, from January 1st 2015 to December 31th 2017. Only the first episode on the study period was analyzed for each patient. Main outcome was the occurrence of an adverse outcome, defined by death in ICU or need for vital support. Predictors of adverse outcome were assessed by Cox regression model.

Results: Four hundred and eighty-eight patients were included during the study period. Reasons for ICU admission were mainly SCD related, with acute chest syndrome being the first one (47.5%). Adverse outcome occurred in 81 (16.6%) patients, with 9.4% patients requiring invasive mechanical ventilation, 5.9% non-invasive mechanical ventilation, 6.6% vasopressor support, 3.7% renal replacement therapy and 1.6% ECMO (arterio-veinous or veno-venous). Those patients had more often high blood pressure, chronic kidney failure and pulmonary hypertension than those without adverse outcome. Sixteen (3.3%) patients died in the ICU, mainly of multi-organ failure following sickle cell crisis or sepsis. In multivariable analysis, independent predictors of adverse outcome were mean arterial pressure (OR 0.98 IC 95% (0.96–1), p = 0.034), respiratory rate (OR 1.09 IC 95% (1.01–1.47), p = 0.035), hemoglobin level (OR 1.22 IC 95% (1.01–1.47)) and creatinine clearance (OR 0.98 IC 95% (0.97–0.98) p < 0.0001) in ICU, and blood exchange transfusion before ICU admission (OR 5.75 IC 95% (1.32–25.03), p = 0.02).

Conclusion: This multicentric study confirmed well-known predictors of adverse outcome, and identified for the first-time low mean arterial blood pressure in ICU, and blood exchange transfusion before ICU admission as independent predictors of adverse outcome in SCD. This last finding interrogates the need for systematic referral to ICU of SCD patients requiring blood exchange transfusion.

Compliance with ethics regulations: Yes.

CO-18 In-Hospital Mortality-Associated Factors of Thrombotic Antiphospholipid Syndrome Patients Requiring Intensive Care Unit Admission

Marc Pineton De Chambrun¹, Romaric Larcher², Frédéric Pene³, Laurent Argaud⁴, Julien Mayaux⁵, Rémi Coudroy⁶, Elie Azoulay⁷, Yacine Tandjaoui-Lambiotte⁸, Stanislas Faguer⁹, Charles-Edouard Luyt¹, Alain Combes¹, Zahir Amoura¹⁰

¹Sorbonne Université, APHP, Hôpital La Pitié–Salpêtrière, Institut de Cardiométabolisme et Nutrition (ICAN), Service de Médecine Intensive-Réanimation, Paris, France; ²Service de Médecine Intensive-Réanimation, Hôpital Lapeyronie, Centre Hospitalier Universitaire (CHU) de Montpellier PhyMedExp, Université de Montpellier, INSERM, CNRS, Montpellier, France; ³Service de Médecine Intensive-Réanimation, Hôpital Cochin, Hôpitaux Universitaires Paris Centre, APHP & Université Paris Descartes, Paris, France; ⁴Service de Médecine Intensive-Réanimation, Hôpital Edouard-Herriot, Hospices Civils de Lyon, Lyon, France; ⁵APHP, Hôpital La Pitié–Salpêtrière, Service de Pneumologie, Médecine Intensive et Réanimation Médicale, Département R3S, Paris, France; ⁶Service de Médecine Intensive-Réanimation, INSERM CIC1402, Groupe ALIVE, Université de Poitiers, CHU de Poitiers, Poitiers, France; ⁷Service de Médecine Intensive-Réanimation, Hôpital Saint-Louis, APHP, Paris, France; ⁸Service de Réanimation Médico-Chirurgicale, Hôpital Avicenne, APHP, HUPSSD, Bobigny, France; ⁹Département de Néphrologie et Transplantation d’Organes, Unité de Réanimation, Centre de Référence des Maladies Rénales Rares, Hôpital Rangueil, CHU de Toulouse, Toulouse, France; ¹⁰Sorbonne Université, Assistance Publique-Hôpitaux de Paris (APHP), Hôpital La Pitié–Salpêtrière, Institut E3M, Service de Médecine Interne 2, Centre de Référence National Lupus Systémique, Syndrome des Anticorps Anti-phospholipides et Autres Maladies Auto-Immunes Systémiques Rares, Paris, France

Correspondence: Marc Pineton de Chambrun (marc.dechambrun@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-18

Rationale: The antiphospholipid syndrome (APS) is a systemic autoimmune disease defined by thrombotic events that can require ICU admission because of organ dysfunction related to macrovascular and/or microvascular thrombosis. Critically ill thrombotic APS patients were studied to gain insight into their prognoses and in-hospital mortality-associated factors.

Patients and methods: This French national, multicenter, retrospective study included all APS patients with any new thrombotic manifestation (s) admitted to 24 ICUs (January 2000-September 2018).

Results: During the study period, 134 patients (male/female ratio, 0.4) with 152 APS episodes were admitted to the ICU (at mean age 46.0 ± 15.1 years). In-hospital mortality of their 134 last episodes was 35/134 (26.1%). The Cox multivariable model retained (HR [95% CI]): age ≥ 40 years (11.4 [3.1–41.5]; P < .0001), mechanical ventilation (11.0 [3.3–37]; P < .0001), renal replacement therapy (2.9 [1.3–6.3]; P = .007) and in-ICU anticoagulation (0.1 [0.03–0.3]; P < .0001) as independently associated with in-hospital mortality. For the subgroup of “definite/probable CAPS”, the Cox bivariable model including the SAPS II score retained double therapy (corticosteroids + anticoagulant: 0.2 [0.07–0.6]; P = .005) but not triple therapy (corticosteroids + anticoagulant + intravenous immunoglobulins or plasmapheresis: HR 0.3 [0.1–1.1]; P = .07) as independently associated with in-hospital mortality (Fig. 1).

Discussion: Triple therapy is the recommended first-line treatment of CAPS. However, herein, it was not significantly associated with better survival in critically ill, thrombotic APS patients. For the subgroup of “definite/probable CAPS” patients, double and triple regimens were associated with survival. But the bivariable analyses including the day-0 SAPS II showed that survival was linked to in-ICU anticoagulation and corticosteroids—not IVIg or plasmapheresis. Our findings indicate that corticosteroids should probably be added to in-ICU anticoagulation to treat “definite/probable CAPS”. Frequent fever and elevated C-reactive protein in all thrombotic APS patients suggest a marked inflammatory state that could explain corticosteroid efficacy. Neither plasmapheresis nor IVIg impacted the prognosis of “definite/probable CAPS”, but that finding could be explained by a lack of power compared to CAPS Registry data.

Conclusion: In-ICU anticoagulation was the only APS-specific treatment independently associated with survival for all patients. Double—but not triple—therapy was independently associated with better survival of “definite/probable CAPS” patients. In these patients, double therapy should be used as first-line therapy while the role of triple therapy requires further evaluation.

Compliance with ethics regulations: Yes.

Fig. 1

Kaplan-Meier estimated probability of survival for the 134 APS patients’ last episodes requiring intensive care unit admission with comparison according triple therapy use

CO-19 Coagulation disorders in critically ill HLH patients: a prospective study

Sandrine Valade¹, Michaël Darmon¹, Amélie Launois², Bérangère Joly², Jehane Fadlallah³, Lionel Galicier³, Claire Fieschi³, Lara Zafrani¹, Virginie Lemiale¹, Anne Claire Lepretre⁴, Adrien Mirouse¹, Jean-Jacques Tudesq¹, Agnès Veyradier², Elie Azoulay¹, Eric Mariotte¹

¹Saint-Louis Hospital, Medical ICU, Paris, France; ²Hematology Department and Research Unit EA3518, Institute of Hematology, French Reference Center for Thrombotic Microangiopathies, Lariboisière Hospital, University Paris Diderot, Paris, France; ³Department of Clinical Immunology, Saint-Louis Hospital, Paris, France; ⁴Saint-Louis Hospital, Transfusion Department, Etablissement Français Du Sang, Paris, France

Correspondence: Sandrine Valade (sandrine.valade@aphp.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-19

Rationale: Hemophagocytic lymphohistiocytosis (HLH) is a rare condition that can be severe and lead patients to the ICU. Coagulation disorders are commonly observed during HLH, the most frequently reported being a decreased fibrinogen level. Hemostasis impairment has been associated with increased risk of bleeding and death in previous studies. The main objective of this study was to describe coagulation defects during HLH in order to identify early markers associated with bleeding events.

Patients and methods: In this prospective study conducted in the ICU and the hematological wards of one university hospital between April 2015 and December 2018, all the patients with a new diagnosis of HLH were included. Blood samples were retrieved at day 1 and day 7 to explore hemostasis. Coagulation disorders were defined as PT < 50% and/or fibrinogen < 2 g/L. Results are presented as median [interquartile range] and number (percent).

Results: During the study period, 47 patients aged 54 years [42–67] were included. Seventy-nine percent required ICU admission, mainly for acute respiratory (n = 14; 30%) or hemodynamic failure (n = 10; 21%). Patients fulfilled 5 [4–5] HLH 2004 criteria and their HScore was 244 [221–276]. Fever was almost constant and histological hemophagocytosis was found in 68% of the patients. HLH etiology was hematological malignancy in 35 patients (74%), infectious disease in 7 patients (15%) and auto-immune disease in 2 patients (4%). Three additional patients had an alternative diagnosis or unknown etiology. Thirty patients (64%) presented coagulation disorders and 11 (23%) experienced a bleeding event. At day 1, fibrinogen level was 2.65 g/L [1.61–5.66], ADAMTS13 activity 22% [12–33], PT 64% [48–72], fibrin degradation products 8.69 [5–31] (N < 6 µg/L), PAI-1 94.1 [45–188] (N = 4–43 ng/mL), tPA 45.2 [30.7–66.6] (N = 2–12 ng/mL). Fifteen patients (32%) required mechanical ventilation and 17 (36%) vasopressors. Etoposide was administered to 72% of the patients. Eighteen (38%) patients died during hospital stay. In multivariate analysis, the occurrence of a severe hemorrhage (OR 3.215 [1.194–8.653], p = 0.021) and SOFA score (OR 1.305 per point [1.146–1.485], p < 0.001) were associated with a higher mortality rate. No early biological marker was associated with bleeding.

Conclusion: Coagulation disorders are frequent during HLH. Severe bleedings occur in almost one in four patients and confer an increased risk of death. This is the first prospective study specifically targeting hemostasis disorders in HLH. Investigations on specialized hemostasis function are ongoing, in order to determine the mechanisms leading to coagulopathy.

Compliance with ethics regulations: Yes.

Fig. 1

Survival curve according to severe bleeding

CO-20 Thrombotic thrombocytopenic purpura related neurological manifestations: clusters at presentation and long-term prognosis

Adrien Mirouse¹, Stéphane Legriel², Sylvie Chevret¹, Agnès Veyradier¹, Lionel Galicier¹, Lara Zafrani¹, Eric Mariotte¹, Elie Azoulay¹

¹Hôpital Saint-Louis, Paris, France; ²Centre Hospitalier de Versailles, Le Chesnay, France

Correspondence: Adrien Mirouse (adrien.mirouse@aphp.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-20

Rationale: Thrombotic thrombocytopenic purpura (TTP) is a thrombotic microangiopathy with frequent neurological manifestations. Data are lacking concerning precise description of neurological manifestations, timing of symptoms, correlation between clinical manifestations and neuro-imaging, and recovery after a severe episode of neurological TTP. This study aimed at describing neurological manifestations during TTP episodes and long term neurological outcome.

Patients and methods: Prospective study in one French ICU including all adult patients with an acute TTP diagnosis with neurological manifestations. Patients were included if they had thrombocytopenia and microangiopathic hemolytic anemia with signs of visceral ischemia and an ADAMTS13 activity < 10%. Neurological clusters were identified using a non parametric unsupervised cluster analysis. Long term neurological recovery was assessed with Glasgow Outcome Scale (GOS).

Results: 108 patients were included from 1997 to 2019. Neurological symptoms were migraine-like symptoms (64%), limb weakness/paresthesia (49%), pyramidal syndrome (39%), confusion or cognitive impairment (34%), obtundation (31%), seizure (19%), visual symptoms (20%), and cerebellar syndrome (18%). Time between neurological symptoms and ICU admission was 7 [3–18] days. A cerebral CT-scan performed in 48 (44%) patients and an MRI in 67 (62%) were abnormal in 9 (19%) and 27 (40%) cases, respectively. Twenty-seven (25%) patients had an electroencephalogram, abnormal in 12 (44%) cases. Three clusters of patients were identified. Cluster 1 included younger patients (37 [27–48] vs. 41 [32–52] and 48 [35–54], p = 0.045), with headaches (75% vs. 27% and 36%, p < 0.0001). Cluster 2 patients presented ataxic gait and cerebellar syndrome (77% vs. 0% and 0%, p < 0.0001), and dizziness (50% vs. 0% and 0%, p < 0.0001). Cluster 3 patients presented confusion (36% vs. 0% and 9%, p < 0.0001), obtundation (58% vs. 0% and 24%, p < 0.0001), and seizure (36% vs. 0% and 14%, p < 0.0001). All patients were treated with plasma exchange therapy. Median ICU length-of-stay was 8 [6–16.5] days. During ICU management, 31 (29%) patients required mechanical ventilation, 18 (17%) vasopressor use, and 16 (15%) renal replacement therapy. Six (6%) patients died in ICU. After a median follow-up of 34 [12–71] months, 100 (93%) patients were alive. Patients from cluster 1 were more frequently GOS 5 compared to cluster 2 and 3 at 3 months (44 [98%] vs. 13 [65%] and 21 [60%], p < 0.0001), 6 months (44 [100%], 15 [68%], and 23 [69%], p < 0.0001), and 1 year (40 [100%] vs. 15 [79%] and 20 [57%], p < 0.0001).

Conclusion: Neurological recovery may be delayed after a TTP episode. One year full neurological recovery range from 57% to 100% depending on neurological initial presentation.

Compliance with ethics regulations: Yes.

Table 1 Patients outcome according to cluster

CO-21 Relationship between diaphragm thickening fraction and transdiaphragmatic pressure in healthy and mechanically ventilated patients: a breath by breath analysis

Thomas Poulard¹, Quentin Fossé², Jean-Luc Gennisson³, Marie-Cécile Niérat², Jean-Yves Hogrel¹, Thomas Similowski², Alexandre Demoule², Damien Bachasson¹, Martin Dres²

¹Institute of Myology, Neuromuscular Investigation Center, Neuromuscular Physiology Laboratory, Paris, France; ²AP-HP, Groupe Hospitalier Pitié-Salpêtrière Charles Foix, Service de Pneumologie, Médecine Intensive et Réanimation, (Département “R3S”), Paris, France; ³Imagerie par Résonance Magnétique Médicale et Multi-Modalités (IR4M), CNRS UMR8081, Université Paris-Saclay, Orsay, France

Correspondence: Thomas Poulard (t.poulard@institut-myologie.org)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-21

Rationale: Diaphragm thickening fraction (TFdi) measured by ultrasound (US) is widely used in clinical research to evaluate diaphragm function in order to guide clinicians in providing optimal ventilator support. Studies reported TFdi cut-off values that could help in predicting weaning outcome in mechanically ventilated (MV) patients, but surprisingly, very little is known on the relationship between TFdi and the changes in transdiaphragmatic pressure (ΔPdi), the reference method. The present study investigated the relationship between ΔPdi and TFdi in healthy subjects and in MV patients.

Patients and methods: Pdi was monitored with gastric and esophageal catheters and US was performed at the zone of apposition of the right hemi-diaphragm. Healthy subjects breathed against an external inspiratory threshold load of 0–50% of maximal inspiratory pressure. MV patients were tested under several ventilator assistances before performing a spontaneous breathing trial. A breath by breath analysis was performed to confront ΔPdi and TFdi for a given breathing cycle. Pearson correlation coefficients (r) were used to determine within-individual and overall relationships between ΔPdi and TFdi.

Results: Fifteen healthy volunteers and 22 MV patients were studied. One healthy subject displayed a significant positive correlation between ΔPdi and TFdi (r = 0.32, p < 0.01). Only three of the 22 MV patients presented with a significant positive correlation between ΔPdi and TFdi (mean r = 0.67, 95% CIs [0.41, 0.83] in patients with significant ΔPdi-TFdi correlation, all p < 0.01). Overall relationship between ΔPdi and TFdi in MV patients was weak (R = 0.17, 95% CIs [0.04, 0.29], p < 0.05). Individual relationships between ΔPdi and TFdi and averaged values for every condition tested are presented in Fig. 1.

Conclusion: These findings show that TFdi does not generally correlate with Pdi suggesting that TFdi may not be used as a surrogate for Pdi. The explanations for this lack of correlation deserve further studies.

Compliance with ethics regulations: Yes.

Fig. 1

Intra-individual relationships between ΔPdi and TFdi for mechanically ventilated (MV) patients (A) and healthy subjects (C). Relationships between ΔPdi and TFdi when breathing cycles were averaged for all participants during each condition for MV patients (B) and healthy subjects (D). − 25%: initial settings minus 25% inspiratory help, + 25%: initial settings plus 25% more inspiratory help, PEP 0: zero positive end-expiratory pressure, SBT: spontaneous breathing trial. Healthy subjects performed spontaneous breathing (SB) and ventilation against inspiratory threshold at 10, 20, 30, 40 and 50% of maximal inspiratory pressure (MIP)

CO-22 Accuracy of clinical estimation of the respiratory effort under pressure support ventilation

Samuel Tuffet, François Perier, Anne-Fleur Haudebourg, Nicolas De Prost, Keyvan Razazi, Armand Mekontso Dessap, Guillaume Carteaux

Réanimation médicale CHU Henri Mondor, Créteil, France

Correspondence: Samuel Tuffet (samuel.tuffet@hotmail.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-22

Rationale: One of the main aims of partial ventilatory support is to maintain the patient’s respiratory effort within a normal range. In clinical routine, the respiratory effort is assessed by clinical judgment, whose accuracy is unknown. In this study, we assessed the accuracy of clinical estimation of the respiratory effort under pressure support ventilation (PSV).

Patients and methods: In this prospective monocenter study, patients under mechanical ventilation were included within the first 2 days after switching from assist control ventilation to PSV. Flow, airway pressure, and esophageal pressure were recorded and respiratory effort indices were computed using the FluxMed device (MBMED). Respiratory effort was classified as normal (esophageal pressure time product (PTPes) between 50 and 150 cmH₂O s min⁻¹ (1)), insufficient (PTPes < 50 cmH₂O s min⁻¹) or excessive (PTPes > 150 cmH₂O s min⁻¹). A senior physician, a resident and a nurse were independently asked to clinically evaluate patient’s respiratory effort using a five levels scale: very low, low, normal, high and very high respiratory effort. Data are reported as median [1st–3rd quartile] or number (percentage).

Results: Thirty patients, aged 65 years [57–71], with a median SOFA score of 6 [4–9], were included so far. They had spent 9 days [6–12] under mechanical ventilation before inclusion. Median PTPes was 175 cmH2O s min− 1 [98–244]. At the time of assessment, respiratory effort was insufficient in two patients (7%), normal in 10 (33%), and excessive in 18 (60%). Senior physician’s clinical estimation misclassified the respiratory effort in 12 patients (40%). Sensibility/specificity of the senior physician estimation to detect insufficient, normal and excessive respiratory effort were 0.50/0.89, 0.50/0.70 and 0.67/0.75, respectively. Accuracy of the resident and nurse evaluations was consistent with that of senior physician. By univariate analysis, higher SOFA score at inclusion (p = 0.02) and higher age (p = 0.03) were significantly associated with non-detection of excessive respiratory effort.

Conclusion: During the 2 days after switching from assist control ventilation to PSV, our preliminary data suggest that: 1/excessive respiratory efforts are frequently observed and 2/the clinical judgment frequently fails to adequately classify the level of respiratory effort, making the excessive efforts underdiagnosed.

Compliance with ethics regulations: Yes.

Reference

• 1. The PLeUral pressure working Group (PLUG—Acute Respiratory Failure section of the European Society of Intensive Care Medicine), Mauri T, Yoshida T, Bellani G, Goligher EC, Carteaux G, et al. Esophageal and transpulmonary pressure in the clinical setting: meaning, usefulness and perspectives. Intensive Care Medicine. sept 2016;42 (9):1360–73.

CO-23 Influence of body mass index on respiratory mechanics in acute respiratory distress syndrome: a multicenter cohort study

Rémi Coudroy¹, Jean-Luc Diehl², Damien Vimpere³, Nadia Aissaoui³, Romy Younan³, Clotilde Bailleul³, Amélie Couteau-Chardon³, Aymeric Lancelot³, Emmanuel Guérot³, Chen Lu⁴, Laurent Brochard⁴

¹Poitiers Hospital, Poitiers, France; ²Assitance Publique Hopitaux de Paris, Paris, France; ³Georges Pompidou European Hospital, Paris, France; ⁴Interdepartmental Division of Critical Care Medicine, Toronto, Canada

Correspondence: Rémi Coudroy (r.coudroy@yahoo.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-23

Rationale: Overweight and obesity are increasingly prevalent worldwide and account for about 30–40% of patients with acute respiratory distress syndrome (ARDS). How body mass index (BMI) influences respiratory mechanics in ARDS is unclear.

Patients and methods: This study is a secondary analysis of 2 cohorts of ARDS according to the Berlin definition: a bicenter Canadian study of 45 ARDS of any BMI enrolled in a prospective study (NCT02457741), and a French monocenter cohort of selected ARDS with a BMI > 40 kg/m2. Airway pressure, flow and esophageal pressure were collected in all patients and we report data at a set positive end-expiratory pressure (PEEP) of 5 cmH2O. Presence of complete airway closure and airway opening pressure were assessed using a low-flow inflation pressure–volume curve. End expiratory lung volume (EELV) was measured using the nitrogen washout/washin technique. The ratio EELV to predicted functional residual capacity was calculated. Patients were sorted in 3 groups according to the World Health Organization overweight classification (BMI < 30, from 30 to < 40, and ≥ 40 kg/m2).

Results: Among the 54 patients included, 18 patients (34%) had BMI < 30 kg/m2, 16 (30%) between 30 and 40 kg/m2, and 19 (36%) ≥ 40 kg/m2. The median PaO2/FiO2 was 138 mmHg with a PEEP of 15 cmH2O, and mortality was 32% without difference across groups. Airway closure occurrence increased with BMI (22%, 38% and 58%, p = 0.04). When present, airway opening pressure was 9.6 cmH2O (8.5–13.2) and similar between the 3 groups. With increasing BMI, total PEEP increased from 6.0 to 9.0 cmH2O between groups (p = 0.02). All values of esophageal pressure increased with BMI. End-expiratory esophageal pressure was strongly correlated with BMI (rho = 0.71, p < 0.001), as illustrated in Fig. 1. Consequently end-expiratory transpulmonary pressure decreased from − 2.7 to − 9.3 cm H2O with increasing BMI (p = 0.008). The ratio of EELV to predicted functional residual capacity was negatively correlated with end-expiratory pressure (Rho = − 0.39, p = 0.01), but not with BMI. Driving pressure and elastance of the respiratory system, chest wall and lung were similar across all ranges of BMI. Likewise, EELV was similar between groups.

Conclusion: In ARDS, increasing BMI is associated with increased occurrence of airway closure and increased values of esophageal pressure. Conversely, chest wall elastance is not influenced by BMI, as well as lung elastance. Including BMI in interpreting respiratory mechanics in ARDS patients can provide additional information for the clinical management.

Compliance with ethics regulations: Yes.

Fig. 1

Relationship between body mass index and end-expiratory esophageal pressure measured at a set positive end-expiratory pressure of 5 cmH₂O

CO-24 Impact of tidal volume during the “transition period” following neuromuscular blockade cessation in ARDS: an observational study

Safaa Nemlaghi¹, Anne-Fleur Haudebourg², François Perier², Nicolas De Prost², Keyvan Razazi², Guillaume Voiriot¹, Muriel Fartoukh¹, Armand Mekontso Dessap², Guillaume Carteaux²

¹Médecine Intensive Réanimation, Hôpital Tenon, Paris, France; ²Réanimation Médicale, CHU Henri Mondor, Créteil, France

Correspondence: Safaa Nemlaghi (safaa.nemlaghi@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-24

Rationale: Low tidal volume is the cornerstone of protective ventilation in the initial phase of ARDS (1). Whether such low tidal volume can still be achieved when the patient is allowed to breathe spontaneously under pressure support ventilation (PSV) is unknown. In moderate-to-severe ARDS patients receiving neuromuscular blockade, we assessed the tidal volume and its potential association with the outcome during the “transition period” following neuromuscular blockade.

Patients and methods: Retrospective observational study in two university intensive care units. Patients fulfilling moderate-to-severe ARDS criteria less than 72 h after intubation and receiving neuromuscular blockers were included upon entry in the “transition period”. We defined the “transition period” as the 72 h following neuromuscular blockers cessation. Ventilatory and hemodynamic parameters were recorded every 3 h during the “transition period”. Primary outcome was the association between mean tidal volume under pressure support ventilation (PSV) during the “transition period” and the 28-day mortality after adjustment for confounding factors. Data are reported as median [1st–3rd quartile] or number (percentage).

Results: One hundred nine patients were included, with a PaO2/FiO2 ratio of 100 mmHg [76–165] at intubation and 194 mmHg [158–244] at inclusion and a SOFA score at 7 [4.5–10]. Patients had been ventilated 2 days [1–3.8] before inclusion. During the “transition period”, 88 patients (80.7%) were switched to PSV. The median duration of PSV was 42 h [27–48]. The mean tidal volume under PSV was significantly lower in survivors than in non survivors at day 28 (7.1 ml/kg [6.3–7.9] vs. 7.8 ml/kg [6.8–9.4] respectively, p = 0.007). By multivariate analysis (Cox proportional hazards regression model), mean tidal volume during PSV remained independently associated with the 28-day mortality after adjusting for SOFA score and immunosuppression. Patients with a mean tidal volume above 8 mL/kg under PSV during the “transition period” had a lower cumulative probability of survival at day 28 as compared with others (Log rank test, p = 0.008) (Fig. 1).

Conclusion: In patients with moderate-to-severe ARDS, a higher tidal volume under PSV within the 72 h following neuromuscular blockers cessation is independently associated with the 28-day mortality.Compliance with ethics regulations: Yes.

Fig. 1

Kaplan-Meier estimate of the cumulative probability of survival according to the mean tidal volume (Vt)—lower of higher than 8 ml/kg—under pressure support ventilation (PSV) during the “transition period”

Reference

• 1. Papazian L, Aubron C, Brochard L, Chiche J-D, Combes A, Dreyfuss D, et al. Formal guidelines: management of acute respiratory distress syndrome. Ann Intensive Care. 2019 Dec;9 (1):69.

CO-25 Impact of Perioperative Anemia and red blood Cells Transfusion on post-operative complications after oncological surgery

Xavier Chapalain¹, Yves Ozier¹, Catherine Le Niger², Olivier Huet¹, Cécile Aubron³

¹Department of anesthesiology and intensive care unit, CHRU de Brest Hôpital de la Cavale Blanche, Brest, France; ²Haemovigilance Unit, CHRU de Brest, Brest, France; ³Medical Intensive Care unit, CHRU de Brest, Brest, France

Correspondence: Xavier Chapalain (xavier.chapalain@chu-brest.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-25

Rationale: Between 36 and 75% of oncologic patients experience anemia and anemia is associated with poor prognosis. Up to 40% of surgical oncologic patients receive red blood cells (RBC). However, transfusion is associated with adverse events, and equipoise remains on the optimal transfusion strategy in oncologic patients in surgical setting.

Patients and methods: This is a retrospective, single center study. All adults admitted to the intensive care unit (ICU) after oncologic surgery from January 2017 to December 2018 were eligible. The following types of surgery for cancer or metastasis resection with a high risk of bleeding were eligible: thoracic, abdominal, neurosurgery, gynecologic, urologic, otorhinolaryngology or spinal surgery. The primary outcome was a composite outcome including post-operative complications (respiratory, cardiac, renal, thromboembolic, infectious and/or hemorrhagic) and/or hospital mortality.

Results: Of the 287 patients included, 142 patients (49.5%) had anemia (based on the WHO definition: hemoglobin level 10–11.9 g/dl for female; hemoglobin level 10–12.9 g/dl for male), 69 patients (24%) had moderate anemia (hemoglobin level: 8–9.9 g/dl) and 32 patients (12.5%) severe anemia (hemoglobin level < 8 g/dl). Fifty-six patients (19.6%) received at least one RBC transfusion during their hospital stay. Patients exposed to moderate and severe anemia required more often renal replacement therapy (RRT) for acute kidney injury (AKI) (1.1% vs. 10.9%; p = 0.003), had more surgery-related infections (7.6% vs. 22.8%; p = 0.004). Patients who received RBC had more often AKI with RRT (0.9% vs. 19.6%; p < 0.001), thromboembolic events (2.2% vs. 8.9%; p = 0.039), sepsis (4.3% vs. 16.1%; p = 0.004), pneumonia (4.8% vs. 14.3%; p = 0.024), surgical site infections (8.7% vs. 30.4; p < 0.001) and second surgery for infection (3% vs. 16.1%; p = 0.001). The multivariate analysis found an association between moderate and severe anemia (moderate anemia: OR 15.03 [2.73–282.3]; severe anemia: OR 16.65 [2.71–325.7]; p = 0.011) and severe post-operative complications (Fig. A). There was also an association between RBC transfusion and severe post-operative complications (OR 4.3 [2.2–8.8]; p < 0.001) (Fig. B).

Conclusion: Anemia was frequent in oncologic surgical patients. Anemia, including moderate anemia, was independently associated to patient outcomes; however, RBC transfusion also negatively impacts on patients’ prognosis. Our study highlights the need for further research to identify the optimal hemoglobin threshold for RBC transfusion in surgical oncologic patients.

Compliance with ethics regulations: Yes.

Fig. 1

Multivariate analysis analyzing the risk factors for the primary outcome (severe post-operative complications and/or mortality) including anemia (a) or transfusion (b)

CO-26 Impact of blood products transfusions on the susceptibility to ICU-acquired infections in septic shock

Edwige Péju¹, Jean-FrançOis Llitjos¹, Jean-Paul Mira¹, Jean-Daniel Chiche¹, Alain Cariou¹, Julien Charpentier¹, Matthieu Jamme², Frédéric Pene¹

¹Centre Hospitalier COCHIN (APHP), Paris, France; ²Centre Hospitalier TENON (APHP), Paris, France

Correspondence: Edwige Péju (edwigepeju@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-26

Rationale: Transfusions of blood products are common in critically ill patients, and have a potential for immunomodulation. The aim of this study is to address the impact of transfusion of blood products on the susceptibility to intensive care unit (ICU)-acquired infections in the high-risk patients with septic shock.

Patients and methods: This single-center retrospective study was conducted in the medical ICU of a tertiary care center over a 10-year period (January 2008 to December 2017). All consecutive patients diagnosed for septic shock within the first 48 h of ICU admission were included. Patients who were discharged or died within the first 48 h were excluded. The amounts of red blood cell, platelet and fresh frozen plasma transfusions were collected in all patients, and eventually censored at the onset of ICU-acquired infection. Statistical analysis was carried out using a multivariate time dependent Cox cause-specific proportional hazard model.

Results: 1152 patients were admitted for septic shock, with 893 patients remaining alive in the ICU after 48 h of management. The median age was 68 years (58–78), with a majority of male patients (63.7%). Thirty-six percent of them were deemed immunocompromised at the time of ICU admission. A first episode of ICU-acquired infection occurred in 28.3% of the 48-h survivors, with a majority of pulmonary infections (57%). Patients with ICU-acquired infections were more likely to have received red blood cell (52% vs. 38%, p < 0.001), platelet (28% vs. 17%, p = 0.009) and fresh frozen plasma (26% vs. 14%, p < 0.001) transfusions beforehand. In multivariate analysis, transfusion of platelets (OR = 1.45 [1.06–1.99], p = 0.02) and fresh frozen plasma (OR = 1.40 [1.02–1.93], p = 0.03) were independently associated with the occurrence of ICU-acquired infections.

Conclusion: Transfusions of platelet and fresh frozen plasma account for independent risk factors of ICU-acquired infections in patients recovering from septic shock. These results support a restrictive transfusion policy in critically ill patients, though indications of platelet and fresh frozen plasma transfusions have not been well investigated in this setting.

Compliance with ethics regulations: Yes.

CO-27 Exploring the microvascular impact of Red Blood Cell Transfusion in intensive care unit patients

Geoffroy Hariri¹, Simon Bourcier¹, Zora Marjanovic², Jérémie Joffre¹, Jeremy Lemarie³, Jean-Rémi Lavillegrand¹, Dominique Charue³, Thomas Duflot⁴, Naike Bige¹, Jean-Luc Baudel¹, Eric Maury¹, Mohamad Mohty², Bertrand Guidet¹, Jeremy Bellien⁵, Olivier Blanc-Brude³, Hafid Ait-Oufella¹

¹Medecine intensive Reanimation, Hôpital Saint-Antoine, Paris, France; ²Service d’hématologie, Hôpital Saint-Antoine, Paris, France; ³Inserm U970, Paris, France; ⁴Inserm U1096, Rouen, France; ⁵Medecine Inserm U1096, Rouen, France

Correspondence: Geoffroy Hariri (geoffroyhariri@hotmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-27

Rationale: Red blood cell (RBC) transfusion is a common treatment for hospitalized patients. However, the effects of RBC transfusion on microvascular function remain controversial.

Patients and methods: In a medical intensive care unit (ICU) in a tertiary teaching hospital, we prospectively included anemic patients requiring RBC transfusion. Skin microvascular reactivity was measured before and 30 min after RBC transfusion. Plasma was collected to analyze intravascular hemolysis and draw the lipidomic and cytokine profiles.

Results: In a cohort of 59 patients, median age was 66 [55–81] years and SAPS II was 38 [24–48]. After RBC transfusion, endothelium-dependent microvascular reactivity improved in 35 (59%) patients, but worsened in 24 others (41%). Comparing clinical and biological markers revealed that baseline blood leucokyte counts distinguished improving from worsening patients (10.3 [5.7; 19.7 vs. 4.6 [2.1; 7.3.109/L; p = 0.001) and correlated with variations of microvascular reactivity (r = 0.36, p = 0.005). Blood platelet count was also higher in improving patients (200 [97; 280] vs 160 [40; 199].103/mL, p = 0.03) but did not correlate with variations of microvascular reactivity. We observed no intravascular hemolysis (HbCO, heme, bilirubin, LDH), but recorded a significant increase in RBC microparticle levels in improving patients after transfusion (292 [108; 531] vs. 53 [34; 99] MP/µL; p = 0.03). The improve in microvascular dilation was positively correlated with RBC microparticle levels (R = 0.83, P < 0.001) and conversion of arachidonic acid into vasodilating eicosanoids.

Conclusion: Patients displaying an improved microvascular reactivity after RBC transfusion, had high blood leukocyte counts, increased RBC microparticles formation and enhanced metabolism of arachidonic acid into vasodilating lipids. Our data suggested a contribution of recipient leukocytes to the vascular impact of RBC transfusion.

Compliance with ethics regulations: Yes.

CO-28 Effectiveness of a transfusion protocol to reduce the inappropriate use of red blood cell transfusions in critically ill patients: a before and after study

Eric Thiry, Eloïse Natalis, Deeba Ali, Julien Guntz, Philippe Devo, Pierre Demaret

chc liège, liège, belgium

Correspondence: Eric Thiry (eric.thiry@chc.be)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-28

Rationale: Red blood cell transfusions (RBCTs) are frequently prescribed in intensive care unit (ICU), but they are an expensive and scarce resource and they are not without risks. Appropriate transfusion strategies are thus required to optimize their use, and interventions aiming to reduce the rate of inappropriate transfusions should be implemented and evaluated.

Patients and methods: Prospective observational before-and-after study conducted in a 32-bed ICU affiliated to a non-university teaching hospital. All RBCTs were prospectively recorded during the 4-month-long phases I and III while a transfusion protocol was implemented in the ICU during the 2-month-long phase II. An RBCT was considered as inappropriate if not prescribed in accordance with the protocol.

Results: While the number of admissions to ICU was similar between phase I (745 admissions) and phase III (743 admissions), 236 RBCTs were prescribed during phase I versus 191 during phase III, corresponding to a 19%-reduction. We retained 358 RBCTs for analysis (203 during phase I, 155 during phase III). The mean (± standard deviation) pre-transfusion hemoglobin (Hb) was 72 ± 10.5 g/L during phase I, indicating restrictive transfusion practices. Despite this, the proportion of inappropriate RBCTs decreased from 100/203 (49.3%) during phase I to 59/155 (38%) during phase III (p = 0.035). Admission during phase III was independently associated with a reduced risk of inappropriate RBCT (adjusted OR 0.5, 95% confidence interval 0.3–0.85, p = 0.01).

Discussion: Our data indicate that an RBCT protocol may improve transfusion practices. However, in a recent international survey, only 29% of the respondents stated that they had an ICU-specific transfusion protocol in their unit. This simple tool is thus probably underused: our findings could help to increase its adoption. Several limitations of our study have to be acknowledged. It may be difficult to assess the appropriateness of some RBCT and our binary classification is probably not appropriate to reflect the complexity of the decisional process leading to some RBCTs. Furthermore, our study is susceptible to selection bias from secular changes in practice from phase I to phase III, and to the Hawthorne effect, i.e. an initial improvement in performance created by the act of observing the performance. Despite these limitations, our findings support the implementation of protocols to improve transfusion practices in ICU.

Conclusion: The implementation of an RBCT protocol in our ICU was associated with a 19%-reduction in the number of RBCT prescribed and was independently associated with a 50%-reduction in the risk of being inappropriately transfused.

Compliance with ethics regulations: Yes.

CO-29 Effect of extracorporeal CO₂ removal on right ventricular function in patients with acute respiratory distress syndrome: the ECCO2Rea pilot study

Suzanne Goursaud, Xavier Valette, Julien Dupeyrat, Cedric Daubin, Damien Du Cheyron

CHU de Caen Normandie, Service de Réanimation Médicale, 14000 Caen, France

Correspondence: Suzanne Goursaud (goursaud.suzanne@wanadoo.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-29

Rationale: Right ventricular (RV) failure is a common complication in moderate to severe acute respiratory distress syndrome (ARDS). RV failure is exacerbated by hypercapnic acidosis and overdistension induced by mechanical ventilation. Veno-venous extracorporeal CO2 removal (ECCO2R) might allow ultraprotective mechanical ventilation strategy with a low tidal volume (VT) and plateau pressure (Pplat). This study investigated if ECCO2R therapy could have beneficial effects on RV function.

Patients and methods: This prospective monocentric pilot study was conducted in a French ICU from January 2017 to March 2019. Patients with moderate to severe ARDS with PaO2/FiO2 ratio between 80 to 150 mmHg were enrolled. Ventilation parameters, arterial blood gases, echocardiographic parameters performed by transthoracic echocardiography (TTE), low-flow ECCO2R system operational characteristics, outcomes and adverse events were collected during the protocol. Primary end point was evolution of RV echocardiographic parameters with ultraprotective ventilation strategy at 4 mL/Kg PBW during the 24-h following the start of ECCO2R.

Results: Eighteen patients were included. Efficacy of ECCO2R allowed an ultraprotective strategy in all patients. We observed a significant improvement of RV systolic function parameters assessed by TTE (Fig. 1). Tricuspid annular plane systolic excursion (TAPSE) increased significantly under ultraprotective ventilation compared to baseline (from 22.8 to 25.4 mm; p < 0.05). Systolic excursion velocity (S’) also increased after 1-day protocol (from 13.8 m/s to 15.1 m/s; p < 0.05). A significant improvement of aortic velocity time integral (VTIAo) under ultraprotective ventilation settings was observed. There were no significant differences in the values of systolic pulmonary arterial pressure (sPAP). When patients were separated in two groups according to baseline PaCO2 level above or under 50 mmHg, we showed the deleterious effect of hypercapnia on RV function, and observed in both groups a beneficial impact of an ultraprotective ventilation strategy on TAPSE. No severe adverse events directly related to ECCO2R were observed in our small cohort.

Conclusion: The low-flow ECCO2R allows ultraprotective ventilation strategy and improve RV function in moderate to severe ARDS patients. Similarly to prone positioning, ECCO2R could become a strategy that enables to reconcile lung protective approach with RV protective approach in ARDS patients. Large-scale clinical studies, including patients with severe RV dysfunction, will be required to confirm these results and to assess the overall benefits, in particular the best timing of beginning ECCO2R in ARDS patients.

Compliance with ethics regulations: Yes.

Fig. 1

Time course of echocardiography variables during the study period

CO-30 Typology of Published Randomized Controlled Trials in Critically Ill Patients with Acute Respiratory Failure

Guillaume Dumas¹, Sylvie Chevret², Marine le Corre³, Elie Azoulay¹

¹Medical Intensive Care Unit, Saint-Louis University Hospital, Paris, France; ²ECSTRA team, Biostatistics and clinical epidemiology, UMR 1153 (center of epidemiology and biostatistic Sorbonne Paris Cité, CRESS), INSERM, Paris Diderot University, Paris, France; ³Department of Anesthesiology and Critical Care, Groupe Hospitalier Pitié-Salpêtrière Charles Foix, Paris, France

Correspondence: Guillaume Dumas (dumas.guillaume1@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-30

Rationale: We conducted a methodological review of published RCTs on ARF to report success rates according to primary and secondary endpoints.

Patients and methods: We searched MEDLINE, Cochrane Central Register of Controlled Trials, and Web of Science for RCTs in critically ill patients with ARF. Published trials in adult patients between January 1995 and December 2018 were evaluated. Clinical endpoints were classified into 3 different categories, namely, ventilation-based endpoints, clinically-based endpoints, or mortality.

Results: Seventy-one RCTs were included, which mainly compared oxygenation/ventilation strategies (94%), mostly in patients with various causes of ARF (63%), most often (58%) performed in a single center, and stopped prematurely in 20% of the cases. Twenty-two distinct primary endpoints were used. Classified into three different categories, the most frequently used primary endpoints were ventilation-based (43 RCTs, 60%), and consisted in the need for intubation (79%) assessed at 6 different timepoints followed by clinically-based endpoints in 19 trials (27%), and mortality in 9 (13%). Overall, 41 (58%) RCTs were positive (i.e. demonstrated a significant difference on the primary endpoint). A positive result was reported in 53% of the studies with ventilation-based endpoints, 79% in studies with clinically-based endpoints, and 33% in studies swith mortality endpoints. Adjusted on study quality, 3 factors were associated with a positive result: clinically-based primary endpoints (Odds Ratio (OR): 7.20, 95% Confidence Interval (95% IC): 1.08–48.17, p-value: 0.04), the use of a predefined standard of care in both the control and the intervention groups (OR: 4.60, 95% IC: 1.05–22.79, p-value:) and single center trials (OR: 3.17, 95% IC: 1.03–10.66, p-value:0.05).

Conclusion: In critically ill patients with ARF, the majority of published RCTs assessed the effectiveness of oxygenation/ventilation strategies, using the need for intubation as the primary endpoint, and were performed in a single center. The use of a predefined standard of care to manage patients in both groups is associated with more frequent positive trials. These results should be used to frame future trial designs in this field, and guide clinicians and researchers towards optimal research transfer to the bedside.

Compliance with ethics regulations: Yes.

CO-31 Impact of radiation therapy exposure on outcome of cancer patients admitted in an intensive care unit for acute respiratory failure

Lucile Martin¹, Michaël Darmon², Virginie Lemiale², Alexandre Demoule³, Frédéric Pene⁴, Anne-Pascale Meert⁵, Achille Kouatchet⁶, djamel MOKART⁷, Fabrice Bruneel⁸, Elie Azoulay², Martine Nyunga¹

¹Centre Hospitalier de Roubaix, Roubaix, France; ²Hôpital Saint-Louis, Paris, France; ³Hôpital Pitié-Salpêtrière, Paris, France; ⁴Hôpital Cochin, Paris, France; ⁵Institut Jules Bordet, Brussels, Belgium; ⁶Centre Hospitalier Régional Universitaire, Angers, France; ⁷Institut Paoli Calmettes, Marseille, France; ⁸Hôpital André Mignot, Versailles, France

Correspondence: Lucile Martin (lucile.martin.lm@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-31

Rationale: Cancer patients are increasingly admitted to the intensive care unit (ICU), and some have been exposed to radiation therapy (RT). Indeed, up to 40% of a cancer population is concerned by radiotherapy. The aim of our study was to assess impact of RT on outcome of these patients admitted in ICU for acute respiratory failure (ARF).

Patients and methods: We studied a cohort of cancer patients admitted in ICU for ARF. Data were extracted from 5 prospective and retrospective datasets performed by our study group. Results are reported as median (IQR) or n (%). Adjusted analyses were performed using cox model, then double adjustment (propensity score matching on risk of receiving RT then Cox model).

Results: Of the 490 patients, 151 (31%) have been exposed to RT. Preexisting history of surgery (22.1% vs 37.7% in patients without and with RT respectively, p = 0.001), female gender (43.4 vs. 53.0% p = 0.05) and allogeneic stem cell recipients (11.5% vs. 20.5% p = 0.012) had more frequently previous exposure to radiation therapy. Mortality rate on day 28 was 39.7%. RT was not associated with outcome (mortality of respectively 33.1% vs 41.9% in patients with and without RT, p = 0.0875). Lung cancer (p = 0.005) and need for vasopressors (p < 0.0001) were independently associated with mortality. After propensity score for matching, two groups of 122 patients were compared. Then RT was associated to survival (mortality 36.1% vs 52.5% in patients with and without RT respectively, p = 0.01), and did not impact clinical presentation of ARF. After adjustment on confounders, age (HR = 1.03 95% IC = [1–1.07] p = 0.033) and need for vasopressors (HR = 2.48 95% IC = [1.19–5.18] p = 0.015) were associated with mortality. RT was independently associated with survival (HR = 0.51 95% IC = [0.32–0.83] p = 0.0065).

Conclusion: This assessment is the first overview of the impact of radiotherapy in this population. Results suggest that exposure to radiation therapy does not worsen prognosis nor survival of cancer patients admitted in the ICU for ARF, regardless the underlying malignancy. Furthermore, radiotherapy is not associated with significant differences in the clinical presentation of respiratory failure. In this study, RT exposure in cancer patients admitted to the ICU for ARF is not associated with a poor prognosis after adjustment for confounders nor to changes in ARF features.

Compliance with ethics regulations: Yes.

Fig. 1

Probability of survival at day-28, after matching

CO-32 Benefit-to-risk balance of bronchoalveolar lavage in the critically ill. A prospective, multicenter cohort study

Toufik Kamel¹, Julie Helms², Ralf Janssen-Langenstein³, Achille Kouatchet⁴, Antoine GUILLON⁵, Jeremy Bourenne⁶, Damien Contou⁷, Christophe Guervilly⁸, Rémi Coudroy⁹, Marie-anne Hoppe¹⁰, Jean-Baptiste Lascarrou¹¹, Jean-Pierre Quenot¹², Gwenhael Colin¹³, Paris Meng¹⁴, Jerome Roustan¹⁵, Christophe Cracco¹⁶, Mai-anh Nay¹, Thierry Boulain¹

¹Centre Hospitalier Régional d’Orléans, Service de Médecine Intensive Réanimation, Orléans, France; ²CHU de Strasbourg-Hôpital Civil, Service de Réanimation Médicale, Strasbourg, France; ³Médecine Intensive Réanimation, Hôpital de Haute pierre, Hôpitaux Universitaires de Strasbourg, Strasbourg, France; ⁴CHU d’Angers, Service de Réanimation Médicale et de Médecine Hyperbare, Angers, France; ⁵CHRU de Tours-Hôpital Bretonneau, Service de Réanimation Polyvalente, Tours, France; ⁶Médecine Intensive Réanimation, Réanimation des Urgences, Hôpital La Timone, Marseille, France; ⁷CH d’Argenteuil, Service de Réanimation Polyvalente, Argenteuil, France; ⁸Assistance Publique-Hôpitaux de Marseille, Hôpital Nord, Médecine Intensive Réanimation, Marseille, France; ⁹Médecine Intensive et Réanimation, CHU de Poitiers, Poitiers, France; ¹⁰CH de La Rochelle-Hôpital Saint-Louis Service de Réanimation Polyvalente, La Rochelle, France; ¹¹Service de Médecine Intensive Réanimation, CHU de Nantes, Hôtel Dieu, Nantes, France; ¹²CHU de Dijon-Complexe du Bocage, Dijon, France; ¹³CHD Vendée–Hôpital de la Roche-sur-Yon, La Roche-Sur-Yon, France; ¹⁴Hôpital Raymond Poincaré, APHP, Service de Médecine intensive Réanimation, Garches, France; ¹⁵Centre hospitalier de Montauban, service de réanimation polyvalente, Montauban, France; ¹⁶CH d’Angoulême Service de Réanimation Polyvalente, Angoulême, France

Correspondence: Toufik Kamel (toufik.kamel@chr-orleans.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-32

Rationale: Bronchoalveolar lavage (BAL) is usually deemed to allow the diagnosis of a large array of pulmonary diseases and is usually considered as well tolerated in intensive care unit (ICU) patients. However, recent data suggest that the diagnostic yield of BAL could be rather low (1), and may question its innocuity (2). The present study aimed at assessing the benefit-to-risk balance of BAL in ICU patients.

Patients and methods: The study was approved by the appropriate Ethics Committee and registered with clinicaltrials.gov (NCT03098888). In 16 ICUs, from April 2017 to October 2018, we prospectively collected adverse events (AE) during or within 24 h after BAL and assessed the BAL input for decision-making in consecutive adult patients. AEs were categorized in 5 grades of increasing severity. The occurrence of a clinical AE at least of grade 3, i.e. sufficiently severe to need therapeutic action (s), including modification (s) in respiratory support, defined poor BAL tolerance. The BAL input for decision-making was declared satisfactory if it allowed to interrupt or initiate one or several treatments.

Results: We included 483 BAL in 483 patients (age 63 yrs [IQR 53–72]; female gender: 162 [33.5%]; Simplified Acute Physiology Score II: 48 [IQR 37–61]; immunosuppression 244 [50.5%], chronic pulmonary disease [163/483 (33.7%)]). BAL was performed either in non-intubated patients receiving standard O2 therapy (n = 56 [11.6%]), or noninvasive ventilation (n = 4 [0.8%]), or high-flow nasal cannula O2 therapy (45 [9.3%]), or in patients under invasive mechanical ventilation (n = 378 [78.3%]). A total of 710 AEs were observed in 415 (85.9%) patients. Sixty-seven (13.9%) patients reached the grade 3 of AE or higher. The main predictor of poor BAL tolerance identified by logistic regression was the association of a BAL performed by a non-experienced physician (non-pulmonologist, or intensivist with less than 10 years in the specialty or less than 50 BAL performed) in non-intubated patients (OR: 31.8 [95% confidence interval 11.6–87.6]; P < 0.0001). Ordinal regression also showed that when BAL was performed by a non-experienced physician in a non-intubated patient, this was associated with an increased risk of AE of any grade (OR: 12.66 [6.27–25.57]). A satisfactory BAL input for decision-making was observed in 227 (47.0%) cases and was not predictable using logistic regression.

Conclusion: Adverse events related to BAL in ICU patients are frequent, and sometimes serious. Our findings call for an extreme caution when envisaging a BAL in ICU patients and for a mandatory accompaniment of the less experienced physicians.

Compliance with ethics regulations: Yes.

CO-33 Meningitis is a rare complication of critically ill patients with severe pneumococcal community-acquired pneumonia

Paul Jaubert, Julien Charpentier, Jean-Daniel Chiche, Frédéric Pene, Alain Cariou, Guillaume Savary, Marine Paul, Jean-Paul Mira, Mathieu Jozwiak

Cochin, Paris, France; ² Mignot, Versailles, France

Correspondence: Paul Jaubert (paul.jaubert@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-33

Rationale: Severe pneumococcal community-acquired pneumonia (PCAP) is a frequent infection requiring intensive care unit (ICU) admission. Pneumococcal meningitis associated with PCAP has been reported and could worsen the prognosis of patients. However, this complication is difficult to predict and lumbar puncture is not systematically performed, regardless the severity of PCAP. Thus, we investigated the characteristics of patients with PCAP associated with pneumococcal meningitis.

Patients and methods: We retrospectively included all patients admitted for PCAP in our ICU between 2006 (inception of our electronic medical sheet) and the end of 2018. Community-acquired pneumonia was defined according to the criteria of the American Thoracic Society. We excluded all patients admitted in ICU with initial suspicion of meningitis. Variables regarding epidemiology, clinical and microbiological characteristics, management and prognosis of these patients were collected and analyzed.

Results: Among the 264 patients admitted for PCAP (62 ± 17 years old, SAPS II 55 ± 22, 59% of men), 59% of the patients required mechanical ventilation and 29% vasopressors infusion. The ICU mortality was 16%. S. pneumoniae was documented by a positive antigen test in 81% of the patient and/or by a positive sputum smear, tracheal aspirate or distal protected airway specimen in 54% of the patients, and/or by pleural aspirate in 5% of the patients and/or by positive blood culture in 32% (n = 84) of the patients. A lumbar puncture was performed in 39% (n = 33) of the patients with bacteriemia and in 30% (n = 54) of the patients without bacteriemia, with a median delay of 12 h [interquartile range: 6–32] after the onset of antibiotherapy. All lumbar punctures (n = 87) were performed for neurological signs: 50% of coma, 46% of confusion and 1% of seizures. When a lumbar puncture was performed, meningitis was diagnosed in 24% (n = 8) of the patients with bacteriemia and in 2% (n = 1) of the patients without bacteriemia (p < 0.05). The ICU mortality (22% vs. 16%, respectively), age (58 ± 19 vs. 63 ± 17 years old, respectively), SAPS II (65 ± 27 vs. 54 ± 22, respectively) or ICU length of stay (17 ± 24 vs. 11 ± 16 days, respectively) were not different between patients with and without meningitis (each p = NS).

Conclusion: Meningitis is a rare complication of PCAP and is more frequent in patients with bacteriemia. Suprisingly, meningitis is not associated with higher ICU mortality. Further analyses are ongoing to identify independent risk factors of meningitis in patients with PCAP.

Compliance with ethics regulations: Yes.

CO-34 Acute varicella zoster encephalitis admitted to the ICU: a French multicentric cohort

Adrien Mirouse¹, Romain Sonneville², Keyvan Razazi³, Sybille Merceron⁴, Laurent Argaud⁵, Naike Bige⁶, Stanislas Faguer⁷, Pierre Perez⁸, Guillaume Geri⁹, Claude Guérin⁵, Anne-Sophie Moreau¹⁰, Laurent Papazian¹¹, René Robert¹², François Barbier¹³, Frédérique Ganster¹⁴, Julien Mayaux¹⁵, Elie Azoulay¹, Emmanuel Canet¹⁶

¹Hôpital Saint-Louis, Paris, France; ²Hôpital Bichat, Paris, France; ³Hôpital Henri Mondor, Créteil, France; ⁴Hôpital André Mignot, Le Chesnay, France; ⁵CHU de Lyon, Lyon, France; ⁶Hôpital Saint-Antoine, Paris, France; ⁷CHU de Toulouse, Toulouse, France; ⁸CHU de Nancy, Nancy, France; ⁹Hôpital Cochin, Paris, France; ¹⁰CHU de Lille, Lille, France; ¹¹CHU de Marseille, Marseille, France; ¹²CHU de Poitiers, Poitiers, France; ¹³CHU de Orléans, Orléans, France; ¹⁴CH de Mulhouse, Mulhouse, France; ¹⁵Hôpital Pitié-Salpêtrière, Paris, France; ¹⁶CHU de Nantes, Nantes, Frances

Correspondence: Adrien Mirouse (adrien.mirouse@aphp.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-34

Rationale: Varicella zoster virus (VZV) is the second most common cause of infectious encephalitis in France. The purpose of this study was to describe the clinical features and prognosis of VZV encephalitis in adult patients admitted to the intensive care unit (ICU).

Patients and methods: Retrospective multicenter cohort study including adult patients admitted to 18 French ICUs between 01/01/1999 and 01/09/2017 with a confirmed diagnosis of VZV encephalitis, defined by a positive VZV PCR in the cerebrospinal fluid (CSF). Neurological outcome was evaluated at one year with the modified Rankin Scale (mRS). Data are presented as median [interquartile range] or numbers (percentage).

Results: Fifty-five patients (29 men, 53%) with an age of 54 years [36–66] at diagnosis were included. Forty-three patients (78%) had an underlying cause of immunosuppression. Patients were admitted to the ICU 1 day [0–3] after the onset of the first neurological symptoms. The main reasons for ICU admission were confusion (n = 21, 38%), coma (n = 20, 36%) and status epilepticus (n = 10, 18%). At ICU admission, the Glasgow Coma Scale (GCS) score was 12 [7–14], and 17 (31%) patients had a CGS ≤ 8. The SOFA score at day 1 was 6 [4–9]. Twenty-five (55%) patients had fever and 37 (67%) patients had a skin rash. CSF analysis revealed lymphocytic pleocytosis (68 cells/mm³ [13–129], lymphocytes 70% [27–95]), mildly elevated CSF protein levels (1.37 g/L [0.77–3.67]) and CSF glucose level of 3.8 mmol/L [2.6–5.0]. A cerebral computed tomography was performed in 33 (60%) patients, normal in 23 (70%) cases, and a brain magnetic resonance imaging was performed in 30 (55%) patients, of whom 21 (70%) had pathologic findings. Patients were treated intravenously with acyclovir 15 mg/kg/day [15–15] during 14 days [9–20]. During ICU stay, 46 (84%) patients required invasive mechanical ventilation, 22 (40%) received vasopressors, and 28 (51%) required renal replacement therapy. The ICU length of stay was 15 days [6–38] and 41 (75%) patients were discharged alive from the ICU. One year after ICU admission, 31 (63%) patients were alive, of whom 20 (41%) had favorable neurological outcome (mRS 0–2), and 18 (37%) were deceased (lost to follow-up, n = 6, 11%).

Conclusion: Severe VZV encephalitis in adult patients requiring ICU admission occurs mainly in patients with a history of immune deficiency and is associated with a severe prognosis. At one year, 41% of the patients had a favorable neurological outcome, 22% had moderate to severe disability, and 37% were dead.

Compliance with ethics regulations: Yes.

CO-35 Comparative analysis of host responses to septic and non-septic shock

Fabrice Uhel¹, Brendon Scicluna¹, Lonneke van Vught¹, Olaf Cremer², Marc Bonten³, Marcus Schultz⁴, Tom van der Poll¹

¹Center for Experimental and Molecular Medicine, Amsterdam University Medical Centers, location Academic Medical Center, University of Amsterdam, Amsterdam, Netherlands; ²Department of Intensive Care Medicine, University Medical Center Utrecht, Utrecht, Netherlands; ³Department of Medical Microbiology, University Medical Center Utrecht, Utrecht, Netherlands; ⁴Department of Intensive Care Medicine, and Laboratory of Experimental Intensive Care and Anesthesiology (L·E·I·C·A), Amsterdam University Medical Centers, location Academic Medical Center, University of Amsterdam, Amsterdam, Netherlands

Correspondence: Fabrice Uhel (fabrice.uhel@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-35

Rationale: Shock is the clinical expression of a circulatory failure that results in inadequate cellular oxygen utilization. Whereas the host response to septic shock has been extensively described, knowledge of the pathogenesis of non-septic shocks remains limited. We aimed to characterize the systemic host response in shock related to non-septic conditions (NSsh) as compared with septic shock (Ssh).

Patients and methods: We performed a prospective study in two intensive care units (ICUs) in patients admitted for Ssh (n = 931) or NSsh (n = 1338). Immune responses were determined upon ICU admission by measuring 17 plasma biomarkers reflecting host response pathways implicated in the pathogenesis of critical illness (in 573 Ssh and 287 NSsh patients), and by applying genome-wide blood mRNA expression profiling (in 267 Ssh and 136 NSsh patients).

Results: Compared with NSsh, patients with Ssh had more chronic comorbidities, greater disease severity (APACHE IV score 89 [IQR 70–112] vs. 71 [IQR 56–98], p < 0.001) and worse outcomes resulting in higher mortality rates up to one year after ICU admission (53.8% vs. 38.8%, p < 0.001). Plasma biomarker analysis revealed severely disturbed host responses in both Ssh and NSsh patients. However, Ssh patients displayed more prominent inflammatory responses, endothelial cell activation, loss of vascular integrity and a more pro-coagulant state relative to NSsh patients. Blood leukocyte genomic responses were more than 80% common between Ssh and NSsh patients relative to health (Fig. 1A), comprising overexpression of innate pro- and anti-inflammatory pathways, and underexpression of lymphocyte and antigen-presentation gene sets. Direct comparison of Ssh to NSsh patients matched for severity (Fig. 1B) showed overexpression of genes involved in mitochondrial dysfunction and specific metabolic pathways, and underexpression of lymphocyte, NF-κB and cytokine pathways.

Conclusion: Patients with Ssh and NSsh present with largely similar host response aberrations at ICU admission; however, patients with septic shock show more dysregulated inflammatory and vascular host responses, as well as specific leukocyte transcriptome alterations consistent with greater metabolic reprogramming and more severe immune suppression.

Compliance with ethics regulations: Yes.

Fig. 1

Leukocyte genomic responses in patients admitted to the intensive care unit with a septic or non-septic shock matched for severity upon admission

CO-36 HepatoDengue: Severity and outcome of acute hepatitis in Dengue Fever in New Caledonia

Mathieu Série¹, Darless Clausse¹, Marine Noel¹, Paul Abraham¹, Carole Forfait², Sophie Gaimard¹, Delphine Betton¹

¹CHT, Nouméa, France; ²DASS, Nouméa, France

Correspondence: Mathieu Série (mathieuserie@aol.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-36

Rationale: Dengue is a self-limited systemic viral infection transmitted by Aedes with one of four dengue viruses (DENV), and stands for the most common arbovirose worlwide. In 2009, the World Health Organization introduced a revised classification scheme consisting of the three following categories: dengue without warning signs, dengue with warning signs, and severe dengue, the last representing 1% of clinical cases requiring management in the ICU. In New Caledonia, hepatic injuries, seems to become a major cause of case severity during dengue fever epidemy. The aim of this study was to describe evolution and characteristics of hepatic injuries in patients with severe dengue fever admitted in New Caledonia territory-wide healthcare system.

Patients and methods: After approval by local ethic committee, all patients admitted in New Caledonia health care system with Dengue diagnosis during 2009 and 2017 epidemics were retrospectively included. Severity (WHO definition), hepatic failure and mortality were compared. Univariate and multivariate analysis between patients with and without severe hepatitis disease were also performed for patients from 2017 epidemy.

Results: Respectively 94 and 385 hospitalized patients in New Caledonia from 2009 and 2017 Dengue fever epidemics were included. A significant difference was found between 2009 and 2017 epidemics in terms of severity rates (respectively 15% vs. 34%, p = 0.006), acute hepatitis (AST or ALT ≥ 1000 units/L) (3.1% vs. 13.7%, p = 0.008) but not hepatic dysfunction (PTT < 50%) (Fig. 1). Mortality rates were not different between the two epidemics (3.1% in 2009, 3.3% in 2017), however hepatic dysfunction was associated with mortality. Indeed, in multivariate analysis, patients with acute severe hepatitis were younger (OR 4.74 [1.64–17.70], p = 0.003 for patients between 20 and 30 years old), had more sepsis (OR 3.2, p = 0.01). DENV Serotype 1 and 3 were more frequently associated with hepatitis (OR 6.15 and 10.39, p < 0.05). Absence of acetaminophen intake was also associated with acute severe hepatitis (OR 2.64 [1.21–5.77], p = 0.01).

Conclusion: Dengue fever severity seems to be increasing in New Caledonia within last epidemics with a bigger proportion of Hepatitis presentation. Dengue fever patients with acute hepatitis require close monitoring considering their higher mortality, and possibly bigger clinical burden. Acetaminophen in our study appears not to be responsible of this severe acute hepatitis.

Compliance with ethics regulations: Yes.

Fig. 1

Hepatic injuries for 2009 and 2017 Dengue epidemics

CO-37 Long-term prognosis after ICU discharge in kidney transplant recipients

Romain Arrestier¹, Nathanael Lapidus², Xavier Monnet³, Muriel Fartoukh⁴, Armand Mekontso Dessap⁵, Philippe Grimbert⁶, Antoine Durrbach⁷, Eric Rondeau⁸, Matthieu Jamme⁹

¹Medical Intensive Care Unit, Henri Mondor University Hospital, Assistance Publique-Hôpitaux de Paris, Créteil, France; ²Sorbonne University, INSERM, Institut Pierre Louis d’Epidémiologie et de Santé Publique IPLESP, Public Health Department, Saint-Antoine Hospital, Assistance Publique-Hôpitaux de Paris, Paris, France; ³Service de réanimation médicale, Hôpital de Bicêtre. Université Paris-Sud, Faculté de médecine Paris-Sud, Inserm UMR S 999, Le Kremlin-Bicêtre, France; ⁴AP-HP, Hôpital Tenon, Service de Réanimation médico-chirurgicale Faculté de médecine Sorbonne Université Groupe de Recherche Clinique CARMAS, Collégium Gallilée, UPEC, Paris, France; ⁵Service de Réanimation Médicale, Hôpitaux Universitaires Henri Mondor-Albert Chenevier, Assistance Publique-Hôpitaux de Paris. Groupe de Recherche Clinique CARMAS, Université Paris-Est Créteil, IMRB, Créteil, France; ⁶AP-HP, Nephrology and Renal Transplantation Department, Institut Francilien de Recherche en Néphrologie et Transplantation (IFRNT), Groupe Hospitalier Henri-Mondor/Albert-Chenevier Université Paris-Est-Créteil, DHU, VIC, IMRB, Equipe 21, INSERM U 955, Créteil, France; ⁷Nephrology Service, Centre Hospitalier de Bicetre UMRS1197, INSERM, Villejuif University Paris Sud, Orsay, Le Kremlin Bicêtre, France; ⁸Kidney transplant department, Tenon Hospital, Assistance Publique-Hôpitaux de Paris, Paris, France; ⁹Service de Réanimation Médico-Chirurgicale, CHI Poissy/Saint-Germain-en-Laye, Poissy, France

Correspondence: Romain Arrestier (romain.arrestier@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-37

Rationale: After kidney transplantation, 10% of kidney transplant recipients experience a hospitalization in intensive care unit (ICU). However, despite this high prevalence, knowledge about mortality and graft loss for patients discharged alive from ICU is scarce. The objective of our study is to determine incidence and risk factors of long term death and graft loss for these patients.

Patients and methods: We conducted a multicenter retrospective study in adult kidney transplant recipients discharged alive from ICU between 2013 and 2018. Data were collected from APHP shared database. Risk factors associated with composite criterion «death and/or graft-loss» were identified with a multivariate analysis by logistic regression, and each component of the composite criterion was studied in a secondary analysis.

Results: We identified 118 patients. Median age of patients was 62 years old [50–65] and median follow-up after ICU discharge was 9.6 months [4–21.2]. Nineteen (18.1%) patients died or lost their graft with a median time of 4.5 months [2–8] after ICU discharge. Acute rejection before ICU (OR = 4.3 [1.3; 15.4], p = 0.02) and need for dialysis in ICU (OR = 4.9 [1.6; 16.8], p = 0.01) were associated with the composite criterion. Subgroup analysis found that hospitalization in ICU for acute kidney injury was associated with graft loss (p = 0.038).

Conclusion: 18.1% of kidney transplant recipients experience graft loss or death in the first year after ICU discharge. History of acute rejection before ICU and the need of dialysis during hospitalization are associated with a poor outcome.

Compliance with ethics regulations: Yes.

CO-38 Outcome of septic patients with augmented renal clearance after community-acquired infection

Romain Courcelle, Diego Castanares, Pierre-François Laterre, Philippe Hantson

Departement of intensive care unit, Cliniques Saint-Luc, Université Catholique de Louvain, Brussels, Belgium

Correspondence: Romain Courcelle (romcourcelle@hotmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-38

Rationale: The true incidence of augmented renal clearance (ARC) has not been well established yet in septic patients with community-acquired infection. In this study, we specifically investigated the effects of ARC on clinical outcome and relapse of infection.

Patients and methods: This single-center, retrospective, observational, cohort study included critically ill patients aged more than 18 years-old, hospitalized for a sepsis or a septic shock secondary to a community bacterial infection. Kidney function was assessed by the 24-h measured creatinine clearance (Ccr). ARC patient was defined as Ccr > 130 ml/min/1.73 m². Primary endpoint was mortality at 28 days and at one year. Secondary endpoints were the incidence of ARC, recurrence of infection during hospitalization, the length of intensive care unit (ICU) and hospital stay.

Results: From 2015 to 2017, 128 patients were enrolled. Sources of infection were abdomen and lung in respectively 28% and 21% of the patients. Septic shock was present in 75% of the patients and 46% were ventilated. ARC was present in 23% of the patients. ARC patients were younger (p < 0.001), had less of chronic kidney failure (p = 0.001), less cardiovascular risk factors, less chronic cardiac failure (p = 0.023) and had a lower APACHE II score (p = 0.003). The SOFA score tended to be lower in ARC group (p = 0.16). Ccr of ARC patients at day 3 was significantly higher (141 vs. 42 ml/min/1.73 m²; p < 0.001). Renal replacement therapy was required in 23% of no-ARC patients vs. 4% of ARC patients (p = 0.014). In Kaplan–Meier analysis, mortality was significantly lower in ARC group at day 28 (10% vs. 38%) and at one year (28% vs. 51%). In multivariate logistic regression, ARC was found to be protective (adjusted OR: 0.1; 95% IC [0.02–0.51]; p = 0.006). SOFA score, cirrhosis and mechanical ventilation were recognized to be associated with mortality. In multivariate Cox regression analysis, ARC was associated with lower mortality (adjusted HR: 0.36; 95% IC [0.16–0.84]; p = 0.018). SOFA score, cirrhosis, mechanical ventilation and renal replacement therapy were found to be risk factors for mortality. There was no difference of the recurrence of infection between both groups (p = 0.89), nor in the length of ICU (p = 0.41) and hospital stay (p = 0.52).

Conclusion: In this limited series, we found a positive association between ARC and outcome in community-acquired infection.

Compliance with ethics regulations: Yes.

CO-39 Transition from Acute Kidney Injury to recovery: A cohort study

Moustafa Abdel-Nabey, Etienne Ghrenassia, Eric Mariotte, Sandrine Valade, Virginie Lemiale, Lara Zafrani, Elie Azoulay, Michaël Darmon

Hopital Saint-Louis, Paris, France

Correspondence: Moustafa Abdel-Nabey (moustafa.abdelnabey@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-39

Rationale: AKI is associated with short and long term mortality and morbidity. Although recovery has been demonstrated to be associated with outcome of critically ill patients, interpretation of available data is limited by time dependent nature of recovery and by competing risks. Our objective was to describe renal recovery, pattern of recovery according to ADQI definitions and risk factor of this later.

Patients and methods: Monocenter retrospective cohort study. Adult patients admitted in our ICU from July 2018 to December 2018 were included. AKI was defined according to KDIGO criteria and recovery according to ADQI definition. Incidence of recovery at each time point was depicted using competing risk survival analysis. Risk of transition between AKI and no-AKI was assessed by a semi-Markov model. Last, a trajectoire analysis was performed to depict most frequent recovery patterns. Results are reported as n (%) or median (IQR).

Results: 350 patients were included with a median age of 57 (45–72). Median SOFA score at admission was 4 [2–7]. At ICU admission, 85 patients (24.2%) had an AKI stage 1, 44 patients (12.6%) an AKI stage 2 and 37 patients (10.6%) an AKI stage 3. According to ADQI criteria, AKI was defined as rapidly reversed in 42 patients (25.3% of AKI patients), persistent AKI in 16 patients (9.7%) and as acute kidney disease (AKD) in 64 patients (38.6%), remaining patients couldn’t be classified (n = 44).

Risk of recovery was of 9% per day until day 7 then 15% per day (Fig. 1a). Fine and Gray model, taking into account death as competing risk, identified 3 risk factors negatively associated with renal recovery, namely SOFA score (sHR = 0.94 per point; 95% IC = [0.89–0.99]), preexisting hypertension (sHR = 0.56; 95% IC = [0.32–0.98]) and AKI severity (stage 3 vs. stage 1 sHR = 0.17; 95% IC = [0.06–0.56]). Risk of de novo AKI was maximal during the first 7 days and ranged from 25 to 36% per day.

Trajectoire model identified 3 clusters of patients (Fig. 1b), closely associated with patients’ outcome: a) low patients’ severity and no or mild AKI (n = 251; hospital mortality: 8%); b) moderate to severe AKI but little associated organ dysfunction (n = 70, hospital mortality: 12.9%); c) severe AKI and multiple organ failure (n = 29; hospital mortality: 89.7%).

Conclusion: This study, assessing AKI recovery patterns, is the first to our knowledge using ADQI definition. Despite the high rate of early recovery and of rapidly reversed AKI, up to 40% of AKI patients had not recovered at day 7 and could therefore be classified has having AKD.

Compliance with ethics regulations: Yes.

Fig. 1

Evolution of the risk of renal recovery and overall patient trajectory

CO-40 Attributable mortality and host response alterations associated with transient and persistent acute kidney injury in critically ill patients with sepsis

Fabrice Uhel¹, Hessel Peters-Sengers¹, Fahimeh Falah¹, Brendon Scicluna¹, Lonneke van Vught¹, Marc Bonten², Olaf Cremer ³, Marcus Schultz⁴, Tom van der Poll¹

¹Center for Experimental and Molecular Medicine, Amsterdam University Medical Centers, location Academic Medical Center, University of Amsterdam, Amsterdam, Netherlands; ²Department of Medical Microbiology, University Medical Center Utrecht, Utrecht, Netherlands; ³Department of Intensive Care Medicine, University Medical Center Utrecht, Utrecht, Netherlands; ⁴Department of Intensive Care Medicine, and Laboratory of Experimental Intensive Care and Anesthesiology (L·E·I·C·A), Amsterdam University Medical Centers, location Academic Medical Center, University of Amsterdam, Amsterdam, Netherlands

Correspondence: Fabrice Uhel (fabrice.uhel@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-40

Rationale: Sepsis is the most frequent cause of acute kidney injury (AKI). The “Acute Disease Quality Initiative Workgroup” recently proposed new definitions for AKI, classifying it as transient or persistent. We aimed to determine the incidence, attributable mortality and host response characteristics of transient and persistent AKI in patients with sepsis.

Patients and methods: We performed a prospective observational study comprising consecutive admissions for sepsis in 2 intensive care units (ICUs) in the Netherlands, stratified according to the presence and evolution of AKI. Attributable mortality fraction (excess risk for dying with persistent AKI relative to transient AKI) was determined using a logistic regression model adjusting for confounding variables. In a subset of 866 sepsis patients, 16 plasma biomarkers indicative of major pathways involved in sepsis pathogenesis were measured. In a second subset of 392 patients, whole-genome blood-leukocyte transcriptomes were analyzed.

Results: 1545 sepsis patients were included. AKI occurred in 37.7% (n = 577), of which 18.4% (n = 106) was transient and 81.6% (n = 471) persistent. Patients with persistent AKI had higher disease severity scores on admission than patients with transient AKI or without AKI and more frequently had severe (Injury of Failure) RIFLE AKI-stages on admission (n = 322, 68.4%) than transient AKI patients (n = 33, 31.1%, P < .001). Persistent AKI, but not transient AKI, was associated with increased mortality by day-30 (adjusted OR 2.42, 95% CI 1.28–4.58; P = .006) (Figure) and up to 1-year (adjusted OR 2.10, 95% CI 1.12–3.92; P = .020). The attributable mortality of persistent relative to transient AKI by day-30 was 14.0% (95% CI 3.7–24.2%). Persistent AKI was associated with enhanced and sustained inflammatory and procoagulant responses during the first 4 days, and a more severe loss of vascular integrity compared with transient AKI. Baseline blood gene expression showed minimal differences with respect to the presence or evolution of AKI.

Conclusion: Persistent AKI is associated with higher sepsis severity, sustained inflammatory and procoagulant responses, and loss of vascular integrity as compared with transient AKI, and independently contributes to sepsis mortality.

Compliance with ethics regulations: Yes.

Fig. 1

Kaplan-Meier 30-day survival plot of patients with sepsis stratified according to the evolution of acute kidney injury after admission to the intensive care unit

CO-41 In-hospital cardiac arrest in intensive care units: evolution of patient profile, management and outcomes from 1997 to 2015

Clotilde Bailleul¹, Etienne Puymirat¹, Philippe Aegerter², Bertrand Guidet¹, Alain Cariou¹, Nadia Aissaoui¹

¹APHP, Paris, France; ²APHP, Boulogne-Billancourt, France

Correspondence: Clotilde Bailleul (clotilde.bailleul@aphp.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-41

Rationale: To address the paucity of data on the epidemiology of patients admitted to intensive care units (ICUs) with in-hospital cardiac arrest (IHCA), we examined key features, mortality and trends in mortality in a large cohort of patients admitted in 33 French ICUs over the past 18 years.

Patients and methods: From 1997 to 2015 database of the Collège des Utilisateurs de Bases de données en Réanimation (CUB-Réa), we determined temporal trends in the characteristics of IHCA, patients’ outcomes and predictors of ICU mortality.

Results: Of the 376 325 ICU admissions, 33 126 (8.8%) were cardiac arrests and 15 324 were IHCA (4.1%). During the study period, the age of IHCA patients increased by 0.7 years (P = 0.04) and patients presented more comorbidities (chronic heart disease, chronic kidney disease and cancer). Patients were also more critically ill over the period as reflected by the increase of SAPS-II by 2.3% (P < 0.001). Paradoxically, in-hospital management became lighter through the time with reduced respiratory support (p < 0.001), renal support (p < 0.001) and use of vasoactive drugs (p < 0.001). Crude in-ICU mortality decreased from 78% to 62.5% over the past eighteen years (P < 0.001), Fig. 1. In multivariate analysis, a more recent time-period was an independent correlate of decreased mortality [OR 0.40, 95% CI 0.35–0.46]. Factors independently associated with increased ICU mortality were older age [ (age ≥ 75 years; OR 1.71, 95% CI 1.54–1.90], SAPS II [OR 1.03, 95% CI 1.02–1.03], and admission for intracranial hemorrhage [OR 2.95, 95% CI 2.22–3.92]. Interestingly, regarding the associated diagnosis with IHCA, myocardial infarction [OR: 0.62; 95% CI 0.56–0.69] and arrhythmia [OR 0.76, 95% CI 0.68–0.85] were associated with a reduced risk of ICU mortality.

Conclusion: Over the 1997–2015 period, IHCA patients became older and more critically ill while the ICU mortality rate decreased suggesting improved overall management. However, mortality rates remain high, warranting further research specifically dedicated to this population.

Compliance with ethics regulations: Yes.

Fig. 1

In-ICU mortality in IHCA patients according to time period

CO-42 Can the cahp score help to predict the mode of death post cardiac arrest?

Marine Paul¹, Sarah Benghanem², Stéphane Legriel¹, Florence Dumas³, Alain Cariou²

¹Intensive Care Unit, Centre Hospitalier de Versailles-Site André Mignot, Le Chesnay, France; ²Intensive Care Unit Cochin hospital, Paris, France; ³Emergency Department, Cochin hospital, Paris, France

Correspondence: Marine Paul (marine.1604@hotmail.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-42

Rationale: Mortality after cardiac arrest (CA) remains very high, chiefly caused by post-resuscitation syndrome that may lead to refractory shock, severe brain damage and brain death. Withdrawal of care due to comorbid status and recurrence of cardiac arrest also contribute to this poor outcome. The Cardiac Arrest Hospital Prognosis (CAHP) score demonstrated a high discrimination value to predict global outcome after CA. Here we assessed its effectiveness in predicting the mode of death.

Patients and methods: We performed a retrospective monocentric study aiming to describe the mode of death after CA according to the CAHP score. All consecutive patients admitted in a tertiary CA center between 2007 and 2017 with a sustained return of spontaneous circulation (ROSC) were studied. Causes of death were recurrent sudden CA, refractory hemodynamic shock, neurological withdrawal of care, brain death and comorbid withdrawal of care. Risk of ICU death was calculated using the validated CAHP score including age, setting, initial cardiac rhythm, no flow as the time from CA to cardio-pulmonary resuscitation (CPR) and low flow as the time from CPR to ROSC, arterial pH, and epinephrine dose. The distribution of death modalities was reported for each of the 3 pre-defined subgroups of CAHP score: low risk (< 150 points), medium risk (150 to 200 points), and high risk (> 200 points).

Results: Among 1445 post-CA patients, CAHP score was available for 1293 patients. Median age was 64 years, median no flow and low flow were 3 min (IQR, 0–7) and 17 min (IQR, 10–25), respectively. Patients had an initial shockable rhythm in 50%, received epinephrine during CPR in 67% and 71% of patients developed a post-resuscitation shock. The median CAHP score was 173 (IQR, 135–206). ICU mortality was 62% (799/1293) with the following distribution: neurological withdrawal of care: 45%, refractory hemodynamic shock: 31.3%, brain death: 14.3%, recurrent sudden CA: 5.6% and comorbid withdrawal of care: 3.8%. Causes of death (Figure) differed significantly according to the CAHP subgroups (p < 0.001). Proportion of refractory shock as the cause of death increased in parallel with severity, especially in the most severe subgroup (> 200 points) (p < 0.001).

Conclusion: According to the CAHP score, the proportion of post–cardiac arrest patients dying from a refractory hemodynamic shock was more frequent in the most severe patients (CAHP score > 200). The CAHP may help to identify specific subgroups of post-cardiac arrest patients when testing new treatments.

Compliance with ethics regulations: Yes.

Fig. 1

CAHP score and outcome with detailed mode of death

CO-43 Neuron-Specific Enolase as a Predictor of Death or Poor Neurological Outcome After Out-of-Hospital Cardiac Arrest in patient with nonshockable rhythm

Arnaud-Félix Miailhe¹, Alain Cariou², Gwenhael Colin³, Jean-Pierre Quenot⁴, Didier Thevenin⁵, Nicolas Pichon⁶, Sylvie Vimeux⁷, Thierry Boulain⁸, Jean-Pierre Frat⁹, Elisabeth Coupez¹⁰, Stéphane Legriel¹¹, Amélie le Gouge¹², Bruno Giraudeau¹², François Desroys du Roure¹³, Jean-Baptiste Lascarrou¹⁴

¹Medical Intensive Care Unit, University Hospital Center, Nantes, France; ²Medical Intensive Care Unit, Cochin University Hospital Center, Paris, France; ³Medical Intensive Care Unit, District Hospital Center, La Roche Sur Yon, France; ⁴Medical Intensive Care Unit, University Hospital, Dijon, France; ⁵Medical-Surgical Intensive Care Unit, General Hospital Center, Lens, France; ⁶University Hospital Center, Limoge, France; ⁷General Hospital Center, Montauban, France; ⁸Regional Hospital Center, Orleans, France; ⁹University Hospital Center, Poitiers, France; ¹⁰Medical Intensive Care Unit, University Hospital, Clermont Ferrand, France; ¹¹Medical Intensive Care Unit, General Hospital, Versailles, France; ¹²INSERM CIC1415, CHRU de Tours, Tours, France; ¹³Biological unit, District Hospital Center, La Roche Sur Yon, France; ¹⁴Medecine Intensive Reanimation, University Hospital Center, Nantes, France

Correspondence: Arnaud-Félix Miailhe (flexmiailhe@hotmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-42

Rationale: The prognostication of hypoxic-ischaemic brain injury after resuscitation from cardiac arrest is based on a multimodal approach including biomarkers. The most used biological marker is the neuron-specific enolase (NSE). Currently, no multicenter randomized blinded biomarker study is interested in NSE after out-of-hospital cardiac arrest (OHCA) in patient with a nonshockable rhythm due to any cause. Our goal is to evaluate NSE as a predictor of death and cerebral injuries in this population.

Patients and methods: All patients included in this study were part of the HYPERION trial. The HYPERION trial was an blinded-outcome-assessor, multicenter, randomized clinical trial conducted in France between January 2014 and January 2018 with the objective of the randomized multicenter HYPERION trial was to assess whether, compared to targeted normothermia (37 °C). The NSE analysis was performed in 12 intensive care units (ICUs). Serum blood sample were taken from the patients at 24, 48, 72 h after return on spontaneous circulation. All samples were pretreated at the different sites, aliquoted and frozen to − 80 °C before shipment to the biology laboratory of La Roche sur Yon, where the NSE analyses were performed using a COBAS e601 line with an Electro-Chemi-Luminescent—Immuno-Assay.

Results: NSE analyses were performed for a total of 106 patients and 237 analyzable blood samples. Median NSE values were 22.6 ng/ml versus 33.6 ng/ml, 18.1 ng/ml versus 76.8 ng/ml, and 9 ng/ml versus 80.5 ng/ml for good versus poor outcome at 24, 48, and 72 h, respectively (p < 0.04; p < 0.0029;p < 0.001). At 48 h and 72 h, NSE predicted neurological outcome with areas under the receiver-operating curve of 0.79 and 0.9, respectively. The cut-off values at 72 h after ROSC provided the best capacity to predict outcome in OHCA in a nonshockable rhythm due to any cause. Management at 33 °C or 37 °C did not significantly affect NSE levels (p = 0.58).

Conclusion: We confirm in a large randomized multicentre trial on OHCA in a nonshockable rhythm due to any cause, that NSE values at 72 h are good predictors of poor outcome: for example, a value of 34.9 ng/ml was associated with a poor outcome with a Specificity of 100%.

Compliance with ethics regulations: Yes.

Fig. 1

NSE values and NSE cutoff values at 24, 48, 72 h after ROSC for outcome prediction according to CPC at 6 months

CO-44 Long-term disabilities in awakened survivors of out-of-hospital cardiac arrest: the Hanox study

Charles-Edouard Luyt¹, Alain Cariou², Nicolas Deye³, Emmanuel Guérot⁴, Julien Mayaux¹, Romain Sonneville⁵, Pascale Pradat-Diehl¹, Anne Peskine¹

¹Pitié-Salpêtrière Hospital, Paris, France; ²Cochin Hospital, Paris, France; ³Lariboisière Hospital, Paris, France; ⁴Georges Pompidou European Hospital, Paris, France; ⁵Bichat Hospital, Paris, France

Correspondence: Charles-Edouard Luyt (charles-edouard.luyt@aphp.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-44

Rationale: Little is known about long-term disabilities of awakened survivors of out-of-hospital cardiac arrest (OHCA). We performed this multicenter prospective observational study to describe 18-month outcome of survivors of OHCA who awoke during the first 2 weeks after cardiac arrest (CA), and risk factors associated with poor outcome.

Patients and methods: All survivors of OHCA who awoke during the first 2 weeks after CA (Glasgow coma score ≥ 12) were enrolled in 6 intensive care units from Paris and followed during 18 months (with evaluations performed at 3, 6, 12 and 18 months) in 4 rehabilitation departments. Primary outcome measure was Glasgow outcome scale-extended (GOS-E) at 18 months. Secondary outcome measures included evaluation of neurological, behavioral, and cognitive disabilities, as well as health-related quality of life (compared to sex- and age-matched population), anxiety and depression. Moreover, we explored risk factors of poor outcome (GOS-E ≤ 6) at 18 months.

Results: One-hundred and thirty-nine patients were included, of whom 98 were evaluable for the primary outcome measure. At 18-months follow-up, 64 (65%) patients had full recovery or minor disabilities (GOS-E > 6), 18 (18%) had moderate disabilities but were autonomous for activities of daily living (GOS-E = 6), 12 (12%) had poor autonomy (GOS-E < 6 but > 1) and 4 were dead. Proportion of patients with GOS-E > 6 increased significantly over time (Fig. 1). At 18 months, no patients had major neurological disabilities, 20% had cognitive disabilities, 1/3 had anxiety symptoms, 1/4 had depression symptoms, and their quality of life was impaired as compared to sex- and age-matched population. Except risk factors associated traditionally with poor outcome in OHCA patients (no-flow, low-flow, SOFA score at admission and mechanical ventilation at day 3 and 7), no specific risk factors of being classified with a GOS-E ≤ 6 at 18 months were observed.

Conclusion: Among patients who awoke (GCS ≥ 12) after OHCA, 35% had moderate–to–heavy disabilities or were dead at 18 months. Health-related quality of life, cognitive and behavioral functions were also altered in some survivors. Risk factors associated with poor outcome (GOS-E ≤ 6) were no- and low-flow times, severity on admission and persistence of mechanical ventilation on day 3 and 7.

Compliance with ethics regulations: Yes.

Fig. 1

Outcomes of patients during follow-up. Proportion of patients in each Glasgow outcome scale extended (GOS-E) category at 3, 6, 12 and 18 months. Numbers of patients evaluated at each time point are given at the bottom

CO-45 Non-inferiority study in pediatric cardiac surgery compared a short 24-h antibiotic prophylaxis with 48-h protocol

Jeanne Bordet, Lucie Petitdemange

CHU Strasbourg, Strasbourg, France

Correspondence: Jeanne Bordet (jeanne.bordet.pro@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-45

Rationale: In surgery, prophylaxis antibiotic aims at preventing the occurrence of post-operative infections. For adults, it is currently recommended to only use prophylactic antibiotic therapy during the time of the intervention. But in pediatric cardiac surgery, there is no consensus around the optimal duration of use of antibiotic prophylaxis. The protocol was modified in 2018 in the ICU and its time reduced to 24 h. We aimed to determine whether 24 h of post-sternotomy antibiotic prophylaxis was not less effective than 48 h treatment to help prevent care-associated infections.

Patients and methods: After agreement of the ethics committee of our institution, we performed a retrospective non inferiority study, with an inferiority margin to 10%. The primary objective is to compare the incidence of care-related infections between a second-generation cephalosporin (C2G) antibiotic prophylaxis during 48 h and a 24-h protocols. The secondary objectives are to determine the infection’s incidence, to identify the risk factors for nosocomial infections and to compare the incidence of multidrug-resistant infections.

Results: Between January 2013 and July 2019, 402 children underwent cardiac surgeries with sternal opening. 299 received 48 h of C2G antibiotic prophylaxis and 103 received 24 h of C2G treatment. Five previously infected children have been excluded. Both groups were demographically and surgically similar. The median age was 7 months (range a few hours of life to 14.5 years old) and the median weight was 6.7 kg. In the intent-to-treat analysis, incidence of care-related infections is at 10.03% in the C2G-48 h group and 10.68% in the C2G-24 h group. A multivariate analysis shows that the shorter 24-h time antibiotic prophylaxis is not inferior regarding infection prevention compared to 48 h of antibiotic prophylaxis, p = 0.046. As in the per protocole analysis, the C2G-48 h group rate was 9.1% and 6.6% for the G2G-24 h group.

Conclusion: It demonstrates that shortening the antibiotic prophylaxis treatment time to 24 h does not affect or increase the rate of infections after a pediatric sternotomy surgery compared to 48-h protocole. Prophylaxis in pediatric cardiac surgery should be short-lived. A multicenter prospective study would allow a consensus and confirm this decision.

Compliance with ethics regulations: Yes.

Fig. 1

Non-inferiority testing, different in confidence interval of infections rate between intent-to-treat and Per-protocol analysis

CO-46 Management of pediatric acute liver failure with extra corporeal therapies

Megan Nallet-Amate¹, Benoit Colomb¹, Etienne Javouhey²

¹Neonatal and Pediatric Intensive Care Unit, Department of Pediatrics, Dijon, Burgundy University, Dijon, France; ²Pediatric Intensive Care Unit, Department of Pediatrics, Lyon-Bron, Claude Bernard University, Lyon, France

Correspondence: Megan Nallet-Amate (megan.amate@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-46

Rationale: Pediatric acute liver failure (ALF) is a life-threatening emergency leading to multiorgan failure and requiring management in pediatric intensive care unit (PICU). Stabilization of the child during the waiting time to spontaneous recovery or to liver transplantation (LT) is crucial to prevent evolution to organ failures and death. In this study, we describe our experience on the use of Renal replacement therapy (RRT) in ALF or acute-on-chronic liver failure (AoCLF).

Patients and methods: We performed a retrospective, monocenter, observational study in Lyon-Bron PICU. Patients aged from 0 to 18 years, admitted to PICU between January 2010 and January 2018, who presented liver failure (definitions of the PALF Study group) and were treated either with continuous hemodiafiltration combined or not with molecular adsorbents recirculating system (MARS), were included. We compared biological (bilirubin, ammonia, transaminase, creatinine, lactate, aromatic amino acids) and clinical (hepatic encephalopathy clinical and electrophysiological grades) parameters of liver dysfunction before and after replacement therapy (continuous veno venous hemo-dia-filtration (CVVHDF) and MARS.

Results: Thirty-nine children were included (medium age: 38 months): 14 (35.9%) underwent MARS treatment, 18 (46.1%) had an AoCLF. Medium PELOD score at admission was 23 [21.0–32.0]. ALF were widely due to congenital (23.1%) or hematologic diseases (20.5%). All biological parameters, except bilirubin, trend to an improvement after 96 h of CVVHDF. Twenty-one children (53.8%) presented clinical signs of encephalopathy. A large part of those children had a grade III encephalopathy (N = 13). We observed a reduction of EEG grade III and IV after 48 h of RRT. Children who received MARS treatment had an improvement of all biological parameters but without statistically significance. The Fischer index increased during MARS treatment from 1.13 to 1.35 (p = 0.09). Overall survival was 62.5% in our cohort, 31% of children survived without liver transplant. For children less than 12 months, we observed a mortality rate of 56%. No side effect of MARS treatment was observed.

Conclusion: The results of this observational study suggest that CVVHDF could be a useful treatment to improve liver function (or to stabilize the patient), and that MARS treatment might be an additional safe support. Inscription on liver transplant list should not be delayed, regardless of the initiated technic, according to the literature.

Compliance with ethics regulations: Yes.

CO-47 An automated modified Wood’s clinical asthma score to strengthen respiratory severity assessment in big data

Sally Al Omar^1,2, Alex Lepage-Farrell^1,2, Atsushi Kawaguchi^1,2,3, Gabriel Masson^1,4, Philippe Jouvet^1,2, Guillaume Emeriaud^1,2

¹Pediatric Intensive Care Unit, Sainte-Justine Hospital, Montreal, Canada; ²Department of Pediatrics, Université de Montréal, Montreal, Canada; ³Department of Pediatrics, University of Ottawa, Ottawa, Canada; ⁴Department of Anesthesiology and Intensive Care Medicine, University Hospital of Lille, Lille, France

Correspondence: Sally Al Omar (sally.alomar@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-47

Rationale: The use of “big data” is getting increasingly popular in the medical field, especially in intensive care where large amounts of data are continuously generated. However, big data can be misleading when essential clinical data are missing. The adequate adjustment for potential confounding factors (e.g., severity of respiratory distress) should be the key procedure in the big data analyses; however, it is challenging to capture the clinical severity within large electronic databases. Bronchiolitis is one main reason for admission to pediatric intensive care unit (PICU). The modified Wood’s clinical asthma score (mWCAS) is widely used to assess the severity of bronchiolitis. The objective of the study is to build an automated mWCAS (A-mWCAS) to continuously assess the severity of respiratory distress in critically ill children.

Patients and methods: This retrospective study included all infants < 2 years old with a clinical diagnosis of bronchiolitis, ventilated with non-invasive neurally adjusted ventilatory assist, in a Canadian PICU, between October 2016 and June 2018. We developed an algorithm, using Python 3.7, which was directly connected to the electronic medical record. The components of the score were collected using structured query language (SQL) queries and processed to derive the A-mWCAS. For validation, the A-mWCAS score was compared to the mWCAS manually computed by a clinical expert (M-mWCAS).

Results: Sixty-four infants were included in the study, for which 256 of A-mWCAS and M-mWCAS were generated respectively. The Cohen’s Kappa coefficient was applied to estimate the agreement between the two scores which was 0.71 (95% confidence Interval) (Table 1) which corresponds to 78.5% of complete agreement. 17.5% of the A-mWCAS scores were within ± 0.5 of the M-mWCAS. The Kappa coefficient for the each score component were: 0.91 for the oxygen saturation, 0.79 for the expiratory wheezing, 0.91 for the inspiratory breath sounds, 0.89 for the use of accessories muscles and 0.51 for the mental status, respectively.

Discussion: The largest discrepancy was observed in the mental status, which clinical evaluation is relatively subjective and varies among care team members (doctor, nurse, respiratory therapist…). The automated score likely decreases this variability by consistently using the same source (respiratory therapist), but its validity should be confirmed in a prospective study.

Conclusion: The A-mWCAS provides a valid estimation of the mWCAS that is fast and robust. After external prospective validation, it may help to add some clinical sense within large electronic databases, with improved assessment of the respiratory distress.

Compliance with ethics regulations: Yes.

Fig. 1

The agreement matrix showing the agreement between the 256 pairs of A-mWCAS and M-mWCAS

CO-48 The experience of parents of critically ill children during their stay in a French paediatric intensive care unit

Clémentine Fort, Etienne Javouhey, Fabienne Bordet

Hôpital Femme Mère Enfant, Lyon, France

Correspondence: Clémentine Fort (clementine.fort@chu-lyon.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-48

Rationale: In paediatric intensive care units (PICU), survival rates have dramatically improved. This has been accompanied by increased morbidity, including psychological morbidity. These new impairments, that can affect the survivors and their families have been conceptualized under the frame of post-intensive care syndrome (PICS) and PICS-family. The aim of this study was to explore the experience of critically ill children parent’s during the stay in PICU, and its impact on the family.

Patients and methods: We planned a prospective, single centre study for 3 months. We collected qualitative written data from parents whose child had been admitted to the PICU for the first time, for at least two nights.

Results: Fifty-seven questionnaires were analysed from thirty-seven admissions. PICU admissions were mostly unplanned. Among parents 40% experienced very painful memories during admission and 61% have feared for their child’s life. During the stay, noise has bothered 30% of parents, and many have described difficulties to rest at night. 81% had the sensation that their child was suffering, mostly from pain, tiredness, anxiety or fear. During PICU stay, 75% of parents had to stop working, and siblings schooling was impacted in 25% of cases, 77% of parents considered themselves to be useful for their child and 81% have participated to nursing care. More than 70% were satisfied about information given and communication, 77% appreciated empathy and support from care givers. Parents received support from family, friends, and also from other parents of hospitalized children. Parents expressed relief (46%) and serenity (38%) to leave PICU, 58% of them were in demand to meet PICU staff again after discharge.

Conclusion: PICU parent’s experience is tough, and the impact on family is clear. These are known risks factors for PICS. On a very positive note, parents seemed to be satisfied by family-centred care, and were able to preserve their parental role. However, there is still room for improvement of practices.

Compliance with ethics regulations: Yes.

CO-49 Study of gastrointestinal dysfunction in ventilated adults with shock

Paul Decamps¹, Amélie le Gouge², Julie Boisramé-Helms³, Laurent Brisard⁴, Jean-Baptiste Lascarrou⁵, Nadia Anguel⁶, Laurent Argaud⁷, Karim Asehnoune⁸, Pierre Asfar⁹, Anne Bretagnol¹⁰, Hoang-nam Bui¹¹, Emmanuel Canet¹², Daniel da Silva¹³, Michaël Darmon¹⁴, Vincent Das¹⁵, Jérôme Devaquet¹⁶, Michel Djibre¹⁷, Maïté Garrouste¹⁸, Stéphane Gaudry¹⁹, Olivier Gontier²⁰, Claude Guerin²¹, Bertrand Guidet²², Christophe Guitton²³, Jean-Etienne Herbrecht²⁴, Jean-Claude Lacherade²⁵, Philippe Letocart²⁶, Frédéric Martino²⁷, Emmanuelle Mercier²⁸, Jean-Paul Mira²⁹, Saad Nseir³⁰, Gaël Piton³¹, Jean-Pierre Quenot³², Jack Richecoeur³³, Jean-Philippe Rigaud³⁴, René Robert³⁵, Carole Schwebel³⁶, Michel Sirodot³⁷, Didier Thevenin³⁸, Jean Reignier¹

¹Service de Médecine Intensive-Réanimation, CHU de Nantes, Nantes, France; ²CIC INSERM 1415, Hôpital Bretonneau, Tours, France; ³Medical Intensive Care Unit, Nouvel Hôpital Civil, Hôpitaux Universitaires de Strasbourg, Strasbourg, France; ⁴CHU de Nantes, Hôpital Laennec, Département d’Anesthésie et Réanimation, Nantes, France; ⁵Médecine Intensive Réanimation, Chu De Nantes, Nantes, France; ⁶Medical Intensive Care Unit, CHU de Bicêtre, Assistance Publique-Hôpitaux de Paris (AP-HP), Le Kremlin-Bicêtre, Paris, France; ⁷Medical Intensive Care Unit, Hospices Civils de Lyon, Hôpital Edouard Herriot, Lyon, France; ⁸Surgical Intensive Care Unit, Hotel Dieu, CHU de Nantes, Nantes, France; ⁹Medical Intensive Care and Hyperbaric Oxygen Therapy Unit, Centre Hospitalier Universitaire Angers, Angers, France; ¹⁰Medical Intensive Care Unit, CHR Orléans, Orléans, France; ¹¹Medical Intensive Care Unit, Hôpital Pellegrin, CHU Bordeaux, Bordeaux, France; ¹²Médecine Intensive Réanimation, CHU de Nantes, Nantes, France; ¹³Medical-Surgical Intensive Care Unit, Centre Hospitalier de Saint-Denis, Saint-Denis, France; ¹⁴Medical-Surgical Intensive Care Unit, University Hospital, Saint Etienne, Saint Etienne, France; ¹⁵Medical-Surgical Intensive Care Unit, Centre Hospitalier Intercommunal André Grégoire, Montreuil, France; ¹⁶Medical-Surgical Intensive Care Unit, Hôpital Foch, Suresnes, Suresnes, France; ¹⁷Medical-Surgical Intensive Care Unit, Tenon University Hospital, Assistance Publique-Hôpitaux de Paris (AP-HP), Paris, France; ¹⁸Medical-Surgical Unit, Hôpital Saint-Joseph, Paris, France; ¹⁹Medical-Surgical Intensive Care Unit, Hôpital Louis Mourier, Assistance Publique-Hôpitaux de Paris (AP-HP), Colombes, France; ²⁰Medical-Surgical Intensive Care Unit, Hôpital de Chartres, Chartres, France; ²¹Medical Intensive Care Unit, Groupement Hospitalier Nord, Hospices Civils de Lyon, Lyon, France; ²²Medical Intensive Care Unit, Hôpital Saint-Antoine, Assistance Publique-Hôpitaux de Paris (AP-HP), Paris, France^{; 23}Medical-Surgical Intensive Care Unit, Hôpital du Mans, Le Mans, France^{; 24}Medical Intensive Care Unit, Hôpital de Hautepierre, Hôpitaux Universitaires de Strasbourg, France^{; 25}Médecine Intensive Réanimation, Centre Hospitalier Départemental de la Vendée, La Roche sur Yon, France^{; 26}Medical-Surgical Intensive Care Unit, Hôpital Jacques Puel, Rodez, France; ²⁷Medical-Surgical Intensive Care Unit, CHU de Pointe-à-Pitre, Pointe-à-Pitre, France; ²⁸Médecine Intensive Réanimation, Hôpital Bretonneau, CHU Tours, Tours, France^{; 29}Intensive Care Unit, Cochin University Hospital, Assistance Publique-Hôpitaux de Paris (AP-HP), Paris, France^{; 30}Medical Intensive Care Unit, CHU Lille, Lille, France; ³¹Intensive Care Unit, CHRU Besançon, Besançon, France; ³²Medical-Surgical Intensive Care Unit, François Mitterrand University Hospital, Dijon, France^{; 33}Medical-Surgical Intensive Care Unit, Hôpital de Beauvais, Beauvais, France^{; 34}Medical-Surgical Intensive Care Unit, Hôpital de Dieppe, Dieppe, France^{; 35}Medical Intensive Care Unit, CHU Poitiers, Poitiers, France; ³⁶Medical Intensive Care Unit, CHU Albert Michallon Grenoble, Grenoble, France; ³⁷Medical-Surgical Intensive Care Unit, Centre Hospitalier Annecy-Genevois, Metz-Tessy, France^{; 38}Medical-Surgical Intensive Care Unit, Centre Hospitalier Docteur Schaffner, Lens, France

Correspondence: Paul Decamps (paul.decamps@hotmail.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-49

Rationale: The Gut has been suspected to be involved in multiple organs dysfunction syndrome (MODS) in the Intensive Care Unit (ICU). Studies suggested a link between gastrointestinal dysfunction (GID) and outcomes. But these studies included very few patients and most of them were retrospective.

Patients and methods: This study is a secondary analysis of data from a previous study that included patients from 44 French ICUs. GID is defined as the association of vomiting and constipation or diarrhea during the first week after ICU admission. Patients included were treated with vasopressors and mechanical ventilation. The first goal was to determine if GID is a risk factor of 28-day mortality in this population. Secondary goals were to assess the impact of GID on nosocomial infections.

Results: Among 2410 included patients, 238 (9.9%) had GID. By day-28, 76 (32%) of the 238 patients with GID and 780 (36%) of the 2172 patients without GID had died (Odds Ratio 0.84 [0.96–1.01]; p = 0.25). Multivariable regression model did not show any association between gastrointestinal dysfunction and increased risk of 28-day mortality in patients (Odds Ratio 0.76 [0.56–1.04], p = 0.083). Gastrointestinal dysfunction was strongly associated with other secondary outcomes (Table 1). Patients with GID had longer ventilation duration, ICU length of stay and hospital length of stay. They also had more nosocomial infections, in particularly ventilator-associated pneumonia. This association still existed in a multivariable regression model for prediction of nosocomial infection including the same variables than the previous model (Odds Ratio 1.68 [1.19–2.37], p = 0.0028). No association with day-90 mortality was observed.

Conclusion: Gastrointestinal dysfunction was not a risk factor of day-28 mortality but was associated with an increased risk of nosocomial infection and an increased length of stay. This study is observational and no causality link can be done. However, our data suggest further studies on strategies aimed to limit GID.

Compliance with ethics regulations: Yes.

Table 1 Outcomes

CO-50 Incidence and risk factors of upper digestive tract ischemia after Out-of-Hospital Cardiac Arrest, the ENTRACT study

David Grimaldi¹, Stéphane Legriel², Nicolas Pichon³, Philippe Colardelle², Sarah Leblanc⁴, Florence Canoui-Poitrine⁵, Omar Ben Hadj Salem⁶, Gregoire Muller⁷, Nicolas de Prost⁸, Sofia Herrmann⁹, Sophie Marque¹⁰, Aurore Baron¹⁰, Bertrand Sauneuf¹¹, Jonathan Messika¹², Jacques Creteur¹³, Jean pierre Bedos¹⁴, Fabio Taccone¹³, Emmanuelle Boutin⁵, Alain Cariou⁶

¹USI Hôpital Erasme, Bruxelles, Belgium; ²CH Versailles, Le Chesnay, France; ³CHU Dupuytren, Limoges, France; ⁴Gastro-enterologie CHU Cochin, Paris, France; ⁵Biostatistique CHU henri Mondor, Creteil, France; ⁶CHU Cochin, Paris, France; ⁷MIR CH Orléans, Orléans, France; ⁸CHU Henri Mondor, Creteil, France; ⁹Gastroentérologie CH Orléans, Orléans, France; ¹⁰Réanimation CH Sud Francilien, Corbeil, France; ¹¹Réanimation CH Public du Cotentin, Cherbourg, France; ¹²Réanimation CHU Louis Mourier, Colombes, France; ¹³USI Hôpital Erasme, ULB, Bruxelles, Belgium; ¹⁴Réanimation CH Versailles, Le Chesnay, France

Correspondence: David Grimaldi (david.grimaldi@erasme.ulb.ac.be)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-50

Rationale: Post-cardiac arrest (CA) ischemia/reperfusion induces organ injury. Gastro-intestinal ischemia can lead to endotoxin translocation and worsening of organs failure. The ischemic injury of the gastro-intestinal (GI) tract after CA is suggested by digestive biomarkers and by frequent endotoxemia. However its incidence and its risk factors after out-of-hospital CA are unknown. The main objective was to measure the incidence of upper digestive tract ischemia after OHCA. The secondary objectives were to determine its severity and its risk factors.

Patients and methods: The ENTRACT study was a prospective, interventional, non-controlled, multicentric study. Included patients underwent a gastroscopy 2 to 4 days after OHCA if still intubated. Digestive ischemia was determined by the gastroenterologist. Severe lesion were defined as ulceration or necrosis, mild as mucosal oedema or erythema. We compared the patients’ caracteristics available before the gastroscopy according to the presence or not of digestive ischemia. Multivariable regression analysis were performed to identify variables associated with digestive ischemia.

Results: 221 patients were included in 9 ICUs, 214 were suitable for complete analysis. Mean age was 62, 74% were male. CA was witnessed in 90%, initial rhythm was VT/VF in 53%, median no-flow was 5 (0–9) min and low flow 21 (15–30) min. 121 (57% 95% IC 50–63%) patients had at least one ischemic lesion of upper digestivetract including 103 (85%) that didn’t present any digestive symptoms before endoscopy. Fundus was the most frequent localization of ischemia followed by antrum, duodenum and esophagus. Ischemic lesions were severe in 55/121 (45%) patients and mild in the others. Patients with digestive ischemia compared to patients without had similar demographic characteristics but were less frequently treated with PPI before CA (16% vs 31%), frequence of NSAID use (including aspirin) and H. pylori positive serology were similar in both groups. CA characteristics were similar except that patients with digestive ischemia received more epinephrine (2 (1–4) vs 1 (0–3) mg) despite similar low-flow. Multivariate regression identified peripheral artery disease (OR 0.3 (0.09–0.94)), previous PPI (OR 0.49 (0.23–1.03)) and hypochloremia (OR 0.9 (0.84–0.97)) as protective factors whereas hematocrit (OR 1.04 (1.00–1.09)) and epinephrine dose during CPR (OR 1.17 (1.03–1.32)) were associated with upper digestive ischemia.

Conclusion: In this multicentric study, upper digestive tract ischemia was frequent after CA (57% of the patients) and was associated with epinephrine dose during CPR. Digestive ischemia involved mainly the stomach, necrosis was rare. Main limitation is that patients extubated or dead before day-2 were not included.

Compliance with ethics regulations: Yes.

CO-51 Severe acute cholangitis in French ICUs: a retrospective multicenter study. ICUte Cholangitis group

Jean-Rémi Lavillegrand¹, Emmanuelle Mercier-des-Rochettes¹, Claire Pichereau², Gaël Piton³, Baron Elodie⁴, Frédéric Pene⁴, Idriss Razach⁵, Merad Monchi⁶, Damien Contou⁷, Gabriel Preda⁸, Chloe Molliere⁹, Arnaud Galbois⁹, Arnaud-felix Miailhe¹⁰, Jean Regnier¹⁰, Raphael Favory¹¹, Guillaume Dumas¹², Hafid Ait-Oufella¹

¹Réanimation, Hopital Saint-Antoine, Paris, France; ²Réanimation, CH de Poissy, Poissy, France; ³Réanimation, CHU Besançon, Besançon, France; ⁴Réanimation, Hopital Cochin, Paris, France; ⁵Réanimation, CH de fontainebleau, Fontainebleau, France; ⁶Réanimation, CH de Melun, Melun, France; ⁷Réanimation, CH V Dupouy, Argenteuil, France; ⁸Réanimation, CH de Saint-Denis, Saint-Denis, France; ⁹Réanimation, CH Claude gallien, Qincy Sous Senart, France; ¹⁰Réanimation, CHU de Nantes, Nantes, France; ¹¹Réanimation, Hopital Salengro, Lille, France; ¹²Réanimation, CHU Saint-Antoine, Paris, France

Correspondence: Jean-Rémi Lavillegrand (JRLavillegrand@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-51

Rationale: Acute cholangitis (AC), a bacterial infection related to an obstruction of the biliary tree, may be responsible for life-threatening organ failure. However, little is known about the outcome and the predictive factors of mortality of critically ill patients admitted in ICU for acute cholangitis. We aimed to describe characteristics of patients admitted in ICU for AC and to analyze predictive factors of in-hospital mortality including the time to biliary drainage procedure.

Patients and methods: Retrospective study of all cases of acute cholangitis admitted in 11 French ICUs (5 tertiary hospitals and 6 non-tertiary hospitals) for continuous period between 2005 to 2018. Data are expressed as percentage or median (1^st IQR; 3^rd IQR).

Results: Overall, 382 patients (age: 72 [62; 81] years; 64% men) were included, in-hospital mortality was 29%. SOFA score at admission was 8 [5; 11] and SAPS2 was 50 [38; 64]; 38% of patients underwent mechanical ventilation and 61% received vasopressor therapy. Biliary obstruction was mainly related to gallstone (53%) and cancer (22%).

Median total bilirubin and PCT were respectively, 83 [50; 147] µmol/L and 19.1 [5.3; 54.8] µg/L. 63% of patients (n = 252) have positive blood culture, mostly Gram negative bacilli (86%) and 14% producing extended spectrum beta lactamase Enterobacteriaecae. At ICU admission, persisting obstruction was frequent (79%) and therapeutic endoscopic retrograde cholangiopancreatography was performed in 76% of them.

In a multivariable analysis, at ICU admission, several factors were significantly associated with in-hospital mortality: SOFA score (OR = 1.14 [95% IC 1.05; 1.24] by point, p = 0.001), arterial lactate (OR = 1.21 [1.08; 1.36] by 1 mmol/L, p < 0.001), total serum bilirubin (OR = 1.26 [1.12; 1.41] by 50 umol/L, p < 0.001), obstruction non-related to gallstones (P < 0.05) and AC complications (liver abcess and/or pancreatitis) (OR = 2.74 [1.45; 5.17] p = 0.002). In addition, time > 48 h between ICU admission and biliary drainage was associated to in-hospital mortality (adjusted OR = 2.73 [1.30; 6.22] p = 0.02).

Conclusion: Acute cholangitis is responsible for high mortality in ICU. Organ failure severity, causes and local complications of cholangitis are predictive factors of mortality as well as delayed biliary drainage.

Compliance with ethics regulations: Yes.

CO-52 Liver transplantation for critically ill cirrhotic patients

Thierry Artzner¹, Baptiste Michard¹, Emmanuel Weiss², Louise Barbier³, Zaid Noorah⁴, Jean-Claude Merle⁴, Catherine Paugam², Claire Francoz², Francois Durand², Olivier Soubrane², Tasneem Pirani⁵, John O’grady⁵, Nigel Heaton⁵, Eleni Theocharidou⁵, William Bernal⁵, Ephrem Salame³, Helene Barraud³, Petru Bucur³, Francois Lefebvre¹, Lawrence Serfaty¹, Camille Besch¹, Philippe Bachellier¹, Francis Schneider¹, Eric Lavesque⁴, François Faitot¹

¹Hôpital de Hautepierre, Strasbourg, France; ²Hôpital Beaujon, Clichy, France; ³Hôpital Trousseau, Tours, France; ⁴Hôpital Henri Mondor, Créteil, France; ⁵King’s College Hospital, London, UK

Correspondence: Thierry Artzner (thierry.artzner@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-52

Rationale: Cirrhotic patients with acute-on-chronic liver failure associated with extra-hepatic organ failure have dismal transplant-free prognosis in the intensive care unit. Liver transplantation is the only effective treatment available for them and a number of studies have shown promising results for critically ill cirrhotic patients. However, there are no recommendations or data in the literature to guide clinicians who need to decide what the optimal transplantability window for these patients is and which patients are too sick to be transplanted. The aim of this study was to develop a prognostic model to predict post-transplantation survival in critically ill cirrhotic candidates for liver transplantation with multiple organ failure.

Patients and methods: Patients with grade-3 acute-on-chronic liver failure (ACLF-3) who underwent liver transplantation between January 1, 2007 and June 30, 2017 in 5 transplant centers (4 in France and 1 in the United Kingdom) were included (n = 152). Predictors of one-year mortality were retrospectively screened and tested on a single center training cohort. A predictive score was developed and tested on an independent multicenter cohort.

Results: Four independent pre-transplantation risk factors were associated with one-year mortality after transplantation in the training cohort: age ≥ 53 years (OR = 5.79, 95% CI = 1.05–32.04, p = 0.044), pre-transplantation arterial lactate level ≥ 4mml/l (OR = 9.99, 95% CI = 1.62–61.46, p = 0.013), mechanical ventilation with PaO2/FiO2 ≤ 200 mmHg (OR = 8.60, 95% CI = 1.30–56.88, p = 0.026) and pre-transplantation leukocyte count ≤ 10G/l (OR = 12.91, 95% CI = 2.26–73.87, p = 0.004). A simplified version of the model was derived by assigning 1 point to each risk factor: the Transplantation for ACLF-3 Model (TAM) score. A cut-off at 2 points distinguished a high-risk group (score > 2) from a low-risk group (score ≤ 2) with one-year survival of 8.3% vs. 83.9% respectively (p < 0.001). The model and its simplified version were validated on the independent multicenter cohort. There was a significant difference between the high-risk and low-risk group with one-year survival of 10% vs. 71.9% respectively (p < 0.001).

Conclusion: Liver transplantation can be an effective treatment for critically ill cirrhotic patients with hepatic and extra hepatic organ failure provided patients are carefully selected and that they are transplanted at the optimal time in the intensive care. The TAM score can help stratify post-transplantation survival and assist clinicians in the transplantation decision-making process at the bedside of ACLF-3 patients.

Compliance with ethics regulations: Yes.

Fig. 1

Survival rate after liver transplantation in the training cohort (A, n = 74) and in the validation cohort (B, n = 74) in high risk (red line) and low risk (black line) patients

CO-53 Does Echocardiography Improve Clinical Outcomes in Shock States?

Habiba Hemamid, Mosbah Nabil, Abdelmalek Hakimi

CHU de Sétif, Setif, ALGERIA

Correspondence: Habiba Hemamid (habibahemamid@yahoo.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-53

Rationale: Trans-thoracic echocardiography (TTE) is commonly used in the initial management of patients with shock in ICU. There is little published evidence for any mortality benefit. We compared the effect of echocardiography protocol versus standard care for survival and clinical outcomes.

Patients and methods: This randomized controlled trial included selected shocked patients (systolic blood pressure < 90 mm Hg and signs of organ hypoperfusion) randomized to early TTE plus standard care versus standard care without TTE. The primary outcome measure was survival to 28 days. Secondary outcome measures included initial treatment and vasopressor weaning.

Results: 202 consecutive subjects with circulatory shock (low systolic arterial blood pressure (SAP) and signs of organ hypoperfusion) at the time of ICU admission are included in the study. In the TTE group: fluid prescription during the first 24 h was significantly lower (1519.62 ± 786.19 ml vs 1828.88 ± 1158.94 ml, P = 0.042), the vasopressor dose during the first 24 h was significantly lower (0.582 µg/kg/min [0.497–0.667] vs. 0.458 µg/kg/min [0.373–0.543] p = 0.044). TTE patients were weaned of vasopressors more quickly than those in the no TTE group (38.4 ± 20.88 vs. 57.8 ± 33.6 h p = 0.005). TTE patients received more dobutamine during the 72 h (165 versus 82 prescriptions). We found no important difference between groups for the primary outcome of survival (TTE group 61 of 100 patients versus standard care 67 of 102 patients).

Conclusion: TTE guided management was associated with, less fluid, less vasopressor dosing, faster vasopressor weaning and increased inotropic prescription. But TTE did not improve survival.

Compliance with ethics regulations: Yes.

CO-54 Prognostic assessment of left ventricular diastolic and systolic function during septic shock

Philippe Vignon¹, Cyril Charron², Marine Goudelin³, Annick Legras⁴, Michel Slama⁵, Gwenael Prat⁶, Stein Silva⁷, David Vandroux⁸, Gregoire Muller⁹, Bruno Lévy¹⁰, Florence Boissier¹¹, Bruno Evrard³, Xavier Repessé², Julien Maizel⁵, Guillaume Geri², Julie Leger¹², Antoine Vieillard-Baron²

¹Réanimation polyvalente/Inserm CIC 1435/Inserm UMR 1092, CHU Dupuytren/Université de Limoges, Limoges, France; ²Médecine Intensive Réanimation, CHU Ambroise Paré, APHP, Boulogne Billancourt, France; ³Réanimation polyvalente/Inserm CIC1435, CHU Dupuytren, Limoges, France; ⁴Médecine Intensive Réanimation, CHU Bretonneau, Tours, France; ⁵Réanimation médicale, CHU Amiens, Amiens, France; ⁶Réanimation médicale, CHU Brest, Brest, France; ⁷Réanimation polyvalente, CHU Purpan, Toulouse, France; ⁸Réanimation polyvalente, CHU Félix Guyon, Saint Denis De La Réunion, France; ⁹Réanimation polyvalente, CHR d’Orléans, Orléans, France; ¹⁰Réanimation médicale, Hôpitaux de Brabois, CHU de Nancy, Vandoeuvre-Lès-Nancy, France; ¹¹Réanimation médicale, CHU La Milétrie, Poitiers, France; ¹²Inserm CIC 1415, CHU Bretonneau, Tours, France

Correspondence: Philippe Vignon (philippe.vignon@unilim.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-54

Rationale: Both the negative prognostic value and reversibility of left ventricular (LV) diastolic dysfunction in septic patients remain debated. The excess of mortality in septic shock patients with hyperdynamic profile has only been reported by small-size studies. Accordingly, the primary objective of the PRODIASYS study was to assess the impact of LV diastolic dysfunction (and its severity) and of LV hyperkinesia echocardiographically identified during the initial phase of septic shock on 28-day survival. The secondary objective was to assess the potential link between LV diastolic dysfunction, cumulative water balance (on day 4), and outcome.

Patients and methods: This was a multicenter, prospective, observational, cohort study. Patients older than 18 years hospitalized in ICU for septic shock (Sepsis-3 definition) were eligible. Exclusion criteria were administration of inotropes, severe left valvular disease, constrictive pericarditis and moribund patients. In each patient, echocardiography was first performed within 12 h after the diagnosis of septic shock and then daily until Day 3, after vasopressor discontinuation, at ICU discharge and on Day 28 or at hospital discharge, whichever occurred first. Vital and biological parameters usually monitored for septic shock management were collected at each echocardiographic assessment. Vital status was collected on Day 28. Associations between LV diastolic dysfunction or LV hyperkinesia and Day-28 mortality were analyzed using a Chi2 test. Adjusted analyses were performed using logistic regression models, including variables known to be linked with the prognosis of septic shock (e.g., severity scores, delay of antibiotherapy). The relationship between the grade (I to III) of LV diastolic dysfunction and 28-day survival were analyzed using a logistic regression model. The relationship between the presence of LV diastolic dysfunction and cumulated water balance on Day 4 were analyzed using a linear regression model adjusted on the body weight on admission. The relationship between the grade of LV diastolic dysfunction and cumulated water balance on Day 4 were analyzed using a linear regression model.

Results: From January 2017 to September 2019, the study enrolled the initially planned 440 patients for a total of more than 1400 echocardiographic examinations performed in the 10 participating sites.

Conclusion: This study promises to determine whether LV diastolic dysfunction using up-to-date consensual definition is independently linked to mortality of septic shock patients or not, whether it is associated with positive fluid balance and potentially reversible in survivors, and whether the hyperkinetic hemodynamic profile is prognostic in septic shock.

Compliance with ethics regulations: Yes.

CO-55 Perfusion index: a non-invasive way to detect the effects of the end-expiratory occlusion test in critically ill patients

Alexandra Beurton, Francesco Gavelli, Jean-Louis Teboul, Nello De Vita, Xavier Monnet

Service de médecine intensive-réanimation, Hôpital de Bicêtre, Hôpitaux universitaires Paris-Saclay, Assistance publique—Hôpitaux de Paris, Inserm UMR S_999, Université Paris-Su, Le Kremlin-Bicêtre, France

Correspondence: Alexandra Beurton (alexandra.beurton@aphp.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-55

Rationale: The end-expiratory occlusion (EEXPO) test consists in interrupting mechanical ventilation for 15-s and observing its effects on cardiac output. An increase in cardiac index (CI) ≥ 5% detects preload responsiveness. However, it has been described only with direct measurements of CI, as demonstrated with pulse contour analysis and echocardiography. The perfusion index (PI) is the ratio between the pulsatile and the non-pulsatile portions of the plethysmography signal. It is in part determined by stroke volume and can detect a positive passive leg raising (PLR) test. We hypothesised that the changes in PI could detect a positive EEXPO test and thus preload responsiveness in a totally non-invasive way.

Patients and methods: In critically ill patients, we measured PI (Radical 7, Masimo) and CI (PiCCO2, Pulsion Medical Systems) before and during a PLR test, a 15-s EEXPO test and, if decided, before and after volume expansion (VE) (500-mL saline).

Results: We included 31 patients. No patient was excluded due to unstable or absent plethysmographic signal. In 19 patients with a positive PLR test (increase in CI ≥ 10%), CI and PI increased during PLR by 17 ± 7% and 49 ± 23%, respectively, and during EEXPO test by 6 ± 2% and 11 ± 8%, respectively. In the 12 patients with a negative PLR test, PI did not significantly change during PLR and EEXPO test (4 ± 4% and 0 ± 1%). Only four patients received VE, causing an increase of CI and PI by 15 ± 3% and 32 ± 47% respectively. The correlation between the PI and CI PLR-induced changes was significant (r = 0.76, p < 0.001). During the PLR test, if PI increased by ≥ 9%, a positive response of CI (≥ 10%) was diagnosed with a sensitivity of 95 (74–100)% and a specificity of 75 (43–95)% (area under the receiver operating characteristics curve (AUROC): 0.94 (0.80–0.99), p < 0.0001 vs. 0.50). The correlation between the PI and CI EEXPO test-induced changes was significant (r = 0.69, p < 0.0001). During the EEXPO test, if PI increased by > 5%, a positive response of CI to the EEXPO test (≥ 5%) was diagnosed with a sensitivity of 87 (60–98)% and a specificity of 94 (70–100)% (AUROC curve: 0.92 (0.77–0.99), p < 0.0001). During the EEXPO test, if PI increased by > 3%, a positive response of CI to the PLR test (≥ 10%) was diagnosed with a sensitivity of 90 (67–99)%, p < 0.0001.

Conclusion: An increase in PI ≥ 5% during an EEXPO test accurately detects a positive response of CI to the EEXPO test. An increase in PI during PLR by 9% accurately detects a positive response of the PLR test as we have already shown.

Compliance with ethics regulations: Yes.

CO-56 Mini and micro fluid challenge predicts preload responsiveness in critically ill patients

Nello de Vita, Francesco Gavelli, Jean-Louis Teboul, Arthur Pavot, Rui Shi, Xavier Monnet

Service de médecine intensive-réanimation, Hôpital de Bicêtre, Hôpitaux universitaires Paris-Saclay, Assistance publique—Hôpitaux de Paris, Inserm UMR S_999, Université Paris-Sud, Le Kremlin-Bicêtre, France

Correspondence: Nello de Vita (Nellodevita@hotmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-56

Rationale: The administration of a fluid bolus of 500 mL (fluid challenge) has been used to test preload responsiveness in critically ill patients. However, if the patient is not preload responder, the amount of fluid given may be detrimental, contributing to fluid overload. Thus, smaller volumes of fluid bolus have been proposed as an alternative method to test preload responsiveness.

Patients and methods: In patients with acute circulatory failure equipped with a PiCCO2 device (Pulsion Medical Systems, Feldkirchen, Germany), we evaluated the ability of the changes in CO induced by the rapid infusion of 50 mL of normal saline over 30 s (micro-fluid challenge) and of 100 mL over 90 s (mini-fluid challenge) to predict preload responsiveness. In preload responders, a fluid bolus of 500 mL of normal saline was administered, according to the judgement of the treating physician.

Results: Between May and September 2019, twenty-three consecutive patients were included. Septic shock was the main cause of acute circulatory failure in 21 patients (91%), while hypovolemia was the main cause in two patients (7%). Norepinephrine was administered to all the enrolled patients at the time of inclusion (0.35 [0.13–0.38] mcg/kg/min). Among the included patients, nine (49%) were preload responders, defined according to a PLR-induced increase in CO ≥ 10%, whereas 14 (61%) were non-responders. Micro-fluid challenge-induced changes in CO were significantly different among preload responders and non-responders (9.4[3.5–15.9]% vs. 0.0[0.0–0.7]%, respectively; p = 0.001). Analogously, mini-fluid challenge-induced changes in CO were different between responders and non-responders (11.2[7.2–15.3]% vs. 0.7[0.0–3.5]%, respectively; p = 0.002). The area under the receiver operating characteristic curve for micro and mini-fluid challenge to predict preload responsiveness was 0.968 ± 0.03 (p < 0.0001), with a sensitivity of 100% and specificity of 86% for both tests. The diagnostic threshold of micro-fluid challenge was 1.1%, whereas for mini-fluid challenge it was 4.1%.

Conclusion: Both micro and mini-fluid challenge reliably predict preload responsiveness in critically ill patients when their effects on cardiac output are detected by the pulse contour analysis method.

Compliance with ethics regulations: Yes.

CO-57 Impact of diaphragm dysfunction on cardiac function and lung aeration during weaning

Emmanuel Rozenberg, Elise Morawiec, Julien Mayaux, Julie Delemazure, Thomas Similowski, Alexandre Demoule, Martin Dres

AP-HP. Sorbonne Université, Hôpital Pitié-Salpêtrière, Service de Pneumologie, Médecine intensive—Réanimation (Département “R3S”), F-75013, Paris, France

Correspondence: Emmanuel Rozenberg (emmanuelrozenberg@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-57

Rationale: Diaphragm dysfunction and weaning induced pulmonary edema are two frequent causes of weaning failure but their coexistence and interaction have been poorly investigated. We hypothesized that diaphragm dysfunction may not induce a sufficient decrease in intra-thoracic pressure to increase venous return and generate a weaning induced pulmonary edema. We therefore investigated whether weaning induced pulmonary edema and diaphragm dysfunction are or not associated and evaluated the effect of diaphragm dysfunction on cardiac function and lung aeration during a spontaneous breathing trial (SBT).

Patients and methods: Patients with readiness to wean criteria who had failed a first SBT were eligible. Before and after a second SBT, diaphragm function was assessed by measuring the change in tracheal pressure induced by a bilateral phrenic nerve stimulation (Ptr, stim), cardiac function (cardiac output, systolic pulmonary arterial pressure) was evaluated with echocardiography and lung aeration was estimated from the lung ultrasound score (LUS). Plasma protein concentration and hemoglobin were also sampled before and after the SBT. Diaphragm dysfunction was defined by Ptr, stim < − 7 cmH2O and weaning induced pulmonary edema was diagnosed in case of SBT failure associated with 1) increase in plasma protein concentration or hemoglobin > 5% during the spontaneous breathing trial and/or 2) early (E) over late peak diastolic velocity ratio > 0.95 or E over peak diastolic velocity ratio > 8.5.

Results: Fifty-three patients were included and 31/53 (58%) failed the SBT. Diaphragm dysfunction was present in 20/24 (83%) of patients with weaning induced pulmonary edema, in 11/22 (50%) patients with SBT success and in 2/7 (29%) patients with other causes of SBT failure (p < 0.01). During the SBT, diaphragm dysfunction induced a significant increase in systolic pulmonary arterial pressure but no change in cardiac output. Patients with diaphragm dysfunction had a higher LUS as compared to their counterparts (10 ± 9 vs. 3 ± 4, respectively, p < 0.01).

Conclusion: Diaphragm dysfunction induces a loss of lung recruitment and a significant increase in systolic pulmonary arterial pressure during the SBT. Coexistence of diaphragm dysfunction and weaning induced pulmonary edema is common in case of SBT failure but weaning induced pulmonary edema appears more likely to be involved than diaphragm dysfunction.

Compliance with ethics regulations: Yes.

CO-58 Shear wave elastography, a new tool for diaphragmatic qualitative assessment in critically ill patients

Yassir Aarab¹, Aurelien Flatres², Fanny Garnier¹, Romaric Larcher¹, David Chapeau¹, Kada Klouche¹, Pascal Etienne³, Gilles Subra⁴, Samir Jaber⁵, Nicolas Molinari⁶, Lucie Gamon⁶, Stefan Matecki², Boris Jung¹

¹Medical Intensive Care Unit, Montpellier University and Montpellier Teaching Hospital, Montpellier, France; ²INSERM U1046, CNRS UMR9214, Université de Montpellier, Montpellier, France; ³Laboratoire Charles Coulomb (L2C), University of Montpellier, CNRS, Montpellier, France; ⁴Institut des Biomolécules Max Mousseron (IBMM), UMR5247 CNRS, ENSCM, Université de Montpellier, Montpellier, France; ⁵Saint Eloi Anesthesiology and Critical Care Medicine, Montpellier University and Montpellier Teaching Hospital, Montpellier, France; ⁶Biostatistics Department, Montpellier University and Montpellier Teaching Hospital, Montpellier, France

Correspondence: Yassir Aarab (y-aarab@f-montpellier.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-58

Rationale: Diaphragmatic weakness in the Intensive Care Unit (ICU) is associated with poor outcome. Prolonged mechanical ventilation is associated either with a decrease (atrophy) or an increase (supposed injury) in diaphragmatic thickness, both associated with prolonged weaning. Shear wave elastography is a non-invasive technique that measures diaphragm shear modulus (SM), a surrogate of its mechanical properties. The aim of this study was to describe the diaphragm shear modulus during the ICU stay and to describe its relation with diaphragm thickness.

Patients and methods: This prospective and monocentric study included all consecutive critically ill patients. Ultrasound examination of the diaphragm (Aixplorer; SuperSonic-Imagine, Aix-en-Provence, France) was obtained by two investigators every other day until ICU discharge. Demographics, diaphragm thickness, SM and outcomes were collected. A mixed model regression was used to study the relation between SM and diaphragm thickness.

Results: We enrolled 102 patients from 2017 December 1st to 2018 June 1st, 88 being invasively mechanically ventilated during the stay. Diaphragm ultrasound evaluation was feasible in 94/102 (96%) patients. The duration of mechanical ventilation during the ICU stay was 7 [4–15] days with 3 [1–5] days spent on controlled mechanical ventilation. SM was 14.3 ± 4.3 kPa and diaphragm end-expiratory thickness was 0.20 ± 0.05 cm upon ICU admission. Increase and decrease ≥ 10% during ICU stay occured in 37 and 49 percent of the patients respectively for diaphragmatic thickness, and in 51 and 41 percent of the patients respectively for diaphragmatic SM. Diaphragm thickness over time was inversely correlated with diaphragm SM and with time spent under mechanical ventilation (Table). Diaphragm SM over time was correlated with time spent under pressure support ventilation or under spontaneous breathing (compared to controlled ventilation) and with time spent under deep sedation. Diaphragm SM was inversely correlated with age, sepsis, exposition to steroids (Table). No association was found between diaphragm SM and outcomes.

Discussion: Our results are in line with the myotrauma concept, suggesting alteration in diaphragm mechanical properties associated with increased diaphragm thickness in critically ill patients. We hypothesize that this observation most likely reflects muscle injury and tissue infiltration with edema and inflammatory cells.

Conclusion: Shear wave ultrasound elastography suggests that in critically ill patients, the increase in diaphragmatic mass is associated with an alteration in diaphragm mechanical properties as measured by SM.

Compliance with ethics regulations: Yes.

Table 1 Factors associated with changes in diaphragm shear modulus and thickness during ICU stay. Adjusted multivariate analysis

CO-59 ICU-acquired weakness but not diaphragm dysfunction strongly impacts long-term survival of critically ill patients

Clément Saccheri, Elise Morawiec, Julie Delemazure, Julien Mayaux, Bruno-Pierre Dubé, Thomas Similowski, Alexandre Demoule, Martin Dres

Service de Pneumologie, Médecine intensive Réanimation, Hôpital Pitié Salpêtrière, Groupe hospitalier universitaire APHP. Sorbonne Université, Paris, France

Correspondence: Clément Saccheri (clement.saccheri@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-59

Rationale: Diaphragm dysfunction and intensive care unit (ICU) acquired weakness (ICU-AW) are associated with poor outcomes in the ICU but their long term impact on prognosis and health-related quality of life (HRQOL) is poorly established. This study sought to determine whether diaphragm dysfunction is associated with negative long-term outcomes and whether the coexistence of diaphragm dysfunction and ICU-AW has a particular impact on two-year survival and HRQOL.

Patients and methods: We used a previous cohort study conducted in our institution to follow up mechanically ventilated patients in whom diaphragm and limb muscle functions were investigated at the time of liberation from mechanical ventilation. Diaphragm dysfunction was defined by tracheal pressure generated by phrenic nerve stimulation < 11 cmH2O and ICU-acquired weakness was defined by Medical Research Council (MRC) score < 48. HRQOL was evaluated with the SF-36 questionnaire.

Results: Sixty-nine of the 76 patients enrolled in the original study were included in the survival analysis and 40 were interviewed. Overall two-year survival was 67% (46/69): 62% (28/45) in patients with diaphragm dysfunction, 71% (17/24) in patients without diaphragm dysfunction, 46% (11/24) in patients with ICU-acquired weakness and 76% (34/45) in patients without ICU-acquired weakness. Patients with concomitant diaphragm dysfunction and ICU-acquired weakness had a poorer outcome with a 2-year survival rate of 36% (5/14) compared to patients without diaphragm function and ICU-acquired weakness (79% (11/14) (p < 0.01)). HRQOL was not influenced by the presence of ICU-acquired weakness, diaphragm dysfunction or their coexistence.

Conclusion: ICU-acquired weakness but not diaphragm dysfunction has a strong negative impact on two-year survival of critically ill patients. The presence of diaphragm dysfunction appears more likely to be a determinant of early prognosis and does not appear to have a significant impact on long-term survival.

Compliance with ethics regulations: Yes.

CO-60 Role of limb weakness on extubation outcome among patients at high-risk of reintubation in the ICU

Arnaud Thille¹, Florence Boissier¹, Michel Muller², Albrice Levrat², Gaël Bourdin³, Sylvène Rosselli³, Jean-Pierre Frat¹, Rémi Coudroy¹, Emmanuel Vivier³

¹CHU de Poitiers, Médecine Intensive Réanimation ALIVE research group INSERM CIC 1402, Poitiers, France; ²Centre Hospitalier Annecy Genevoix, Réanimation Polyvalente, Metz-Tessy, France; ³Centre Hospitalier Saint Joseph Saint Luc, Réanimation Polyvalente, Lyon, France

Correspondence: Arnaud Thille (aw.thille@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-60

Rationale: Whereas ICU-acquired limb weakness may delay extubation in mechanically ventilated patients, its influence on extubation failure is poorly known. Research question: This study aimed at assessing the role of limb weakness on extubation failure in the ICU and the relation between limb weakness and cough strength.

Patients and methods: A secondary analysis of two previous prospective studies including patients at high-risk of reintubation after planned extubation, i.e. age greater than 65 years, with underlying cardiac or respiratory disease, or intubated for more than 7 days prior to extubation. Patients intubated less than 24 h and those with a do-not-reintubate order were not included. Limb and cough strength were assessed by physiotherapist just before extubation. Limb weakness was defined as Medical Research Council (MRC) score < 48 points, and severe weakness as MRC score < 36. Cough strength was assessed using a semi-quantitative 5-Likert scale. Extubation failure was defined as reintubation or death within the first 7 days following extubation.

Results: Among 344 patients at high-risk, 16% experienced extubation failure (56/344). These patients had greater severity, lower MRC score (41 ± 16 vs. 49 ± 13, p < 0.001) and were more likely to have ineffective cough than the others. Extubation failure rate were 12% (25/214) in patients with no limb weakness vs. 18% (12/65) and 29% (19/65) in those with moderate and severe limb weakness, respectively (p < 0.01). MRC score and cough strength were weakly but significantly correlated (Rho = 0.28, p < 0.001). After multivariate logistic regression analyses, the lower the MRC score the greater the risk of reintubation; and severe limb weakness was independently associated with extubation failure, even after adjustment on cough strength at time of extubation and severity score (SAPS II) at admission.

Conclusion: Limb weakness was independently associated with extubation failure in patients at high-risk of reintubation in the ICU.

Compliance with ethics regulations: Yes.

CO-61 Lung bacteriobiota but not mycobiota predicts day 28 mortality among critically ill influenza patients

Renaud Prével¹, Raphaël Enaud², Alexandre Massri³, Erwan Begot¹, Laurent Villeneuve⁴, Fabien Beaufils², Marie-Edith Lafon⁵, Camille Ciccone⁵, Patrick Berger², Alexandre Boyer¹, Laurence Delhaes⁶, Didier Gruson¹

¹CHU Bordeaux, Medical Intensive Care Unit, Bordeaux, France; ²Inserm U1045 Centre de Recherche Cardio-thoracique, Bordeaux, France; ³Intensive Care Unit, CH Francois Mitterrand, Pau, France; ⁴Bacteriology laboratory, CH Francois Mitterrand, Pau, France; ⁵CHU Bordeaux, Virology laboratory, Bordeaux, France; ⁶CHU Bordeaux, Mycology laboratory, Bordeaux, France

Correspondence: Renaud Prével (renaud.prevel@hotmail.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-61

Rationale: Influenza can lead to severe condition with acute respiratory failure and acute respiratory distress syndrome due to a massive pulmonary inflammatory in response to the viral invasion. Lung bacteriobiota has been described to be associated with pulmonary inflammation in chronic respiratory diseases such as chronic obstructive pulmonary disease or cystic fibrosis. Lung mycobiota has been poorly investigated despite the well-known role for fungi in numerous respiratory diseases. The aim of our study was to assess the prognostic value of lung bacteriobiota and mycobiota among critically ill influenza patients.

Patients and methods: We prospectively included influenza patients admitted to ICU. Sputum were stored a -80 °C. Bacterial and fungal DNA were extracted thanks to QIAamp^® PowerFecal^® Pro DNA Kit. 16S rRNA gene V3-V4 regions and ITS2 regions were amplified by PCR and sequenced on Illumina MiSeq^®. Taxonomic assignation was obtained by DADA2 pipeline and microbiota analysis were performed according to day-28 mortality by the mean of Phyloseq package on R 3.6.0 software.

Results: Thirty-nine patients were admitted to ICU for influenza with 23 sputa available and finally 18 DNA samples available after extraction. Bacteriobiota alpha diversity was significantly lower among non-survivors than survivors when expressed by the mean of Shannon index, Simpson index or Evenness (respectively p = 0.01, p = 0.02, p = 0.01). Area under the curve to predict day-28 mortality was 0.91, 95CI [0.77; 1.00] for Shannon index, 0.89 95CI [0.74; 1.00] for Simpson index and 0.91 95CI [0.77; 1.00] for evenness. β-diversity analysis also demonstrated significant differences between survivors and non-survivors (adjusted permutational multivariate ANOVA, p  =  0.01). Nonsurvivors had a higher abundance of Staphylococcus, Haemophilus, Streptococcus and Moraxella. None of the fungal alpha-diversity index nor beta-diversity were significantively different between survivors and non-survivors. Non-survivors had a higher proportion of Candida albicans and Malassezia but not of Aspergillus.

Conclusion: The lung bacteriobiota profile, but not the mycobiota one, of critically ill influenza patients is associated with day-28 mortality and may be used to identify subjects with a poor prognosis at the time of admission.

Compliance with ethics regulations: Yes.

CO-62 Pathway mapping of leukocyte transcriptome in influenza patients reveals distinct pathogenic mechanisms associated with progression to severe infection

Yoann Zerbib¹, Emily Jenkins², Maryam Shojaei², Anthony Mclean², Klaus Schughart³, Benjamin Tang²

¹Department of medical Intensive Care, Amiens University Hospital, Amiens, France; ²Department of Intensive Care Medicine, Nepean Hospital, Sydney, Australia; ³Helmholtz Centre for Infection Research, Braunschweig, Germany

Correspondence: Yoann Zerbib (zerbib.yoann@chu-amiens.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-62

Rationale: Influenza infections produce a spectrum of disease severity, ranging from a mild respiratory illness to respiratory failure and death. The host-response pathways associated with the progression to severe influenza disease are not well understood.

Patients and methods: To gain insight into the disease mechanisms associated with progression to severe infection, we analyzed the leukocyte transcriptome in severe and moderate influenza patients and healthy control subjects. Pathway analysis on differentially expressed genes was performed using a topology-based pathway analysis tool that takes into account the interaction between multiple cellular pathways. The pathway profiles between moderate and severe influenza were then compared to delineate the biological mechanisms underpinning the progression from moderate to severe influenza.

Results: 107 patients (44 severe and 63 moderate influenza patients) and 52 healthy control subjects were included in the study. Severe influenza was associated with upregulation in several neutrophil-related pathways, including pathways involved in neutrophil differentiation, migration, degranulation and neutrophil extracellular trap (NET) formation. The degree of upregulation in neutrophil-related pathways was significantly higher in severely infected patients compared to moderately infected patients. Severe influenza was also associated with downregulation in immune response pathways, including pathways involved in antigen presentation, CD4+ T-cell co-stimulation, CD8+ T cell and Natural Killer (NK) cells effector functions. Apoptosis pathways were also downregulated in severe influenza patients compared to moderate and healthy controls.

Conclusion: These findings showed that there are changes in gene expression profile that may highlight distinct pathogenic mechanisms associated with progression from moderate to severe influenza infection.

Compliance with ethics regulations: Yes.

CO-63 Herpes Simplex Virus and Cytomegalovirus reactivation among severe ARDS patients under veno-venous ECMO

Sami Hraiech¹, Eline Bonnardel², Christophe Guervilly¹, Jean-Marie Forel¹, Melanie Adda¹, Gabriel Parzy¹, Guilhem Cavaille¹, Benjamin Coiffard¹, Laurent Papazian¹

¹Service de Médecine Intensive-Réanimation, APHM, Hôpital Nord, Marseille, France; ²Magellan Medico-Surgical Center, South Department of Anaesthesia and Critical Care, Bordeaux, France

Correspondence: Sami Hraiech (sami.hraiech@ap-hm.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-63

Rationale: Herpesviridae reactivation among non-immunocompromised critically ill patients is associated with impaired prognosis, especially during acute respiratory distress syndrome (ARDS). However, few is known about Herpes Simplex Virus (HSV) and Cytomegalovirus (CMV) reactivation occurring in patients with severe ARDS under veno-venous ExtraCorporeal Membrane Oxygenation (ECMO). We tried to determine the frequency of herpesviridae reactivation and its impact on patients’prognosis during ECMO for severe ARDS.

Patients and methods: We conducted an observational, retrospective study in a medical ICU (ARDS and ECMO referee center) between 2011 and 2017. Patients with a severe ARDS requiring a venovenous ECMO for 2 days or more were included. HSV and/or CMV reactivation occurring after ECMO insertion was screened for these patients. Patients with immunosuppression, antiviral therapy against HSV and/or CMV prior to inclusion, or HSV/CMV reactivation known at the time of ECMO insertion were excluded. HSV reactivation was defined by a positive qualitative throat sample (Virocult^®) PCR or positive broncho-alveolar lavage (BAL) PCR. CMV reactivation was defined by a positive quantitative blood or BAL PCR.

Results: During a five-year period, 123 non-immunocompromised patients with a severe ARDS necessitating a veno-venous ECMO were included. Sixty-seven (54%) experienced HSV and/or CMV reactivation during ECMO course (20 viral co-infection, 40 HSV alone and 7 CMV alone). HSV reactivation occurred earlier than CMV after the beginning of MV (10 (6–15) vs. 19 (13–29) days; p < 0.01) and after ECMO implementation (4 (2–8) vs. 14 (10–20) days; p < 0.01). In univariate analysis, HSV/CMV reactivation was associated with a longer duration of mechanical ventilation (34 (22–52.5) vs. 17.5 (9–28) days; p < 0.01), a longer duration of ECMO (15 (10–22.5) vs. 9 (5–14) days; p < 0.01), and a prolonged ICU (29 (19.5–47.5) vs. 16 (9–30) days; p < 0.01) and hospital stay (44 (29–63.5) vs. 24 (11–43) days; p < 0.01). However, in multivariate analysis, viral reactivation remained associated with prolonged MV only. When comparing patients having CMV (alone or combined with HSV) vs. HSV reactivation alone, CMV positive patients had a longer mechanical ventilation duration and fewer ventilator-free days at day-28 and a longer ICU and hospital length of stay.

Conclusion: Herpesviridae reactivation is frequent among patients with sevre ARDS under veno-venous ECMO and is associated with a longer duration of mechanical ventilation. CMV seems to have a proper negative role on pulmonary fiunction as compared to HSV alone. HSV and CMV deserve to be researched in severe ARDS patients under ECMO.

Compliance with ethics regulations: Yes.

CO-64 Comparison of RSV and influenza infection in ICU patients-the CAPTIF study

Charlotte Vandueren¹, Benjamin Zuber², Eve Garrigues³, Antoine Gros⁴, Nicolas Epaillard⁵, Guillaume Voiriot⁶, Yacine Tandjaoui-Lambiotte⁷, Jean-Baptiste Lascarrou⁸, Florence Boissier⁹, Virginie Lemiale¹⁰, Damien Contou¹¹, Sami Hraiech¹², Anne-Pascale Meert¹³, Bertrand Sauneuf¹⁴, Aline Munting¹⁵, Sylvie Ricome¹⁶, Guillaume Geri³, Jonathan Messika¹⁷, Gregoire Muller¹⁸, Julien Coussement¹⁹, David Grimaldi²⁰

¹USI, Hôpital Erasme, ULB, Bruxelles, Belgium; ²Réanimation Hôpital Foch, Suresnes, France; ³Réanimation Hôpital Ambroise Paré, Boulogne, France; ⁴CH Versailles, Le Chesnay, France; ⁵Réanimation Hôpital Saint Antoine, Paris, France; ⁶Réanimation Hôpital Tenon, Paris, France; ⁷Réanimation Hôpital Avicennes, Bobigny, France; ⁸MIR CHU Nantes, Nantes, France; ⁹MIR CHU Poitiers, Poitiers, France; ¹⁰Réanimation Hôpital Saint Louis, Paris, France; ¹¹Réanimation CH Argenteuil, Argenteuil, France; ¹²Réanimation Hôpital Nord-APHM, Marseille, France; ¹³USI, Hôpital BORDET, ULB, Bruxelles, Belgium; ¹⁴Réanimation CH Public du Cotentin, Cherbourg, France; ¹⁵CHU UCL Namur, Godinne, Belgium; ¹⁶Réanimation CH Aulnay, Aulnay S/bois, France; ¹⁷Réanimation Hôpital Louis Mourier, Colombes, France; ¹⁸CH Orleans, Orleans, France; ¹⁹Maladies Infectieuses, Hôpital Erasme, ULB, Bruxelles, Belgium; ²⁰USI Hôpital Erasme ULB, Bruxelles, Belgium

Correspondence: Charlotte Vandueren (chavandueren@yahoo.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-64

Rationale: Respiratory syncytial virus (RSV) is a common cause of pediatric bronchiolitis and influenza-like illness in adults. Its involvement in severe infections in adults remains unclear. The CAPTIF study aimed at comparing characteristics and prognosis of ICU patients infected with RSV and influenza, assuming that, based on the limited evidence, the mortality of RSV infection would be lower than the influenza related one.

Patients and methods: Multicenter Franco-Belgian retrospective study. Adults admitted to 18 ICUs between 1/Nov/2011 and 30/Apr/2017 with respiratory RSV infection were included and matched 1:1 to influenza patients on center and ICU admission date. Patients’ characteristics, clinical presentation, and outcome were compared between groups using univariate and multivariable analyses.

Results: We report here the results for the first 470 cases among 650 included patients. Mean age was 65.5 (16.7) years and SAPS-2 score was 42 (17), not different between groups. Compared to influenza patients, RSV patients more frequently had chronic respiratory failure (61% vs 39%, p < 0.001) or immune suppression (36 vs 26%, p = 0.03). Frequencies of cardiac, renal and hepatic chronic diseases were similar. Almost all patients had respiratory symptoms (> 95%), extrarespiratory symptoms were more frequent in influenza patients (9 vs 15%, = 0.04). RSV patients more frequently had bronchospasm (51 vs 36%, p = 0.001). Clinical presentation such as ARDS (20%), shock (30%) and pulmonary coinfection (32%) were similar, however SOFA score was higher in RSV patients (4.6 (3.4) vs 5.6 (4), p = 0.004). The P/F ratio was around 210 mmHg in both groups, PaCO₂ was higher in RSV patients (55 vs 47 mmHg, < 0.001). Respiratory assistance at diagnosis tended to differ (p = 0.06), RSV patients receiving more non invasive ventilation (29 vs 19%) and less high flow oxygen therapy (10 vs 14%) but invasive ventilation was required similarly (36 vs 33%). During ICU stay, ARDS was more frequent in RSV patients (21 vs 30%, p = 0.03), accordingly prone position (1.3 vs 4.3%) and ECMO (2.5 vs 9.1%) were more frequently needed. Length of mechanical ventilation (2 days (0–8)) and ICU LOS (5 days (3–12)) were not different. ICU mortality was similar in RSV and influenza patients (18.4% and 21.3%), the multivariate analysis did not find an association between type of virus and mortality.

Conclusion: RSV infection is frequent in adult ICU patients. It presents more frequently than influenza as an acute on chronic respiratory failure with bronchospasm. Despite difference in case mix and clinical presentation, VRS severity and burden appear similar to influenza justifying effort to prevent and treat it.

Compliance with ethics regulations: Yes.

CO-65 Epidemiology and 1-year functional outcome of ICU-admitted stroke patients

Benoit Brassart, Ahmed El Kalioubie, Saad Nseir

Université Lille, Centre Hospitalier Universitaire Lille, Lille, France

Correspondence: Benoit Brassart (brassart.benoit@gmail.com)

Intensive Care 2020, 10 (Suppl 1):CO-65

Rationale: Stroke remains the major cause of severe disability and the second most common cause of death despite advances in prevention and treatments over the last two decades. New techniques of stroke management and ageing of population has led to an increase in ICU-admission of patients with stroke. The aim of our study was to describe the epidemiology and 1-year functional outcome of stroke patients admitted to our intensive care unit and to identify early predictors of poor outcome at admission in this population.

Patients and methods: Single Medical ICU Center at the Lille University Teaching Hospital. Retrospective analysis of patients with acute ischemic stroke (AIS) or non-traumatic intracranial hemorrhage (ICH) requiring intensive care admission between January 2008 and December 2015. Patients with subarachnoid hemorrhage were excluded. Admission (demographics, comorbidities, clinical data, cause of ICU admission), ICU (length of stay (LOS), adverse events, tracheotomy) and outcome data (in hospital and one year mortality, one year Modified Rankin Scale (mRS)) were collected from electronic records. Univariate then multivariate analysis were performed to identify risk factors for poor outcome (1 year mRS > 2).

Results: In the eight-year period, we admitted 556 patients, of whom 309 (56%) were male, with a mean age of 64 (52.76). 291 (52%) with AIS and 265 (48%) with ICH. Median admission NIHSS was 17. 483 patients (87%) were mechanically ventilated, for a median duration of 3 days (1.11). Patients were primarly admitted for neurologic (82.9%), then respiratory (12.9%) failures. Median ICU and hospital LOS were 4 (1.12) and 11 (2.36) days. Ventilation associated pneumonia occurred in 14.9%, and tracheotomy was performed in 13.5% of cases. A do-not-resuscitate decision was taken for 296 (53%) patients. 306 (55.1%) patients died during hospitalization. mRS > 2 (including death) at hospital discharge and at 1 year follow up involved respectively 91.2% and 84.4%. Only 87 (39.9%) of the 218 patients who were alive at one year were functionally independent. Admission risk factors for poor outcome (1-year mRS > 2), for combined AIS and ICH, were age [OR 1.25 95% IC (1.03, 1.53)], IGSII [OR 1.74 95% IC (1.44, 2.11)], endotracheal intubation [OR 2.31 95% IC (1.08, 4.97)]. In a sub-group analysis, vertebro-basilar AIC (101 (35%) patients), and lobar ICH (174 (66%) patients) were independently associated with a favorable 1-year mRS (respectively OR 0.23 95% IC (0.11, 0.50) and OR 0.24 95% IC (0.10.0.58)).

Conclusion: In our retrospective cohort, stroke patients requiring intensive care admission had high mortality rate and survivors were mostly dependent one year after. Further study is needed to better describe this population.

Compliance with ethics regulations: Yes.

CO-66 Association of reason for endotracheal intubation with long-term survival in acute stroke patients requiring mechanical ventilation

Etienne de Montmollin¹, Nicolas Terzi², Claire Dupuis³, Maïté Garrouste⁴, Daniel da Silva⁵, Michaël Darmon⁶, Virginie Laurent⁷, Guillaume Thiery⁸, Johanna Oziel⁹, Guillaume Marcotte¹⁰, Marc Gainnier¹¹, Shidasp Siami¹², Benjamin Sztrymf¹³, Christophe Adrie¹⁴, Jean Reignier¹⁵, Romain Sonneville¹, Jean-François Timsit¹

¹Medical and Infectious Diseases Intensive Care Unit, Bichat-Claude Bernard Hospital, Paris, France; ²Medical Intensive Care Unit, Grenoble University Hospital, La Tronche, France; ³Medical Intensive Care Unit, Gabriel Montpied University Hospital, Clermont-Ferrand, France; ⁴Medical Unit, French British Hospital, Levallois-Perret, France; ⁵Intensive Care Unit, Delafontaine Hospital, Saint-Denis, France; ⁶Medical Intensive Care Unit, Saint-Louis Hospital, Paris, France; ⁷Intensive Care Unit, André Mignot Hospital, Le Chesnay, France; ⁸Intensive care Unit, Hôpital Nord, Saint-Etienne, France; ⁹Intensive Care Unit, Avicenne Hospital, Bobigny, France; ¹⁰Surgical Intensive Care Unit, Edouard Herriot Hospital, Lyon, France; ¹¹Intensive Care Unit, La Timone Hospital, Marseille, France; ¹²Intensive Care Unit, Sud-Essonne Hospital, Etampes, France; ¹³Intensive Care Unit, Antoine Béclère Hospital, Clamart, France; ¹⁴Physiology department, Cochin Hospital, Paris, France; ¹⁵Medical Intensive Care Unit, Nantes University Hospital, Nantes, France

Correspondence: Etienne de Montmollin (edemontmollin@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-66

Rationale: Mortality in acute stroke patients requiring mechanical ventilation ranges from 60 to 90% at 1 year. Studies evaluating indicators of outcome in these patients have limitations, including single-center, retrospective designs and no adjustment for withholding/withdrawal of life-sustaining treatments (WLST). Our objective was to identify factors associated with 1-year survival in acute stroke patients requiring mechanical ventilation.

Patients and methods: Retrospective analysis of a prospective multicenter database between 1997 and 2016. ICU stroke patients entered in the database and requiring mechanical ventilation within 24 h were included. Were excluded patients with stroke of traumatic origin, subdural hematoma or venous cerebral thrombosis. Factors associated with 1-year survival were identified using a Cox model stratified on inclusion center, adjusted on WFLST occurring during the first 48 h. Data are presented as median [Q1-Q3] or percentages. Cox model results are presented as hazard ratios (HR) and 95% confidence intervals (CI).

Results: We identified 419 patients from 14 ICUs, aged 68 [58–76] years and 60% males. On admission, the Glasgow coma score (GCS) was 4 [3–8] and the SAPS 2 score was 58 [47–72]. Types of strokes were ischemic (46%), hemorrhagic (42%) and subarachnoid hemorrhage (SAH) (12%). Ischemic stroke patients received thrombolysis or thromboaspiration in 35/191 (18%) cases, and hemorrhagic stroke/SAH patients received neurosurgery or embolization in 35/228 (15%) cases. Reasons for endotracheal intubation were coma (72%), acute respiratory failure (12%), seizures (8%), cardiac arrest (5%) and elective procedure (3%). Sixty-five (16%) patients received a decision of WFLST in the first 48 h. One-year survival year was 23%. Variables independently associated with 1-year survival were stroke type (ischemic as reference, hemorrhagic HR 0.72 [0.56–0.93], SAH HR 0.61 [0.40–0.91], p = 0.013), specific stroke treatment (HR 1.79 [1.23–2.56], p = 0.002), reason for intubation (elective procedure as reference, seizures HR 0.57 [0.12–2.63], respiratory failure HR 0.24 [0.05–1.02], altered mental status HR 0.21 [0.05–0.88] and cardiac arrest HR 0.08 [0.02–0.37], p < 0.001), GCS score < 8 at admission (HR 0.54 [0.39–0.75], p < 0.001), cardiologic SOFA (per point, HR 0.91 [0.84–0.98], p = 0.017) and decision of WFLST within 48 h (HR 0.41 [0.30–0.56], p < 0.001) (Fig. 1). Inclusion period (1996–2002/2003–2009/2010–2016) or having a stroke unit on site was not associated with 1-year survival.

Conclusion: In acute stroke patients requiring mechanical ventilation, the reason for intubation and the opportunity to receive a specific stroke therapy are independently associated with long-term survival. These variables should be integrated in the decision process regarding initiation of MV in acute stroke patients.

Compliance with ethics regulations: Yes.

Fig. 1

Kaplan–Meier’s survival estimates of patients according to the reason for endotracheal intubation

CO-67 Association between targeted temperature management and long-term neurological prognosis: insights from two large datasets from France and north-America

Jean-Baptiste Lascarrou¹, Florence Dumas², Wulfran Bougouin³, Frankie Beganton⁴, Nadia Aissaoui⁵, Richard Chocron², Nicolas Deye⁵, Daniel Jost⁶, Stéphane Legriel⁷, Lionel Lamhaut⁸, Antoine Vieillard-Baron⁹, Eloi Marijon¹⁰, Xavier Jouven⁴, Alain Cariou⁵

¹Médecine Intensive Réanimation, Nantes, France; ²Emergency Department, Paris, France; ³Medical Surgical Intensive Care Unit, Massy, France; ⁴INSERM, Paris, France; ⁵Médecine Intensive Réanimation, Paris, France; ⁶Brigade Sapeurs Pompiers Paris, Paris, France; ⁷Medical Surgical Intensive Care Unit, Versailles, France; ⁸Medical Intensive Care Unit, Paris, France; ⁹Médecine Intensive Réanimation, Boulogne Billancourt, France; ¹⁰Cardiology Department, Paris, France

Correspondence: Jean-Baptiste Lascarrou (jeanbaptiste.lascarrou@chu-nantes.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-67

Rationale: International guidelines recommend targeted temperature management (TTM) between 32° and 36 °C for out-of-hospital cardiac arrest (CA) patients. However, it is unknown if this treatment is effective whatever the severity of the insult. We aimed to examine the association between TTM and long-term neurological outcome according to the risk evaluated at time of admission in intensive care unit (ICU) using a dedicated and validated score.

Patients and methods: We used data prospectively collected in the Sudden Death Expert Center (SDEC) registry (Great Paris area, France) between May 2011 and December 2017 and in the Resuscitation Outcome Consortium—Continuous Chest Compression (ROC-CCC) between June 2011 and May 2015. We used a modified version of the Cardiac Arrest Hospital Prognosis (mCAHP) score to assess the risk of poor outcome at ICU admission in each of 2 datasets. We finally studied the association between TTM use and long-term neurological prognosis according to mCAHP score at ICU admission divided into tertiles of severity in each of the 2 datasets.

Results: There were 2723 patients analyzed in the French dataset and 4202 in the North-American dataset. The mCAHP identified 3 categories: low risk (score < 80 points, 40% of unfavourable outcome), medium risk (80 ≤ score < 105, 80% of unfavourable outcome) and high-risk group (score > 105, 95% of unfavourable outcome). According to the mCAHP score at ICU admission, TTM was associated with a better long-term neurological prognosis in patients with low risk (aOR = 1.62 [1.15–2.30] (p = 0.006) in SDEC registry and aOR = 1.43 [1.02–2.02] (p = 0.038) in ROC-CCC) and high risk (aOR = 1.94 [1.06–3.54] (p = 0.026) in SDEC registry and OR = 2.42 [1.38–4.24] (p = 0.021) in ROC-CCC) of poor outcome, but the association was not significant in patients with a medium risk (aOR = 1.19 [0.76–1.87] (p = 0.424) in SDEC registry and OR = 1.12 [0.86–1.47] (p = 0.360) in ROC-CCC). The results were consistent in both French and north-America datasets.

Conclusion: TTM is significantly associated with better long-term neurological prognosis for patients with low risk (mCAHP < 80) and high risk (mCAHP ≥ 105) of poor outcome evaluated at ICU admission according to the CAHP score. Furthers studies are needed to individualize ICU care after cardiac arrest especially for TTM.‬‬‬‬‬‬‬.

Compliance with ethics regulations: Yes.

CO-68 Cost-utility analysis of different treatments for acute ischaemic stroke: a Belgian, micro-costing health technology assessment

Evi Mellebeek¹, Anouk Lesenne², Jef Grieten¹, Alain Wibail³, Ludovic Ernon³, Luc Stockx³, Elly Vandermeulen³, Pascal Vanelderen⁴, Joris Vundelinckx⁵, Sofie van Cauter³, Jill Grondelaers³, Elke Panis³, Dieter Mesotten⁴

¹KU Leuven/ZOL Genk, Genk, Belgium; ²Universiteit Gent/ZOL Genk, Gent, Belgium; ³ZOL Genk, Genk, Belgium; ⁴Universiteit Hasselt/ZOL Genk, Genk, Belgium; ⁵ZOL Genk, Genk, Belgium

Correspondence: Evi Mellebeek (evimellebeek503@hotmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):CO-68

Rationale: Acute ischaemic stroke is associated with a high risk of mortality, morbidity and healthcare-related costs. Over the last decades new treatments, such as thrombolysis and thrombectomy, have been introduced. Because of their further improvement, complications have been decreasing. This also led to extending indications for treatment to patients who were previously not eligible. The impact of this evolution on long-term outcome and cost-effectiveness has mainly been assessed in clinical trials and simulation studies.

Patients and methods: This single-centre retrospective study included 564 patients treated for stroke between January 2017 and February 2019. Functional outcome at 90 days was assessed by the modified Rankin Scale (mRS). Cost data were retrieved from individual invoices of 538 patients. Undiscounted total healthcare costs were calculated for the index hospital stay, capped at 90 days. Contribution of 6 cost categories to total costs was analysed. mRS at 90 days was used as a proxy for utilities to define Quality-Adjusted Life Years (QALYs). Multivariate analysis was done for gender, age, Charlson comorbidity index, pre-stroke mRS, stroke severity (NIHSS) and treatment modality (thrombectomy, thrombolysis, thrombectomy + thrombolysis, no intervention). Incremental Cost-Effectiveness Ratios (ICERs), associated to each treatment modality, were calculated.

Results: No intervention was done in 328 patients (61.0%). 93 patients (17.3%) required thrombolysis, 64 (11.9%) thrombectomy and 53 (9.9%) the combination. Total costs were mean 15,787 EUR (IQR 3534–18,177). Hospitalisation costs (mean 11,275 EUR, IQR 2066–12,428) represented 71% of total costs, compared with drug costs (341 EUR, IQR 118–330), procedural costs (3283 EUR, IQR 652–2981), honoraria (290 EUR, IQR 111–395), lab (123 EUR, IQR 22–116) and imaging (475 EUR, IQR 334–539). Mean total costs differed between treatment modalities: 12,553 (IQR 3080–12,947) EUR for no intervention, 11,820 (IQR 3265–11,197) EUR for thrombolysis, 26,103 (IQR 12,308–35,893) EUR for thrombectomy and 30,301 (IQR 12,563–35,488) EUR for the combination (p < 0.0001). Drivers for total costs were treatment modality (p < 0.017) and NIHSS-stroke severity (p < 0.0001). Utility scores were 0.65 (IQR 0.42–0.93) for no intervention, 0.70 (IQR 0.42–0.95) for thrombolysis, 0.37 (IQR 0.38–0.62) for thrombectomy and 0.66 (IQR 0.42–0.95) for the combination (p < 0.0001). Thrombolysis was dominant over no intervention, while no intervention dominated thrombectomy. The ICER of the combination was 22.187.650 EUR/QALY compared with no intervention.

Conclusion: In a 90-day time horizon, thrombectomy or its combination with thrombolysis was systematically dominated or associated with extremely high ICERs. However, their cost-effectiveness needs to be evaluated in long-term health-economic models with a time horizon of at least 5 years.

Compliance with ethics regulations: Yes.

Flash communications

F-001 A Systematic Echography During Intubation Procedure to Predict Cardiovascular Collapse (EPIC)

Vanessa Jean-Michel¹, Gwenael Prat², Pierre Bailly², James Benis³, Montaine Lefevre⁴, Pierre-Yves Egreteau⁴, Christophe Giacardi⁵, Cécile Aubron², Erwan L’her²

¹CH de Tourcoing, Tourcoing, France; ²CHRU de La Cavale Blanche, Brest, France; ³CH de Quimper, Quimper, France; ⁴CH de Morlaix, Morlaix, France; ⁵HIA Clermont Tonnerre, Brest, France

Correspondence: Vanessa Jean-Michel (vanessa.jeanmichel@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-001

Rationale: Emergency endotracheal intubation (ETI) in the Intensive Care Unit (ICU) often concerns hypoxemic patients with hemodynamic instability. A cardiovascular collapse (CVC) after ETI is a life-threatening complication. 2018 French guidelines suggested systematic fluid loading prior to ETI. Our study aimed to predict CVC after ETI, while using echocardiography, and to evaluate the impact of fluid loading.

Patients and methods: A prospective study of 70 consecutive intubations was performed from June 2017 to November 2018 in three ICUs. Patients were selected if mean blood pressure measurements ≥ 60 mmHg before ETI. CVC was defined as mean blood pressure < 60 mmHg within 15 min following ETI. Four echocardiographic examinations were performed: 1–30 min before and 2–30 min after ETI (or when a CVC occurred); 3—after passive leg raising; 4–3 h following ETI. Patients were classified as fluid responders when the left ventricular outflow tract velocity–time integral increased by at least 10% compared with baseline.

Results: 269 echocardiographic examinations were performed. CVC occurred in 32/70 procedures (46%). In CVC group, mean dose of Diprivan, used for fast sequence induction, was higher (2.8 ± 1 mg/kg vs 1.3 ± 0.7 mg/kg, p = 0.02). In the CVC group, fluid responsiveness was considered in 44% patients and left ventricular (LV) systolic dysfunction 13%. LV diastolic dysfunction did not concern any patient in the CVC group. Systolic blood pressure (SBP) < 120 mmHg was the sole independent risk factor for CVC occurrence in multivariate analysis: OR 5.9 CI 95% 1.7–21.2, p = 0.02. Fluid responsiveness independent risk factors for CVC patients was SBP < 120 mmHg (OR 8.7, CI 95% 1.1–65.1, p = 0.01) and tidal volume > 7 ml/kg IBW (OR 7.8 ICI 5% 1.02–60.3, p = 0.048).

Conclusion: CVC frequently occurs following ETI. Fluid loading does not seem to be mandatory for all patients to prevent CVC during the ETI procedure.

Compliance with ethics regulations: Yes.

F-002 Heart rate variability as an indicator of outcome in the ICU? A prospective cohort

Laetitia Bodenes, Victoire Pateau, Julien Dolou, Quang-Thang N’guyen, Erwan L’her

Service de Medecine Intensive et Réanimation, CHRU de Brest, Brest, France

Correspondence: Laetitia Bodenes (Laetitia.bodenes@chu-brest.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-002

Rationale: The autonomic nervous system is highly adaptable and allows the organism to maintain its balance when experiencing stress. Heart rate variability (HRV) is a mean to evaluate cardiac effects of autonomic nervous system activity and a relation between HRV and outcome has been proposed in various types of patients. We attempted to evaluate the best determinants of such variation in survival prediction using a physiological data-warehousing program (ReaSTOC ClinicalTrials identifier NCT 02893462).

Patients and methods: Physiological tracings were recorded at 125 Hz from the standard monitoring system (Intelliview Philips MP70) using the Synapse Software (LTSI INSERM UMR 1099), for a 2 h period, during the 24 h following ICU admission. All measurements were recorded while patients were laying in bed, with the head at 30° and without any medical intervention. Physiological data were associated with metadata collection by a dedicated research assistant. HRV was derived using Kubios HRV, in either temporal ( (SDNN), (RMSSD) and triangular index (TI)), frequency ( (LF), (HF)), non-linear domains (Poincaré plotting) and entropy.

Results: 540 consecutive patients were recorded between May 2014 and April 2019. A lower LF/HF (< 0.9) and SD2/SD1 (< 1.3) ratios on admission were associated with a higher ICU mortality. Multivariate analysis enabled to develop a mortality predictive model (BICUS) associating SpO2/FiO2 and HRV parameters (LF/HF and Shannon entropy) with an AUC = 0.73 (p < 0.0001) for a BICUS value > 1 (Fig. 1).

Conclusion: HRV measured on admission enables to predict prognosis in the ICU, independently of the admission diagnosis, treatment and MV requirements. BICUS may help predict prognosis on a real time basis, using parameters derived from standard routine monitoring.

Compliance with ethics regulations: Yes.

Fig. 1

ROC curve for Brest ICU Score (BICUS)

F-003 A mouse model of ischemic stroke in impaired glucose tolerance condition induced by chronical exposure of High Fat Diet

Coline Grisotto, David Couret

Service neuroréanimation, CHU Sud Reunion, Saint-Pierre, France

Correspondence: Coline Grisotto (grisottocoline@yahoo.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-003

Rationale: Stroke, in the context of type 2 diabetes (T2D) is associated with a worse outcome than in non-diabetic conditions, reflected by an increased ischemic volume and more intracerebral hemorrhage. An unbalanced diet is one of major risk for developing T2D. We aimed at creating a reproducible mouse model of stroke in impaired glucose tolerance condition induced by high fat diet.

Patients and methods: Adult C57BL6mice (28 male and 28 female) were fed for 2 months with either High Fat Diet (HFD, 43% lipids, 21% proteins, 35% carbohydrates) or a normal diet (ND, 8.4% lipids, 19.3% proteins, 72.4% carbohydrates). We used a model of Middle Cerebral Artery Occlusion (MCAO) by a monofilament for 90 min. Oral Glucose Tolerance Test and Insulin Tolerance Test were used for evaluating the pre-diabetic state. Mice were euthanized 20 h after reperfusion. Systemic inflammation, cerebral infarct volume and hemorrhagic transformation were determined.

Results: HFD was associated with an increased glycaemia following the Oral Glucose Tolerance Test. Plasma leptinlevels in stroke conditions were significantly higher in HFD vs ND group. The HFD group presented a significant increase of infarct volume (HFD: 51.86 ± 4.46 mm³ vs ND: 33.23 ± 4.29 mm³ p = 0.016) and hemorrhagic transformation (HFD: 2.67 ± 0.66 vs ND: 0.73 ± 0.28 p = 0.012) (Fig. 1) compared to ND group.

Discussion: In humans, one of the mechanisms leading to insulin resistance is low-grade inflammation. HFD increases gut permeability, which leads microbiota dysbiosis, thereby promoting metabolic endotoxaemia and a low-grade inflammation state. Experimental mouse models available for diabetes studies use leptin receptor deficient mice which develop T2D or destruction of pancreatic beta cells by streptozotocine injection (T1D). Studies using diet-induced insulin resistance models generally feed the mice for 12 weeks or more. However, metabolic disorders could appear earlier such as increase inflammatory markers. In our model, a short exposition to HFD (8 weeks) leads to an increase of the pro-inflammatory markers as plasma leptin and a more severe stroke status (infarct and hemorrhagic transformation).

Conclusion: Two months of HFD in adult mice altered hyperglycemia control. This metabolic disorder was associated with significantly higher leptin production, increased infarct volume and hemorrhagic complications than in normal-fed mice. This new model is particularly relevant to study stroke under pre-diabetic conditions induced by HFD.

Compliance with ethics regulations: Yes.

Fig. 1

Eight weeks of HFD increase ischemic volume and hemorrhagic transformation. (A)-Infarct volume (V) 24 h after reperfusion, all value are mean ± SEM, HFD: V = 51.66 ± 4.46 mm³, n = 14, ND: V = 33.23 ± 4.26 mm³, n = 7, *p = 0.016 (B)—Hemorrhage Transformation (HT) score 24 h after MCAO. All value are mean ± SEM HFD: HT score = 2.67 ± 0.66, n = 15, ND: HT score = 0.73 +/+ 0.28, n = 15 *p = 0.012

F-004 Does metformin act as a protective drug after a massive ischemia reperfusion induced by cardiac arrest? A survival study

Barthélémy Vallee¹, Amélie Rolle¹, Romain Rozier², Frédéric Martino¹, Jean-Ehrland Ricci², Bertrand Pons¹, Michel Carles¹

¹CHU Guadeloupe, Les Abymes, France; ²Equipe 3-INSERM U1065-Université Nice Sophia Antipolis (UNS), Nice, France

Correspondence: Barthélémy Vallee (barthelemy.vallee@chu-guadeloupe.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-004

Rationale: Cardiac arrest (CA), as massive ischemia reperfusion (IR), is an universal health issue. Medication taken at the time of the CA could have prognosis consequences. No medication has proven its benefit on CA prognosis. Pharmacological pre- or postconditioning aims to reduce IR injury but with disappointing results. Metformin (MET) is a worldwide-prescribed antidiabetic drug, and several clinical reports plead for a potential protective effect in various settings of sterile and non sterile inflammation, including IR. Our hypothesis is that MET act as a preconditioning drug against CA-induced IR.

Patients and methods: Retrospective single academic medical center survival study (French West Indies) on resuscitated CA in ICU (institutional ethical committee approval). Data were extracted from medical charts, PMSI, and laboratory DBSYNERGY™ software. Anonymized data were entered on a Excel™ and transferred to IBM^®—SPSS^® software (v 24.0.0.0) for analysis. Univariate study (Chi-2, Fisher exact tests, Student-t test, Mann–Whitney U-test if required) was followed by a multivariate model (Odd ratio OR and 95% IC: Kaplan–Meier estimator and non parametric Logrank test-Mantel Cox model). Assuming an overall in-hospital mortality for CA in ICU of 80% with an expected mortality decrease of 15% by MET, the number of patients to be included is 510.

Results: The inclusion period was 2012 to 2018, with 555 included patients (202 diabetic patients among whom 62 took MET). The D28 mortality was 67% in MET+ patients (n = 62) versus 82% in NoMET patients (n = 493), p < 0.01. Comparing alive (n = 115) versus deceased (n = 440) at D28 in univariate then multivariate analysis, asystole on the first EKG, number of iterative cardiac arrest, SOFA, No-Flow, lactate, Low-flow and SAPSII appear as independent criteria associated with D28 mortality. Conversely, MET intake showed up as a protective criterion (OR 0.477, CI95 0.237–0.957). The survival curve, including strata of Low-Flow duration at the cut-off 20 min, is reported on the Fig. 1. Among diabetic patients (n = 202), the mortality of patients in the MET+ (n = 62) was 67% versus 80% in the NoMET (n = 140), p = 0.04.

Conclusion: In diabetic patients suffering of massive IR related to resuscitated CA, a current treatment by MET is associated with a better survival. These results support a protective effect of MET and are important to initiate prospective evaluations, because of millions diabetic people around the world and the potential benefit of MET. The potential benefit in non diabetic patients and in sterile as well as non sterile inflammation should be addressed.

Compliance with ethics regulations: Yes.

Fig. 1

Survival analysis of the 555 having or not an ongoing treatment by MET at the time of the first CA, groups MET-LF < 20, MET-LF > 20, NoMET-LF < 20, NoMET-LF > 20)

F-005 Ischemic and hemorrhagic complications of patients on venoarterial-extracorporeal membrane oxygenation for cardiogenic shock post myocardial infarction

Paul Masi, Maxence Le Guyader, Charles-Edouard Luyt, Guillaume Franchineau, Loïc Le Guennec, Simon Bourcier, Nicolas Brechot, Matthieu Schmidt, Alain Combes, Guillaume Hekimian

Réanimation Médicale Pitié Salpétrière, Paris, France

Correspondence: Paul Masi (paul.masi@aphp.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-005

Rationale: Short-term ischemic and hemorrhagic complications of patients on venoarterial-extracorporeal membrane (VA-ECMO) for cardiogenic shock post myocardial infarction (MI) have not yet been reported. Our objective was to describe frequencies, outcomes and risk factors for ischemic and hemorrhagic complications.

Patients and methods: From 2015 to 2019, we conducted a retrospective observational study in a single tertiary care center on patients who developed a cardiogenic shock post MI and required VA-ECMO.

Results: During this period, 177 patients were implanted with VA-ECMO for cardiogenic shock post MI. Mean age was 56 years and 78% were male. MI was complicated by cardiac arrest in 61%. Culprit lesion was respectively the left main coronary artery for 23 patients (14.6%), left anterior descending artery for 82 patients (46.3%), circumflex artery for 14 patients (8.9%) and right coronary artery for 36 patients (22.9%). Intraaortic balloon pump or Impella were implanted respectively for 89 patients (50.9%) and eight patients (4.6%). 108 (61.0%) patients died in intensive care, 76 (43.4%) patients had a hemorrhagic complication, 46 patients (26%) had a severe hemorrhagic complication defined by death or/and massive transfusion or/and hemostatic procedure (surgery or endoscopy), 87 (49.1%) patients had thrombopenia < 100 G/L, 50 (28.2%) patients had thrombopenia < 50 G/L, 53 patients (30.1%) and 43 patients (24.6%) required antiaggregation and anticoagulation withdrawal. Nine (5.6%) patients died secondary to bleeding. VA-ECMO membranes thrombosis occurred in six (3.5%) patients. A coronary angiography check was performed for 27 (15.6%) patients, 8 (4.6%) patients had stent thrombosis.

Conclusion: Ischemic and hemorrhagic complications are frequent in patients on VA-ECMO for cardiogenic shock post MI. The main findings of our study were an high rate of severe bleeding complications and thrombopenia requiring antiaggregation withdrawal associated with a very high prevalence of stent thrombosis.

Compliance with ethics regulations: Yes.

F-006 The endoplasmic reticulum stress expression under cardiopulmonary bypass: the relationship with organ failure

Zoé Demailly¹, Thomas Clavier¹, Vincent Richard², Fabienne Tamion¹

¹Department of medical critical care-Rouen university hospital, Rouen, France; ²UMR inserm 1096 Endothelium, Valvulopathy & Heart Failure, Rouen, France

Correspondence: Zoé Demailly (lulu.demailly@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-006

Rationale: During systemic inflammation, the accumulation of misfolded proteins in the endoplasmic reticulum (ER) induces ER stress (ERS). In animal models, the inhibition of ERS reduces inflammatory response and organ failure. Cardiopulmonary bypass (CPB) induces a significant systemic inflammatory response but ERS expression has never been described in cardiac surgery patients. Our objective was to describe the variations of the Glucose Related Protein of 78 kDa (GRP78), the final effector of the ERS, during CPB.

Patients and methods: We conducted a prospective monocenter study including patients undergoing cardiac surgery with CPB. Two samples (PAXGene^® tube + EDTA tube) were taken at three times: before CPB, 2 h after the end of CPB (H2-CPB) and 24 h after (H24-CPB). After RNA isolation and reverse transcription, we performed a quantitative polymerase chain reaction to evaluate the expression of gene encoding for GRP78 and determined the plasma level of GRP78 using Enzyme-Linked Immunosorbent Assay. Our main objective was to study the variation of GRP78 between pre-CPB and H2-CPB samples. Our secondary objectives were to evaluate the association of ERS with morbi-mortality: organ failure at 24 h (catecholamines and/or invasive ventilation and/or acute renal failure), troponinemia and PaO2/FiO2 ratio (lung damage control).

Results: We included 46 patients with a median age of 70[63; 75] years and a median CPB duration of 117[92; 139] min. We showed an increase in GRP78 gene expression (p < 0.0001) and a decrease in its plasma level at H2-CPB (4540[2519; 7575] ng/mL vs. 1902[939; 3133] ng/mL; p < 0.0001). H24-CPB GRP78 plasma levels were lower than baseline in patients with persistent organ failure at 24 h but returned to baseline in patients without persistent organ failure (− 37[− 61;− 18] % vs. 0.33[− 29; 43] %; p < 0.01; Fig. 1). We found an inverse correlation between GRP78 plasma level and troponinemia at H2 (r = − 0.31; 95% CI[− 0.56; − 0.10]; p = 0.037) and a correlation between the PaO2/FiO2 ratio and GRP78 plasma level at H2 (r = 0.37; 95% CI[0.10; 0.59]; p = 0.0064).

Conclusion: We showed a significant relationship between the variation in plasma concentration of GRP78 and post-operative organ failure after CPB. Further studies are needed to better understand the molecular mechanisms of ERS in acute inflammatory organ failure in humans.

Compliance with ethics regulations: Yes.

Fig. 1

GRP78 concentration variations from baseline in patients with or without persistent organ failure at 24 h. GRP78: Glucose Related Protein 78 kDa; OF: organ failure; **p < 0.01

F-007 Acute mesenteric ischemia in patients under mechanical support by veno-arterial ECMO

Marie Renaudier¹, Quentin De Roux¹, Wulfran Bougouin², Johanna Boccara¹, Baptiste Dubost¹, Antonio Fiore¹, Aurélien Amiot¹, Olivier Langeron¹, Nicolas Mongardon¹

¹CHU Henri Mondor, Créteil, France; ²Hôpital Privé Jacques Cartier, Massy, France

Correspondence: Marie Renaudier (Mrenaudier@aol.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-007

Rationale: Acute mesenteric ischemia is a dramatic complication in critically ill patients. However, acute mesenteric ischemia has never been evaluated in patients under veno-arterial ExtraCorporeal Membrane Oxygenation (VA-ECMO). This study was designed to determine the incidence of acute mesenteric ischemia in this population and to evaluate its risk factors, therapeutic modalities and outcome.

Patients and methods: In a retrospective monocentric study (01/2013–01/2017) conducted in cardio-vascular surgical intensive care unit (ICU) in Henri Mondor teaching hospital, all consecutive adult patients who underwent peripheral VA-ECMO were included, with exclusion of those dying in the first 24 h. Diagnosis of acute mesenteric ischemia was performed using digestive endoscopy, abdominal CT-scan or fist-line laparotomy. Significative results in the univariate analysis were analyzed in a multivariate analysis using logistic regression.

Results: 150 VA-ECMO were implanted. Median age was 58 (48–69) years and median SAPS II 54 (38–70). VA-ECMO was implanted after a cardiotomy in 43% of the cases and for a medical reason in 57% of the cases including 26% of refractory cardiac arrest. Patients characteristics are reported in the table. Acute mesenteric ischemia was suspected in 38 patients, with a delay of 4 (2–7) days after ECMO implantation. Digestive endoscopy was performed in 33 patients, CT-scan in five patients and first-line laparotomy in three patients. Acute mesenteric ischemia was confirmed in 14 patients, i.e. an incidence of 9%. Laparotomy was performed in six of the 14 patients, two having a stage I colitis ischemitis with stable conditions and 6 being considered too severe to undergo futile surgery. Overall mortality was 56%. All the patients with acute mesenteric ischemia died in the ICU. Independent risk factors of developing acute mesenteric ischemia were renal replacement therapy (OR 4.5 (95% IC 1.3–15.7, p = 0.02)) and onset of a second shock state within the first 5 days of ICU stay (OR 7.8 (95% IC 1.5–41.3, p = 0.02)). Conversely, early enteral nutrition was negatively associated with acute mesenteric ischemia (OR 0.15 (95% IC 0.03–0.69), p 0.02).

Conclusion: Acute mesenteric ischemia is a relatively frequent condition among patients under VA-ECMO for cardiogenic shock. Its extremely poor prognosis requires low threshold of suspicion.

Compliance with ethics regulations: Yes.

Table 1 Patients characteristics and initial management

F-008 Comparison of pleural and esophageal pressure in supine and prone position in a porcine model of acute respiratory distress syndrome

Nicolas Terzi¹, Sam Bayat¹, Norbert Noury², Jean-Pierre Giliberto³, Sylvie Roulet³, Emanuele Turbil⁴, Walide Habre³, Claude Guérin²

¹CHUGA, Grenoble, France; ²Université de Lyon, Lyon, France; ³Université de Genève, Genève, Switzerland; ⁴Université de Sassarie, Sassarie, Italy

Correspondence: Nicolas Terzi (nterzi@chu-grenoble.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-008

Rationale: Measurement of esophageal pressure (Pes) is the single method currently available to estimate pleural pressure (Ppl) in ICU patients (1). It allows the computation of trans-pulmonary pressure (2) and can be used to set positive end-expiratory pressure (PEEP) (3.4). Prone position (PP) can reduce mortality in patients with acute respiratory distress syndrome (ARDS), but PEEP selection in PP is controversial. In human ARDS end-expiratory Pes at zero flow (PEEPt,es) was not different between supine (SP) and PP at same PEEP (5). As no study measured Ppl in SP and PP in ARDS we aimed at comparing PEEPt,es and end-expiratory Ppl at zero flow (PEEPt,Ppl) in this condition. Our hypothesis was that PEEPt,es was close to dorsal PEEPt,Ppl (PEEPt,Ppldorsal) in SP and to ventral PEEPt,Ppl (PEEPt,Pplventral) in PP.

Patients and methods: In eight female pigs of 40 kgs intubated, sedated, paralyzed and mechanically ventilated, ARDS was induced by repeated saline lavage until PaO2/FIO2 < 100 mmHg under FIO2 1 and PEEP 5cmH2O. Pes was measured by Nutrivent catheter. Ppl was measured by custom-made pouch sensors inserted surgically into the right anterior and posterior sixth intercostal space. Ppl sensors were filled with air. After ARDS induction animals were randomly assigned to SP or PP. In each position, a recruitment manoeuver was performed and PEEP decreased from 20 to 5 cmH2O by steps of 5cmH2O lasting 10 min each, then the animals were crossed over into the alternate position where the same procedure was done. At the end of each step nonstressed volume and correct position (Baydur maneuver) were determined for Pes and Ppl sensors, then a 3-s end-expiratory occlusion was performed and Pes and Ppl recorded. Linear mixed model was used to compare the value of Pes and Ppl at each PEEP and position.

Results: Box-and-whisker plots of Pes and Ppl in SP and PP are shown in Fig. 1. There is marked dorsal-to-ventral gradient in Ppl at each PEEP in SP, which is reverted in PP at PEEP 10 and 15 only. There was no interaction between pressures and PEEP or position. With increasing PEEP Pes increased significantly from PEEP 5 in SP and PP. PEEPt,PplVentral was significantly lower than PEEPt,es in SP but not in PP.

Conclusion: PEEPtes follows PEEPt,Ppldorsal in SP and PEEPt,Pplventral in PP in this ARDS porcine model. References: (1) Akioumaniaki et al. AJRCCM 2014; (2) Yoshida et al. AJRCCM 2018; (3) Talmor et al. NEJM 2008; (4) Beitler et al. JAMA 2019; (5) Mezidi et al. AOIC 2018.

Compliance with ethics regulations: Yes.

Fig. 1

Box-and-whisker plots of esophageal pressure and pleural pressure in supine position and in prone position

F-009 A systematic bench assessment of automatic tube compensation provided by intensive care unit ventilators

Nicolas Terzi¹, Louis-Marie Galerneau¹, Zakaria Riad², Emanuele Turbil³, Carole Schwebel¹, Claude Guérin², Bruno Louis⁴

¹Médecine Intensive Réanimation, Grenoble, France; ²Médecine Intensive Réanimation, Lyon, France; ³Médecine Intensive Réanimation, Sassari, Italy; ⁴INSERM UMR 955 Eq13, Créteil, France

Correspondence: Nicolas Terzi (nterzi@chu-grenoble.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-009

Rationale: Automatic tube compensation (ATC) is an option available in intensive care unit (ICU) ventilators to compensate for endotracheal tube (ETT) resistance. To achieve this ICU ventilator delivers a certain amount of pressure/flow that compensates for the resistive pressure drop across ETT. It requires notifying size of ETT and percent of compensation. We reasoned that if ATC works properly tidal volume (VT) should be the same without ATC and no ETT as with ATC and ETT. We tested the performance of ICU ventilators on a bench, expecting furthermore differences between them.

Patients and methods: Seven ICU ventilators (Evita XL and V500 infinity (Draeger), C6 and S1 (Nihon-Khoden), Elisa 800 (Lowenstein), 980 (Medtronic), Carescape 860 (GE)) were set in pressure support 0 cmH2O, PEEP 4 cmH2O, FIO2 21% and equipped with the same double limb ventilator circuit (Intersurgical) without any humidification device. ASL 5000 bench model was set with 3 inspiratory/expiratory resistance (R) and compliance (C) combinations: R13/12-C54, R12/14-C39 and R22/18-C59 mimicking normal, ARDS and COPD conditions, respectively (1). Inspiratory effort generated by ASL 5000 consisted of 30 consecutive breaths obtained from the esophageal pressure in a real patient at the time of a spontaneous breathing trial. For each ICU ventilator and RC combination, two steps were performed: in the first, ATC was not activated and ventilator attached to ASL 5000 without ETT (ATC-ETT-); in the second, ATC was set on at 100% compensation for an ETT 8 mm ID and such an ETT (Shiley Hi contour, Covidien) joined ICU ventilator to ASL 5000 (ATC+ ETT+). The null hypothesis is that VTATC+ ETT+ minus VTATC-ETT—is 0. Primary end point was the breath by breath paired difference betwen ATC+ ETT+ and ATC-ETT-. It was tested to zero for each ventilator in each RC condition.

Results: Median VT was 213 ml. Table 1 displays mean (± SD) difference in VT (ml) between ATC+ ETT+ and ATC-ETT-: a negative value means that ATC under delivers and a positive value that ATC over delivers VT for a given patient’s inspiratory effort and RC. In four ventilators (C6, S1, Elisa 800 and 980) ATC almost systematically under delivered VT. In several instances under compensation was greater than 10% median VT. By contrast ATC performed better with the other three ventilators (Evita XL, V500 and Carescape 860).

Conclusion: ATC tended to under deliver VT. Furthermore, there were marked differences between ICU ventilators the clinician should be aware of when using the ATC option.

Compliance with ethics regulations: NA.

Table 1 Change in tidal volume between no (Automatic tube compensation) ATC and ATC plus tube

F-010 Impact of the mechanical power and its components on mortality in patients with ARDS: a post hoc analysis of a controlled randomized study

Radj Cally¹, Claude Guérin², Arnaud Gacouin³, Gilles Perrin¹, Anderson Loundou¹, Samir Jaber⁴, Jean-Michel Arnal⁵, Dider Perez⁶, Jean-Marie Seghboyan⁷, Jean-Michel Constantin⁸, Jean-Yves Lefrant⁹, Gwenael Prat¹⁰, Antoine Roch¹, Laurent Papazian¹, Jean-Marie Forel¹

¹AP-HM, Marseille, France; ²CHU de Lyon, Lyon, France; ³CHU de Rennes, Rennes, France; ⁴CHU de Montpellier, Montpellier, France; ⁵Centre Hospitalier de Toulon, Toulon, France; ⁶Centre Hospitalier de Lons-le-Saunier, Lons-Le-Saunier, France; ⁷Hôpital Européen, Marseille, France; ⁸CHU de Clermont-Ferrand, Clermont-Ferrand, France; ⁹CHU de Nîmes, Nimes, France; ¹⁰CHU de Brest, Brest, France

Correspondence: Radj Cally (radj.cally@yahoo.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-010

Rationale: During the last decades, identification of factors associated with ventilation-induced lung injury has led to improved survival in patients with ARDS. The mechanical power of ventilation is the total energy transmitted from the ventilator to the respiratory system per unit of time and comprises three different components: elastic related to PEEP, elastic related to tidal volume and resistive. This integrative variable has been recently proposed as an useful predictor of ventilation-induced lung injury and death among ventilated patients. Our goal was to determine the respective impact of the total mechanical power and its three components on the outcome of patients with ARDS.

Patients and methods: We performed a post hoc analysis of a randomized, controlled study of patients with ARDS with a PaO2/FiO2 ratio < 150. The mechanical power at inclusion and averaged on the first 2 days after inclusion (total and its three different components) was computed according to the following equation: PowerRS (J/min) = 0.098 Respiratory Rate Tidal Volume [PEEP (1) + ½ Driving Pressure (2) +  (Peak Pressure—Plateau Pressure) (3)], where the (1), (2) and (3) parts correspond respectively to the elastic related to PEEP, elastic related to tidal volume and resistive components. The association between each of these four types of mechanical power evaluated during the first 2 days after inclusion and mortality at D90 was assessed one after the other through multiple logistic regression, allowing control for potential confounding variables at inclusion (age, IGS score without age, group of randomization, PaO2/FiO2, arterial pH).

Results: Data from 339 patients were analyzed, among which 115 (33.9%) died before D90. There was no difference concerning the mechanical power at inclusion between survivors and non survivors (either total or its three components). Among the four different types of mechanical power tested during the first 2 days after inclusion, the elastic component related to tidal volume was the only one that was independently associated with mortality at D90 (OR 1.030; 95% CI 1.003–1.058; p = 0.03) (Figure).

Conclusion: Our study shows that only the elastic component of the mechanical power related to tidal volume independently predicted mortality at D90 among patients with ARDS, whereas the total mechanical power, its elastic component related to PEEP and its resistive component did not. Further studies are needed to better define how the mechanical power of ventilation could be useful to synthetize the risk of ventilation-induced lung injury.

Compliance with ethics regulations: Yes.

Fig. 1

Probability of death at D90 as a factor of mean value (on D1-D2) of the elastic component related to tidal volume of the mechanical power. Others potential confounding variables included in the logistic regression model were age, IGS score without age, group of randomization, PaO₂/FiO₂, arterial pH (all at inclusion)

F-011 A physiological systematic review and meta-analysis on positive end expiratory pressure-induced lung recruitment in patients with acute respiratory distress syndrome

Emanuele Turbil¹, Louis-Marie Galerneau², Antonia Koutsoukou³, Jean Jacques Rouby⁴, Jean Dellamonica⁵, Carole Schwebel², Nicolas Terzi², Claude Guérin⁶

¹University of Sassari, Sassari, Italy; ²C.H.U de Grenoble, La Tronche, France; ³University of Athens Medical School, Athens, GREECE; ⁴La Pitié Salpétrière Hospital, Paris, France; ⁵CHU de Nice, Nice, France; ⁶Hôpital de la croix rousse, Lyon, France

Correspondence: Emanuele Turbil (emanuele.turbil@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-0011.

Rationale: Recruited lung volume (Vrec) elicited by positive end-expiratory pressure (PEEP) can be measured using pressure–volume (P–V) curve. Our purpose was to perform a meta-analysis of Vrec measured by P–V curve in patients with Acute Respiratory Distress Syndrome (ARDS). Primary aim was determination of the prevalence of recruiters (R). Secondary aim was comparison of physiologic data and mortality in R and non-recruiters (NR).

Patients and methods: We conducted a PubMed search using the key words: lung recruitment, alveolar recruitment, P–V curve, ARDS, PEEP, humans and adult. All articles were screened by two reviewers. Articles concerning animals, children, case reports and reviews were excluded. Disagreements were resolved by discussion. The same reviewers extracted data in a case record form, which included mortality at ICU discharge, baseline value of anthropometric variables, ARDS cause, physiologic characteristics of patients, ventilator settings, and respiratory mechanics. Vrec ≥ 150 ml was used as the threshold to define R. We first performed a meta-analysis on grouped data, which were pooled applying a random effects model. Then, authors were contacted to obtain individual data. We performed an individual-databased meta-analysis and compared R and NR and survivors and nonsurvivors. Logistic regression (LR) analysis was done to assess the role of Vrec on mortality after adjustment for confounders. These latter were identified by univariate LR analysis between survivors and nonsurvivors. Results were reported as mean, percentage or relative risk (RR), with 95% confidence intervals (CI).

Results: From a total of 817 studies, we retrieved 18 articles that included 361 patients on whom the grouped-data meta-analysis. From data on 283 patients, in part coming from original papers’ authors, the individual-data meta-analysis was performed. The prevalence of R was 74% (95% IC 64–84%) in grouped-data and 66% (60–71%) in individual-data analyses. In both grouped- and individual-data analyses there were no significant differences between R and NR for baseline age, male proportion, PEEP, PaO₂/FIO₂, ARDS cause, and days in ARDS before the start of the investigation. In grouped-data analysis compliance was significantly higher (mean difference 13.80 (0.14–27.52) ml/cmH₂O) in R, but this result was not confirmed in individual-data analysis. Finally, RR of mortality in grouped-data was 1.20 (0.88–1.63), which was in accordance with individual-data analysis (p-value = 1). After adjusting for confounding variables (PEEP, PaO₂/FIO₂ and Plateau pressure), Vrec was not an independent risk factor of ICU mortality.

Conclusion: Most ARDS patients exhibited lung recruitment after increase in PEEP with no correlation with ICU mortality.

Compliance with ethics regulations: Yes.

F-012 Early effects of mechanical ventilation on diaphragm function and its influence on weaning

Ahlem Trifi, Farah Ben Lamine, Cyrine Abdennebi, Foued Daly, Yosr Touil, Sami Abdellatif, Salah Ben Lakhal

Medical ICU, la Rabta hospital, Faculty of Medicine of Tunis, Tunis, Tunisia

Correspondence: Ahlem Trifi (trifiahlem2@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-012

Rationale: To examine the effect of early-stage mechanical ventilation (MV) on diaphragmatic contractility. In the 2nd step, if a diaphragmatic dysfunction was detected, we assessed its influence on the weaning from ventilator.

Patients and methods: We measured prospectively the ultrasound-diaphragmatic thickening fraction (DTF) between 2 groups: a study group versus a control group (n = 30 for each). The study group included all adult patients receiving MV, in whom, the DTF was measured within a minimum of 48 h and a maximum of 5 days of MV. For the control group, were enrolled after their approval for participation, adult volunteers in spontaneous ventilation (SV). Patients with factors affecting the diaphragmatic contractility (neuromuscular disease, severe obesity, and neuromuscular blockers…) were excluded. The ultrasound measurements were obtained at the zone of apposition of the right hemithorax. Teleinspiratory and telexpiratory diameters (tid/ted) were taken on the 3 medio-axillary lines: posterior, median and anterior. The DTF was calculated as following: DTF =  (tid-ted/ted) x 100. At the 1st step, the DTFs were compared and at the 2nd step: the relationship between DTF and weaning was analysed.

Results: Our 2 groups were comparable in corpulence and co morbidities. The SV group was younger (35 vs. 47 years, p < 0.05) with a predominant female composition. The diaphragmatic exploration concluded that in the MV group, the mean tid tended to be higher but without significant difference (29.1 + 7 versus 26.1 + 5 mm, p = 0.09), the mean ted was significantly higher (20.9 + 6 versus 17.6 + 3.2 mm, p = 0.01) and DTF was significantly lower (39.9 + 12.5% versus 49 + 20.5%, p = 0.043). The ventilation mode had no effect on DTF (40.2 + 13% for control volume vs. 38.6 + 9% for PSV mode, p = 0.8). Fourteen among 30 ventilated patients had a successful weaning with a mean duration of 6 days. A negative correlation was found close to significance between DTF and weaning duration (Rho = − 0.464 and p = 0.08). A DTF value > 33% was associated with weaning success (OR = 2, 95% CI = [1.07–3.7] and p = 0.058) with sensitivity = 85.7%, Specificity = 50%, PPV = 60% and NPV = 80%.

Conclusion: The diaphragmatic contractile function was altered from the first days of MV. Weaning duration seemed to be negatively correlated with DTF, and a DTF at the first 5 days of MV greater than 33% was predictive of weaning success.

Compliance with ethics regulations: Yes.

F-013 Shear wave elastography for assessing diaphragm activity in mechanically ventilated patients

Quentin Fossé¹, Thomas Poulard², Jean-Yves Hogrel², Jean-Luc Gennisson³, Thomas Similowski¹, Alexandre Demoule ¹, Marie Cécile Niérat⁴, Damien Bachasson², Martin Dres¹

¹Service de Pneumologie, Médecine Intensive Réanimation, Pitié Salpétrière, GH Universitaire APHP-Sorbonne Université, Paris, France; ²Institut de Myologie, Paris, France; ³IR4M, Orsay, France; ⁴INSERM UMRS1158 Neurophysiologie respiratoire expérimentale et clinique, Paris, France

Correspondence: Quentin Fossé (quentin.fosse@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-013

Rationale: Mechanical ventilation is a life-saving treatment that is however associated with lung injury and/or diaphragm dysfunction. The optimal ventilator settings to provide lung protective ventilation while maintaining safe diaphragm activity are difficult to determine. A noninvasive and bedside evaluation of the diaphragm activity could be helpful in this context. The present study investigated whether changes in diaphragm shear modulus (i.e. stiffness, ΔSMdi) assessed by ultrasound shear wave elastography (SWE) may be used as a surrogate of changes in transdiaphragmatic pressure (ΔPdi) in mechanically ventilated patients.

Patients and methods: Patients had to be ventilated for at least 24 h without contraindications for the placement of an œso-gastric catheter. Pdi was monitored continuously and SMdi was measured at the zone of apposition of the right hemi-diaphragm, at 2 Hz sampling rate. Measurements were performed twice under initial ventilator settings and at the end of a weaning trial. Pearson correlation coefficients (r) were computed to determine within-individual correlations between Pdi and SMdi and changes in Pdi and in SMdi occurring between initial ventilator settings and the end of the SBT were compared by a paired test.

Results: Twenty-five patients were enrolled and 8 displayed a significant correlation between ΔSMdi and ΔPdi (mean r = 0.73, range = 0.62–0.88, all p < 0.05) (Fig. 1A). Compared to their counterparts, patients with significant within correlations had a lower respiratory rate (16.8 ± 4.7 vs 23.9 ± 6.2breath/min. respectively; p < 0.01) and a significant increase in ΔSMdi (7.3 ± 5.5 kPa vs 13.4 ± 9.0 kPa. p < 0.01) between initial ventilator settings and the SBT. Patients without ΔSMdi-ΔPdi correlation only displayed an increase in ΔPdi (7.9 ± 5.9 vs 14.9 ± 7.9 cmH₂O, p < 0.01) at the end of the SBT with no concomitant significant increase in ΔSMdi (7.6 ± 3.8 kPa vs 7.9 ± 5.3 kPa, p > 0.05). (Fig. 1B).

Conclusion: SMdi obtained by SWE appears as a promising technique to assess diaphragm activity in mechanically ventilated patients but technological improvements are necessary to increase SWE sampling rate before enabling its generalization in the ICU.

Compliance with ethics regulations: Yes.

Fig. 1

a Relationship between changes and diaphragmatic shear modulus. b ΔPdi and ΔSMdi under initial ventilator settings and at the end of the SBT

F-014 Inspiratory effort can be measured during spontaneous breathing at the bedside: illustration of feasibility

Davy Cabrio, Filippo Bongiovanni, Philippe Eckert, Lise Piquilloud

CHUV, Centre Hospitalier Universitaire Vaudois, Lausanne, Switzerland

Correspondence: Davy Cabrio (cabriodavy@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-014

Rationale: End-inspiratory (EIP) and end-expiratory (EEP) pauses are commonly used during volume assist control ventilation to assess plateau pressure and total positive end-expiratory pressure (PEEPtot). They can also be used during assisted ventilation (AV) for muscle pressure assessment. It requires ventilators able to perform EIP during AV. Plateau pressure (Pplat) usually increases in AV during EIP due to “hidden” inspiratory effort. Pressure muscular index (PMI) is equal to Pplat minus the sum of PEEPtot (measured during an EEP) and set pressure support (PS); it theoretically reflects patient’s effort without esophageal pressure (Pes) monitoring. Pes is the gold standard method to assess inspiratory muscle pressure (Pmus, difference of Pes drop at neural end-inspiration and correction factor for chest wall elastance and tidal volume). We aimed to illustrate the feasibility of measuring PMI using a standard ICU ventilator at the bedside and study the correlation between Pmus and PMI.

Patients and methods: Measurements were recorded in 4 ICU patients. Pes was measured using an nasogastric probe (equipped with an esophageal balloon) inserted for advanced monitoring (severe acute respiratory distress syndrome—ARDS) or for a study protocol (difficult weaning after COPD exacerbation). Recorded EIP, EEP and Pes were used for post hoc analyses. Results reported as ranges and median [IQR]. Correlation between Pmus and PMI tested with Spearman correlation test.

Results: 25 out of 28 EIP and EEP duos could be analyzed (2-esophageal spasm/1-calibration error). Ventilator mode was pressure support ventilation (PS 0–12 cmH2O). Pplat = 17.9 [15.1–21.4] cmH2O, PEEPtot = 5.5 [5–8.3] cmH2O, Pmus = 3.5 [1.7–9.5] cmH2O, PMI = 4.5 [2.4–7.8]. For all recordings, Spearman r coefficient between Pmus and PMI was 0.438 (p = 0.028).

Conclusion: Muscular effort can be assessed in AV using EIP and EEP using ICU ventilators. However, recordings can be influenced by expiratory muscles contraction. Patient’s ability to follow directions during the maneuvers is an important factor to obtain reliable values. There seem to be a correlation in our small sample between muscular pressure assessed without and with Pes.

Compliance with ethics regulations: Yes.

F-015 Innate T cells during severe pneumonia and Acute Respiratory Distress Syndrome

Youenn Jouan¹, Chloé Boisseau², Antoine Guillon¹, Yonatan Perez¹, Pierre-François Dequin¹, Mustapha Si-Tahar², Christophe Paget²

¹Service de Médecine Intensive Réanimation, Tours, France; ²INSERM U1100, Tours, France

Correspondence: Youenn Jouan (youenn.jouan@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-015

Rationale: Severe pneumonia can culminate in acute respiratory distress syndrome (ARDS). An uncontrolled inflammatory response is a key feature favoring transition towards ARDS. However, the underlying mechanisms remain poorly understood. In this context, the contribution of “innate T cells” (ITC) -a family of non-peptide reactive T cells comprising NKT cells, Mucosal Associated Invariant T (MAIT) cells and γδT cells—has never been explored. ITC have emerged as key players in orchestration of the host response during infections and inflammation processes. For these reasons, these cells are already seen as potential therapeutic targets in other medical fields (especially oncology). Here, we hypothesized that a tight regulation of their functions could be paramount to control the inflammatory response and to prevent ARDS development.

Patients and methods: To explore this, we combined a murine model of influenza A virus (IAV) infection mimicking ARDS symptoms and a clinical study recruiting patients admitted in ICU for severe pneumonia. Using flow-cytometry approaches, we investigated (1) the abundance and dynamics of ITC in various compartments, (2) their pattern of activation/regulation markers (respectively CD69 and PD-1) and (3) their cytokine production.

Results: During experimental IAV pneumonia, ITC were transiently recruited into the airways. Unlike γδT and NKT, MAIT cells phenotype was largely changed, displaying a progressive CD69 overexpression and increased IL-17A production. During the resolution phase, up to 90% of pulmonary MAITs expressed PD-1 (versus < 10% in controls), which can suggest emergence of regulatory functions. Last, using gene-targeted mice, we suggested that MAIT cells confer a protective effect during pneumonia. In the ongoing clinical study, the proportion of circulating MAIT cells in patients was markedly decreased compared to controls (1.0 ± 1.0% versus 5.7 ± 2.8% of T cells), but not for NKT or γδT cells. Notably, some patients with severe ARDS presented detectable levels of MAITs in their respiratory fluids. In addition, circulating MAIT cells in patients overexpressed CD69 and PD-1 (56.5% and 55% respectively), but with a reduced proportion able to produce IL-17 and IFNγ, compared to healthy controls. Lastly, proportion of activated (CD69 +) MAIT cells significantly decreased with clinical improvement.

Conclusion: This translational approach combining in vivo animal experiments and clinical samples with ex vivo experiments indicates a preferential modulation in MAIT cells functions during severe pneumonia. These data justify an in-depth analysis of MAIT cells activation mechanisms and functions in this context, in order to further explore a potential use as a disease-progression marker and -in a long term perspective—as a potential therapeutic target.

Compliance with ethics regulations: Yes.

Fig. 1

Representative flow-cytometry dot-plots of MAIT cells labelling using fluorophore-conjugated MR1 tetramers loaded with 5-OP-RU from lungs of an infected mouse (A) and blood sample of a patient with pneumonia (B). C: frequency of MAIT cells, proportion of CD69 and PD-1 + MAIT cells in bronchoalveolar lavage during experimental murine pneumonia. D: blood frequency of MAIT cells in patients with pneumonia compared with healthy controls (as % of total T cells)

F-016 A dominant circulating PD-1/2B4 CD8+ T-cells pattern at day 1 correlates with mortality but not with secondary infections in septic patients admitted to the ICU

Damien Guinault¹, Marie-Laure Nicolau-Travers², Stein Silva³, Arnaud Del Bello¹, Olivier Cointault¹, Edith Hourcastagnou³, David Rousset³, Laurence Lavayssiere¹, Marie-Béatrice Nogier¹, Arnaud Mari³, Nassim Kamar¹, François Vergez², Stanislas Faguer¹

¹Département de Néphrologie et Transplantation d’Organes, CHU de Toulouse, Toulouse, France; ²Laboratoire d’Hématologie-Immunologie, CHU de Toulouse, Toulouse, France; ³Département d’Anesthésie et Réanimation-Réanimation Médicale, CHU de Toulouse, Toulouse, France

Correspondence: Damien Guinault (guinault.d@chu-toulouse.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-016

Rationale: Immune paralysis following hyperinflammatory states increases the risk of secondary infections and death. Reversing T-cells exhaustion using recombinant IL7 or immune checkpoints inhibitors may improve the prognosis of patients with sepsis admitted to the ICU. However, there is an unmet need to better characterize the state of T-cells exhaustion in these patients, its reproducibility and its correlation with the outcomes before implementing immunotherapy in the therapeutic armamentarium against sepsis.

Patients and methods: Prospective observational cohort study performed in two tertiary-care ICUs in a university hospital. Peripheral blood mononuclear cells were collected at day 1 in 80 adult patients with sepsis admitted to the ICU. The level of CD4+ and CD8+ T-cells exhaustion was quantified using multi-color flux cytometry targeting the following exhaustion markers: PD-1, 2B4 and CD160. CD4+ regulatory T-cells (CD3+ CD4+ CD25hi CD127Lo cells) were also assessed.

Results: The 80 patients included in the study could be split in five clusters according to their dominant pattern of exhaustion markers on CD8+ T-cell (i.e. no markers, PD-1+, 2B4+, 2B4+ CD160+ and 2B4+ PD-1+) and independently of their underlying morbidities. No patients harbored a fully exhausted triple-positive pattern. By multivariate analysis, SAPS2 gravity score at day 1 (p = 0.007), a dominant 2B4 and/or PD-1 CD8+ pattern (p = 0.04) and lung sepsis (p = 0.02) where associated with the risk of death at day 28, whereas hemoglobin level was associated with survival (p = 0.04). No CD4+ or CD8+ exhaustion pattern independently predicted the risk of secondary infections. Neither the level of CD4+ regulatory T-cells nor the dominant CD4+ exhaustion pattern was associated with the outcomes.

Conclusion: CD8+ T-cells of individuals with sepsis may express one or two exhaustion markers at the admission to the ICU but fully exhausted phenotype is rare, suggesting an intermediate state of T-cell activation or differentiation. A dominant 2B4 and/or PD-1 CD8+ pattern at day 1 is independently associated with mortality.

Compliance with ethics regulations: Yes.

F-017 Comparison of one-year outcome of severe community-acquired bacterial or viral pneumonia and pneumonia of unidentified etiology

Frédéric Sangla¹, Pierre-Emmanuel Marti², David Legouis², Amélie Brebion³, Pierre Saint-Sardos⁴, Alexandre Lautrette⁵, Mireille Adda⁵, Bruno Pereira⁶, Bertrand Souweine⁵

¹CHU de Clermont-Ferrand, Clermont-Ferrand, France; ²Service de Soins Intensifs adultes, Hopitaux Universitaires de Genève, Genève, Switzerland; ³Laboratoire de Virologie, CHU Clermont-Ferrand, CNRS, Université Clermont Auvergne, Clermont-Ferrand, France; ⁴Laboratoire de Bactériologie, CHU Clermont-Ferrand, CNRS, INSERM, Université Clermont Auvergne, Clermont-Ferrand, France; ⁵Service de Médecine Intensive et réanimation, CHU Clermont-Ferrand, CNRS, Université Clermont Auvergne, Clermont-Ferrand, France; ⁶Département de Biostatistique, CHU Clermont-Ferrand, Clermont-Ferrand, France

Correspondence: Frédéric Sangla (fred.sangla@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-017

Rationale: There is growing use of multiplex polymerase chain reaction (mPCR) for respiratory virus testing in patients with community-acquired pneumonia (CAP). Data on one-year outcomes in patients with severe CAP of bacterial, viral and unidentified etiology are scarce.

Patients and methods: A single-center retrospective study was performed in 123 intensive care unit (ICU) patients with known one-year survival status who had undergone respiratory virus testing for CAP by mPCR. One year after ICU admission, mortality rates and functional status were compared in patients with CAP of bacterial, viral or unidentified etiology.

Results: There were 19 (15.4%) patients in the bacterial group, 37 (30.1%) in the viral group and 67 (54.5%) with unidentified etiology. One-year mortality was 57.9% (n = 11/19), 27% (n = 10/37) and 28.4% (n = 19/67), respectively (p = 0.046). In multivariate analysis, one-year mortality was higher in the bacterial group than in the viral group (HR 2.92, 95% IC 1.71–7.28, p = 0.02), had a trend to be higher in the bacterial group compared to the unidentified etiology group (HR 2.07, 95% IC 0.96–4.45, p = 0.06) and was not different between the viral and unidentified etiology groups (HR 0.71, 95% IC 0.30–1.65, p = 0.43). Severe dyspnea (mMRC score = 4 or death), major adverse respiratory events (new homecare ventilatory support or death) and severe autonomy deficiencies (ADL Katz score ≤ 2 or death) were observed in 52/104 (50.0%), 65/104 (62.5%) and 47/104 (45.2%) patients, respectively, with no difference between groups.

Conclusion: CAP of bacterial origin was associated with a poorer prognosis than viral or unidentified etiology. Impaired functional status was observed in a substantial proportion at one-year, irrespective of the causative microorganisms involved.

Compliance with ethics regulations: Yes.

F-018 Interest of UNYVERO multiplex PCR (CURETIS) for BAL rapid microbiologic and antibiotic susceptibility documentations in immunocompromised patients under antibiotic therapy

Jean-Luc Baudel¹, Jacques Tankovic², Redouane Dahoumane², Salah Gallah², Laurent Benzerara², Jean-Remy Lavillegrand¹, Razach Abdallah¹, Geoffroy Hariri¹, Naike Bige¹, Hafid Ait-Oufella¹, Nicolas Veziris², Eric Maury¹, Bertrand Guidet¹

¹Service de Réanimation Médicale, Hôpital Saint-Antoine, Paris, France; ²Service de Bactériologie, Hôpital Saint-Antoine, Paris, France

Correspondence: Jean-Luc Baudel (jean-luc.baudel@aphp.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-018

Rationale: Our aim was to evaluate the interest of the Unyvero rapid (4.5 h) multiplex PCR assay (performed on bronchoalveolar lavage [BAL] samples) for the management of immunocompromised patients already treated with antibiotics and diagnosed with pneumonia (according to clinical and radiological findings). We thus performed an observational study that compared the results (and the length of time to obtain them) of routine microbiological evaluation and Unyvero assay.

Patients and methods: From July 2018 to January 2019 and from April to August 2019, we examined BAL samples from immunocompromised patients (coming from hematology, oncology, hepatology, gastroenterology, internal medicine, and neurology units) diagnosed with pneumonia (based on clinical and radiological findings), and already receiving antibiotic treatment. The following data were collected: age, gender, SAPS2 score, lung CT scan (92%) or X-ray (8%) results, duration and content of prior antibiotic therapy, direct examination, culture, antibiogram and Unyvero results, secondary confirmation of pneumonia or not, possible changes in antibiotic therapy that could have been made after obtention of Unyvero results. Informed consent was obtained from all patients.

Results: 40 BAL samples were analyzed in 38 immunocompromised patients (m/f ratio 2.17, SAPS2 51.5 ± 6.8) mostly with hematologic (76%) or oncologic (13%) diseases. The patients received either corticosteroids (47%), or chemotherapy (37%), or immunotherapy (8%). 40% of the patients were under mechanical ventilation, 15% under Optiflow. 32% presented a shock, 22% had aplasia or neutropenia, 24% were allografted, 16% were autografted. The duration of prior antibiotic therapy at the time of BAL were 9.3 ± 5.6 days. Direct examination was positive in 22.5% of the cases, culture (both above and under the classical threshold of 104 CFU/ml) in 60%, Unyvero in 47.5%. A retrospective analysis of all the cases confirmed the initial diagnosis of pneumonia in only 50% of the cases. Compared to culture, the sensitivity of Unyvero was 81%, its specificity 94%. Unyvero could permit to rapidly deescalate antibiotic therapy in 42% of the cases and to rapidly stop it in 32%.

Conclusion: The Unyvero assay on BAL samples is useful in this specific population for rapid obtention of microbiological results and also for confirmation of the negativity of cultures and thus permits a better management of antibiotic therapy, leading to a reduction of antibiotic resistance selection pressure in the ICU.

Compliance with ethics regulations: Yes.

F-019 Do not underestimate RSV pneumonia among critically ill patients

Erwan Begot¹, Suzanne Champion¹, Charline Sazio¹, Benjamin Clouzeau¹, Alexandre Boyer¹, Hoang-Nam Bui¹, Marie-Edith Lafon², Camille Ciccone², Julia Dina³, Didier Gruson¹, Renaud Prével¹

¹CHU Bordeaux, Medical intensive care unit, Bordeaux, France; ²CHU Bordeaux, Virology laboratory, Bordeaux, France; ³National Reference Center for Measles Mumps and Rubella, CHU de Caen, Caen, France

Correspondence: Erwan Begot (erwan.begot@chu-bordeaux.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-019

Rationale: Respiratory syncitial virus (RSV) is a well-known cause of respiratory failure among neonates but its pathogenicity in adults is now emerging as a potential cause of viral pneumonia. Data are limited with conflicting results regarding RSV pneumonia severity in adults. Data are lacking about critically ill RSV patients’ characteristics and outcomes. The aim of this study is to compare RSV patients’ characteristics, care and outcomes to influenza patients’ ones.

Patients and methods: Patients diagnosed with RSV and influenza pneumonia admitted to our medical ICU were included. Data were retrospectively recorded. Quantitative data are expressed by median and interquartile range and compared by use of Mann–Whitney test. Qualitative data are expressed by number and percentages and compared by use of Fischer exact t-test. RSV strains were prospectively collected.

Results: Eighteen critically ill patients with RSV pneumonia and 95 with influenza pneumonia were included. RSV and influenza patients had the same characteristics at admission except for age (respectively 71yo [62; 81] vs 63yo [51; 70], p: 0.03). In particular, they had similar rates of underlying chronic pulmonary disease (respectively 10/18 (56%) vs 43/95 (45%), p = 0.45) or haematological malignancy (2/18 (12%) vs 9/95 (10%), p = 1.00). They exhibited similar presentation with similar SAPSII scores (respectively 46 [38; 63] vs 53 [34; 74], p = 0.55), PaO₂/FiO₂ ratio (respectively 195 [145; 256] vs 157 [100; 226], p = 0.15) and acute respiratory distress syndrome rates (respectively 7/18 (39%) vs 47/95 (49%), p = 0.44). They received similar treatment as suggested by oro-tracheal intubation rates (respectively 6/18 (33%) vs 52/95 (54%), p: 0.12) and antibiotics prescription (respectively 16/18 (89%) vs 88/95 (93%), p: 0.63). RSV and influenza patients also had the same rates of bacterial co-infections (4/18 (22%) vs 28 (29%), p: 0.78). Invasive aspergillosis remained a rare event but also occurred among RSV patients (1/18 (6%) vs 3/95 (3%), p: 0.51). Acute coronary syndromes were as frequent in both groups (respectively 2/18 (12%) vs 9/95 (10%), p = 1.00). Day-28 mortality was similar between RSV and influenza patients (respectively 3/18 (19%) vs 23 (24%), p = 0.76). RSV strains typing is under progress.

Conclusion: Patients suffering from RSV pneumonia admitted to ICU have similar presentation, care and day-28 mortality than critically ill influenza patients. The severity of RSV pneumonia diagnosis should not be underestimated by intensivists.

Compliance with ethics regulations: Yes.

F-020 Determinants of respiratory distress from seawater drowning: a prospective observational study from seven intensive care units on the French Riviera

Alexandre Robert¹, Denis Doyen², Pierre-Marie Bertrand³, Michel Kaidomar¹, Nicolas Clement³, Nicolas Bele¹, Nihal Martis², Hervé Quintard⁴, Hervé Hyvernat², Gilles Bernardin², Jean Dellamonica²

¹Fédération des réanimations du Var Est, Fréjus-Draguignan, France; ²Médecine Intensive Réanimation, Hôpital l’Archet 1, CHU de Nice, Nice, France; ³Médecine Intensive Réanimation, CH de Cannes, Cannes, France; ⁴Réanimation polyvalente, Hôpital Pasteur 2, CHU de Nice, Nice, France

Correspondence: Alexandre Robert (alex_robert@hotmail.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-020

Rationale: Respiratory distress from seawater drowning is commonly considered multifactorial. Etiologies are debatable and include heart failure, infection and acute respiratory distress syndrome (ARDS). Documented bacterial infections seems mostly related to the site of drowning. Data in this regard are scarce with prospective studies lacking. The objective of our study was to describe prospectively the characteristics and determinants of respiratory distress from Seawater Drowning.

Patients and methods: All patients admitted for seawater drowning to seven intensive care units (ICU) on the French Riviera in the summers of 2017 and 2018 were prospectively included. Recorded data included clinical features on examination, personal history, chest X-rays, echocardiography and biological results obtained within the first 48 h. A paired Student’s t-test was used to study statistical differences between quantitative variables on admission and during early evaluation (i.e. first 48 h).

Results: Forty-eight patients were admitted to seven centers of which 45 (94%) were diagnosed as having ARDS, 31 (65%) early pneumonia and 4 (8%) acute cardiogenic pulmonary edema. Twenty-one (44%) respiratory samples were collected but bacterial culture was positive in only 5 cases. Multidrug-resistant bacteria were not observed, and amoxicillin-clavulanate as first-line treatment was effective in all cases. Echocardiography performed in 38 (75%) patients was normal and unable to identify specific patient profiles. The median Clinical Pulmonary Infection Score (CPIS) on admission was 6 (IQR, 5–7) and decreased rapidly and significantly (p < 0.0001) within 24 h to 3 (IQR, 2–3) (Fig. 1).

Conclusion: Data from this multicenter cohort suggest that respiratory distress following seawater drowning can mimic bacterial pneumonia during the first 24 h with subsequent rapid clinical improvement in patients admitted to the ICU. Probabilistic antibacterial therapy should therefore be limited to the most severe patients. Isolate ARDS is often the only etiology found and is resolutive within 24 h. This prospective cohort is the largest of its kind and gives a better insight into the limited impact of cardiogenic and infectious processes on sea drowning-related respiratory distress.

Compliance with ethics regulations: Yes.

Fig. 1

Modifications in Clinical Pulmonary Infection Scores (CPIS) from admission, to day 1 of follow-up in seawater drowned patients admitted to the ICU. Scores were compared using a Student’s paired T-test

F-021 Impact of Extra-Corporeal Membrane Oxygenation (ECMO) support on piperacillin exposure in septic patients: a case–control study

Pierre Fillatre¹, Florian Lemaitre², Nicolas Nesseler³, Matthieu Schmidt⁴, Sebastien Besset⁴, Yoann Launey⁵, Adel Maamar⁶, Pierre Daufresne⁷, Erwan Flecher⁸, Yves Le Tulzo⁹, Jean-Marc Tadié⁹, Pierre Tattevin⁹

¹Service de Réanimation Polyvalente, Saint Brieuc, France; ²Service de Pharmacologie-CHU Pontchaillou, Rennes, France; ³Service de Réanimation de Chirurgie Thoracique et de Chirurgie Vasculaire-CHU Pontchaillou, Rennes, France; ⁴Service de Réanimation Médicale-CHU La Pitié Salpétrière, Paris, France; ⁵Service de Réanimation Chirurgicale-CHU Pontchaillou, Rennes, France; ⁶Service de Maladies Infectieuses et de Réanimation Médicale-CHU Pontchaillou, Rennes, France; ⁷Service d’hématologie-CHU Pontchaillou, Rennes, France; ⁸Service de Chirurgie Thoracique et de Chirurgie Vasculaire-CHU Pontchaillou, Rennes, France; ⁹Service de Maladies Infectieuses et de Réanimation Médicale-CHU Pontchaillou, Rennes, France

Correspondence: Pierre Fillatre (pierrefillatre@hotmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-021

Rationale: Patients treated with “ExtraCorporeal Membrane Oxygenation” (ECMO) are at a higher risk of developing nosocomial infections and they are consequently often treated with beta-lactams. French guidelines recommend obtaining beta-lactam trough concentrations above four times the minimal inhibitory concentration (MIC) of the causative bacteria. The ECMO device may alter the pharmacokinetics of these medications, which may result in underexposure to beta-lactam antibiotics.

Patients and methods: This observational, prospective, multicenter, case–control study was performed in the intensive care units of two tertiary care hospitals in France. ECMO Patients with sepsis treated with piperacillin-tazobactam were enrolled. Control patients were matched according to SOFA score and creatinine clearance. The pharmacokinetics of piperacillin was described based on a population pharmacokinetic model, allowing to calculate the time spent above 4 × the MIC breakpoint for Pseudomonas aeruginosa susceptibility after the first dose and at steady state between two piperacillin infusions.

Results: Forty-two patients were included. The median age was 60 years [49–66], the SOFA score was 11 [9–14], and median creatinine clearance was 47 mL/min [5–95]. There was no significant difference in the time above 4 x MIC in patients treated with ECMO and controls during the first administration (p = 0.184) and at steady state (p = 0.309). There was no significant difference between the trough at steady state (p = 0.535), with 18/42 patients (43%) exhibiting concentrations of piperacillin lower than 4 x MIC. ECMO support was not associated with a steady state trough concentration below 4 x MIC (OR = 0.5 [0.1–2.1], p = 0.378). The only variable independently associated with this risk was a creatinine clearance ≥ 40 mL/min, (OR = 4.3 [1.1–17.7], p = 0.043).

Conclusion: ECMO support has no significant impact on piperacillin exposure. Intensive care unit patients with sepsis are, however, frequently underexposed with piperacillin, which suggest that therapeutic drug monitoring should be strongly recommended for severe infections.

Compliance with ethics regulations: Yes.

Fig. 1

Ratio of the time between two piperacilline infusion spent above 64 µg/mL during the first administration or at steady state in ECMO patients and in control patients

F-022 End-of-life management strategies the opinion of Tunisian intensivists in the Vincent lambert’s case

Meriem Fatnassi¹, Hela Maamouri¹, Nassereddine Foudhaili¹, Nozha Brahmi¹

¹Service de réanimation toxicologique et polyvalente centre Mahmoud Yaacoub, Tunis, Tunisia

Correspondence: Meriem Fatnassi (merifatnassi@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-022

Rationale: Decision-making at the end of life is a very difficult situation frequently encountered by Intensivists. The media debate over the case of Vincent Lambert focused on issue related to the end-of-life management strategies. This study aims to discover the opinion of Tunisian physicians in different hospitals.

Patients and methods: We have realized a questionnaire composed of 17 items. Sixty intensivists participated to our study. They have different status in intensive care and anesthesia: 10 professors, 26 senior doctors and 24 interns.

Results: We collected 60 opinions. 96.6% of intensivists had taken care of patients in vegetative coma and 88.1% (n = 52) agreed with the fact that patients in vegetative coma are eligible to withold or even withdraw treatment. 50.8% were against active treatment limitating decisions. It was related to humanitarian believes in 62.1% of cases and to religious convictions in 20.7%. The other causes were moral, legal and ethic. 83.9% thought that the treatment limitation decision was not prohibited by Islamic religion. Furthermore, 79.3% approved passive limiting care et 83.1% were against euthanasia. In pratice, 72.9% have used deep sedation as palliative care. The common methods of care limiting were: not using catecholamines in 94.9%, stopping enteral and parenteral nutrition in 88.1% and administration of deep sedation while maintaining the intubation in 62.7%. Most of intensivists (96.6%) recognized that Vincent lambert’s affair raised the problem of end-of-life management. 59.3% agreed with the medical and the partner decision. 56.1% found that the lawsuit initiated by parents to human rights was disproportionate and that media influenced negatively the matter. Finally, the majority of doctors suggested that advance directives may help in those situations.

Conclusion: The end of life management remains a subject of controversy. Advance directives and formulated laws such as Leonetti law can be helpful.

Compliance with ethics regulations: Yes.

F-023 Impact of a visual support dedicated to prognosis of patients on symptoms of stress of family members

Pascal Beuret¹, Gabrielle Burelli¹, Arthur Vincent¹, Jeremy Mallard¹, Sarah Meffre¹, Alizee Maarek¹, Sixtine Bonnet¹, Jerome Morel²

¹Réanimation polyvalente/Centre hospitalier, Roanne, France; ²Réanimation polyvalente B/CHUI, Saint Etienne, France

Correspondence: Pascal Beuret (pascal.beuret@ch-roanne.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-023

Rationale: Family members commonly have inaccurate expectations of patient’s prognosis. Adding to classic oral information a visual support, depicting day by day the evolution of the condition of the patient, improves the concordance in prognosis estimate between physicians and family members. The objective of this study was to evaluate the impact of this support on symptoms of anxiety/depression of family members.

Patients and methods: We conducted a bi-center prospective before-after study. All consecutive patients admitted in the two ICUs were eligible. In the before period (3 months), family members received classic oral information. In the after period (3 months), in addition to classic oral information, the visual support (Fig. 1) was available for family members in the patient’s room from the day of admission until discharge from the ICU. At day 5 and 90 from admission, symptoms of anxiety/depression of referent family member were evaluated by Hospital Anxiety and Depression Scale (HADS).

Results: 140 patients and their referent family members were included (77 in period before and 63 after). Characteristics of patients of the two groups were similar regarding age, reason for admission, SAPS II at admission and SOFA score at day 5. Also characteristics of referent family members were comparable in terms of age, sex ratio, type of relationship with the patient and number of visits since admission. At day 5, total HAD score was 17 [9–25] in the group before without the support and 15 [10–22] in the group after with the support (p = 0.43). The prevalence of symptoms of anxiety (HAD-A score > 7) and depression (HAD-D score > 7) was similar in the two groups (respectively 66.2% and 49.4% in the group before, and 68.3% and 36.5% in the group after (NS)). At day 90, total HAD score was 11 in the group before [7–16] and 9 [5–16] in the group after (p = 0.38). By multivariate analysis the following factors were significantly associated with total HAD score > 12 at day 5: age of patient (OR 0.98 [0.96–0.999]), number of visits of referent (OR 2.72 [1.09–6.76]) and previous or current treatment of referent for anxiety or depression (OR 2.76 [1.08–7.06]).

Conclusion: In this study, the use of a visual support dedicated to prognosis of patients did not modify the level of stress of family members.

Compliance with ethics regulations: Yes.

Fig. 1

An example of visual support

F-024 Sedation practice and discomfort during withdrawal of mechanical ventilation in critically ill patients at End-of-Life: a post hoc analysis of a multicenter study

René Robert¹, Amélie Le Gouge², Jean Reignier³

¹Université de Poitiers, Poitiers, France/Inserm CIC 1402, ALIVE, service de Médecine Intensive Réanimation, Poitiers, France; ²Inserm CIC 1415, Tours, Tours, France; ³Université de Nantes, Nantes, France

Correspondence: René Robert (r.robert@chu-poitiers.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-024

Rationale: The use of sedation and opioids at the end of life is a topic of considerable ethical debate. Incidence of discomfort during the end-of-life of ICU patients and impact of sedation on discomfort are poorly known.

Patients and methods: Post-hoc analysis of an observational prospective multicenter study comparing terminal weaning vs. immediate extubation for end-of-life in ICU patients, aimed at assessing the incidence of discomfort events according to levels of sedation. Discomforts including gasps, significant bronchial obstruction or high behavioral pain scale score, were prospectively assessed by nurses from mechanical ventilation withdrawal until death. Level of sedation was assessed using the Richmond Agitation Sedation Scale (RASS).

Results: Among the 450 patients included in the original study, 226 (50%) experienced discomfort after mechanical ventilation withdrawal. Patients with discomfort received lower doses of midazolam and equivalent morphine, and less frequently had deep sedation (RASS -5) than patients without discomfort (59% vs 79%, p < 0.001). After multivariate logistic regression, immediate extubation was the only factor associated with discomfort whereas deep sedation and administrations of vasoactive drugs were two factors independently associated with no discomfort. Death occurred less rapidly in patient with discomfort than in those without discomfort (7.3 h [1.9–25.0] vs 1.6 [0.3–6.0], P < 0.0001) (Figure). Long-term evaluation of psychological disorders in family members of dead patients did not differ between those with discomfort and the others.

Discussion: Despite the theoretically expected anticipatory titrated doses of opioids and benzodiazepines to alleviate any discomfort after withdrawal of mechanical ventilation, half of the patients did not receive sedation or opiate when the decision to withdraw mechanical ventilation was taken. A major point that could interfere with the continuous deep sedation practice until death is the fear of potentially hastening death, and there is much controversy regarding its proper use in end-of-life care.

Conclusion: Discomfort was frequent during end-of-life of ICU patients and was mainly associated with terminal extubation and less profound sedation.

Compliance with ethics regulations: Yes.

Fig. 1

Survival probability in patients with and without discomfort

F-025 Post-death meeting for relatives of ICU deceased patients: a way to alleviate grief symptoms?

Thibault Dekeyser, Dominique Malacrino, Caroline Sejourne, Marie-Christine Dufossez, Mehdi Marzouk, Imen Rahmani, Laura Lstiburek, Christophe Vinsonneau

Service de Médecine Intensive Réanimation, Centre hospitalier de Béthune Germon et Gauthier, 62408, Béthune, France

Correspondence: Thibault Dekeyser (thibault.dekeyser@live.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-025

Rationale: Bereavement in Intensive Care Unit (ICU) is associated with psychiatric disorders on relatives called Post-Intensive Care Syndrome Family (PICS-F). No isolated intervention (such as condolence letter) has shown a positive effect on these disorders, despite a well acceptance by relatives. We thought that a more integrated bereavement program should be considered. The goal of this study is to evaluate a combined psychologist-physician post-death meeting (PDM) in a bereavement program to evaluate needs and adhesion of relatives, and the effect on symptoms of anxiety and depression.

Patients and methods: Monocentric, prospective study focused on relatives of patient admitted > 48 h and deceased in ICU. During patient’s stay, relatives’ presence was allowed on a 24 h-basis and they could meet a clinician psychologist. Formal meeting between relatives and the staff was realized at patient’s admission and after important decision-making treatment. Two weeks after patient’s death, the psychologist called relatives to offer emotional support and to invite to a PDM. PDM occurs 3 weeks after patient’s death with the psychologist and the physician in charge of the patient. The objectives of the meeting were to provide emotional support, to answer medical question, and to detect symptoms of anxiety and/or depression with the Hospital Anxiety and Depression Scale (HADS). We hypothesized that PMD would be able to alleviate PICS-F at 3 months. We aimed to enroll 70 families to detect a 20% lowering of HADS.

Results: The rate of PDM acceptance was lower than expected. After 53 inclusions, only 12 relatives accepted the PDM, whereas the phone call was well perceived (86%). Main association with acceptance of PMD was a short duration of ICU stay (4.3 days [2–5.3] vs 7.3 days [3.7–10.9] P = 0.027) and ICU admission for acute respiratory failure (66.7% vs 19.5%, P = 0.003) (Table 1). We found no relation between the number of in ICU meeting (psychologist of medical staff) and PMD acceptance. For relatives who accept PMD we found a high proportion of symptoms of anxiety and depression (92% and 83%) with a HADS at 28.5 [22–34.5] (Median, IQR). No evaluation was performed at 3 months.

Conclusion: Post death contact appears well perceived by relatives but PMD quite useless. This result may be explained by the inclusion of only late death (> 48 h) where psychologist and medical staff had the opportunity to support relatives. Further study should focus on early death (< 48 h).

Compliance with ethics regulations: Yes.

Table 1 Demographic admission data of the overall study population and subset with post-death meeting acceptance or not

F-026 The pediatric intensivist’s position when confronted to parents opposed to termination of life-sustaining treatments on their child

Delphine Brown, Marcel-Louis Viallard, Laure de Saint-Blanquat, Agathe Béranger, Lucie Derauglaudre, Marion Gregnon, Sylvain Renolleau, Marie-France Mamzer, Laurent Dupic

Necker, Paris, France

Correspondence: Delphine Brown (delphine.brown@aphp.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-026

Rationale: Pediatric intensivists frequently question themselves on the issue of limitation or termination of life-sustaining treatments (LLST) carried out on children. Such a decision comes under the Claeys-Leonetti Law which forbids doctors from applying unreasonable treatment However, every so often, parents oppose themselves to a collegial LLST decision that the medical and paramedical team had taken. Such cases can even end up in Court. In order to sort out this problem, this study focused on the factors that underlie the disagreement and the solution brought forward by pediatricians whenever parents demand to persue treatments although considered as unreasonable obstinacy.

Patients and methods: We carried out a qualitative study involving three multipurpose pediatric critical care unit. All pediatricians operating within these units were contacted. Those who volonteered were met individually for a semi-directed interview. Every interview was recorded and entitled to a complete hand-written retranscription. The interviews were analysed following the phenomenological interpretive analysis method and were subject to dual listing.

Results: 16 pediatricians out of 30 took part in the study. 8/16 claimed they would increase treatments or carry out cardiopulmonary resuscitation acts if asked to do so by parents, even if this went against the initial collegial decision. 8/16 claimed they would persue treatments although not beyond the current level. 2/16 said they would oppose themselves to parents concerning blood transfusion for comfort reasons. Several key factors were identified as leading a doctor to the non-application of a LLST decision: the certainty regarding the child’s death on a short or mid-term basis (16/16), the litigiousness risk (14/16), the apprehension of mediatic pressure (14/16), the fear of a violent reaction from parents (13/16), other self-interest positions within the medical team (12/16), empathy towards parents (11/16), the uncertainty concerning the neurological prognosis (5/16), the lapse of time needed to fully accept the application in force of a decision (5/16). 12 pediatricians out of 16 admitted their own-suffering when confronted to the situation.

Conclusion: This study points out that pediatricians tend to follow parents’ position when confronted to parental opposition. In such situations, pediatricians go against their own decision in order to safeguard the parental alliance even if it leads to unreasonable obstinacy, thus conflicting with medical deontological Code obligations.

Compliance with ethics regulations: Yes.

F-027 Advance directives: a general overview

Ameni Abidi, Hela Maamouri, Meriem Fatnassi, Takoua Khzouri, Nozha Brahmi

Service de réanimation toxicologique et polyvalente, Centre Mahmoud Yaacoub d’assistance médicale urgente, Tunis, Tunisia

Correspondence: Ameni Abidi (drabidiameni@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-027

Rationale: End-of-life management strategies are clearly a worldwide issue of major importance that intensivists have to deal with on a daily basis. Advance directives may be the solution sought to guide physicians to take such difficult decisions. Yet, health care directives are not legislated in Tunisia. The objective of this project was to draw a general descriptive overview to assess patients’ wishes in Tunisia.

Patients and methods: Data were collected from a 39-item-questionnaire based on the French Intensive Care Society’s form for advance directives which was filled by 101 people of general population in Tunisia, including doctors and paramedics, from May to mid-September 2019. All people included were 18 or older and well informed of the form’s utility.

Results: A total of 101 participants were included. The mean age was 34.6 ± 13.5 years with extremes of 18 and 76 and a sex ratio of 1.7. Fourty-one (36.6%) were either doctors or nurses and 3 (3%) did suffer from a severe medical condition. Among all the participants, 30 (29.7%) thought that end-of-life decisions were up to the doctor. For the rest, they willingly chose to be hospitalized in an ICU, to undergo cardiopulmonary rescuscitation and to have ventilation support with orotracheal intubation or tracheostomy respectively in 63 (87.5%), 60 (83.3%) and 67 (93.1%) of the cases. Only 4 (5.6%) refused temporary dialysis. When asked about sequelae they can live with, participants accepted hemiplegia in 63.9% and paraplegia in58.3% of the cases. On the contrary, 68 (94.4%) refused to live in permanent coma and 55 (79.9%) disagreed to undergo tracheostomy and ventilation for life. Moreover, 31 (43.1%) found that serious un aesthetic sequelae was a fatal consequence they could not survive. As well, only 7 (9.7%) consented to live with deep intellectual deficiency. Regarding palliative care, 53 (73.6%) participants wished to be profoundly sedated until death, 65 (90.3%) prefered to die home over 2 (2.8%) in hospital. Sixty-two (84.9%) desired to see a representative of their religion. Furthermore, 65 (89%) were for organ donnation. Gender, being a health care professional and age under 30 versus equal or over30 were not significant in dependent factors (p > 0.05).

Conclusion: It is our duty as health care professionals to spread advance directives awareness and education. Nevertheless, the law should keep the pace with ethics evolution.

Compliance with ethics regulations: Yes.

F-028 Development of a hemodialysis model in rats with septic acute kidney injury

Nicolas Benichou, Stéphane Gaudry, Sandrine Placier, Juliette Hadchouel, Alexandre Hertig, Christos Chatziantoniou, Didier Dreyfuss

French National Institute of Health and Medical Research (INSERM), UMR_S1155, Remodeling and Repair of Renal Tissue, Hôpital Tenon, F-75020, Sorbonne Université, Paris, France

Correspondence: Nicolas Benichou (nico.benichou@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-028

Rationale: Adapted organ support techniques are needed to enhance reliability of preclinical animal experiments in the intensive care setting (Guillon, Annals of Intensive Care-2019). A few renal replacement therapy (RRT) models have already been developed in rats, mostly hemodialysis in chronic kidney disease models or hemofiltration techniques in sepsis experiments. Mounting evidence from clinical (Gaudry, NEJM-2016) and histopathological studies suggest that RRT for acute kidney injury (AKI) could impair renal recovery by acting as a ‘second hit’ leading to a maladaptive repair of tubular epithelium. We aimed to study this hypothesis in a hemodialysis model in rats with septic AKI.

Patients and methods: On day 1, Sprague–Dawley rats were injected with lipopolysaccharide or placebo (NaCl 0.9%) intraperitoneally. On day 2, anesthetized rats underwent femoral artery catheterization for hemodynamic parameters monitoring. At the same time, one femoral vein and one carotid artery were catheterized for arterio-venous sterile extracorporeal circulation with or without passing through a miniature sterile Polyester Sulfone hemodialyzer (20 cm2 surface, 50 kDa pores, MicroKros^®) filled with dialyzate liquid in the outer compartment (Table 1). Vessels were ligated after the procedure and rats allowed to awaken. On day 3, rats were sacrificed.

Results: All rats injected with lipopolysaccharides O127:B8 10 mg/kg survived at day 2. Anesthesia was much challenging: Ketamine + Xylazine and Tiletamine-Zolazepam + Xylazine required induction and maintenance intraperitoneal injections. These medications induced important hemodynamic parameters fluctuations and high mortality. Isoflurane gas inhalation enabled better stability, less hypothermia and quick awakening. Adequate temperature was controlled with a heating pad during the procedure and an incubator after. Supine position was maintained. The whole circuit was anticoagulated with 1 ml of heparinized saline 100 UI/ml, since clots occurred in the absence of anticoagulation and bleeding when higher dosing was used. Circuit (< 1.5 ml including dialyzer) was filled with saline solution before initiation, and total restitution of blood at the end of the experiment prevented any blood transfusion requirement. Hematocrit was determined at beginning (50%) and end of experiment (40%). A peristaltic pump provided a blood flow rate of 1.0 mL/min, (higher rate was not tolerated) for 2 h. Of note, rats who underwent sham procedure (vessels ligature only) survived and did not display AKI. Circulation of a counterflow dialysate in the dialyzer is planned but has not been performed yet.

Conclusion: This hemodialysis system for rats is feasible at a reasonable price and might help research involving RRT in either CKD or AKI.

Compliance with ethics regulations: Yes.

Table 1 Hemodialysis model in rats: characteristics

F-029 Incidence and renal prognosis of positive antibiotic associated crystalluria in infective endocarditis patients

Matthieu Jamme¹, Leopold Oliver², Julien Ternacle², Raphael Lepeule³, Sovannarith San², Amina Moussafeur², Antonio Fiore⁴, Nicolas Mongardon⁵, Michel Daudon⁶, Pascal Lim², Emmanuel Letavernier⁶

¹Renal ICU and Kidney Transplantation, Tenon university hospital, APHP, Paris, France; ²Cardiology, Mondor university hospital, APHP, Creteil, France; ³Infectious mobile unit, Mondor university hospital, APHP, Creteil, France; ⁴Heart surgery, Mondor university hospital, APHP, Creteil, France; ⁵Surgical ICU Mondor university hospital, APHP, Creteil, France; ⁶Physiology unit, Tenon university hospital, APHP, Paris, France

Correspondence: Matthieu Jamme (mat.jamme@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-029

Rationale: Infective endocarditis (IE) is a rare disease characterized by an high mortality. High-dose antibiotic has long been recognized to induce acute kidney injury (AKI) through various routes of injury. Next to acute interstitial nephritis, antibiotic-associated crystal (AAC) nephropathy has been increasingly acknowledged as a cause of severe AKI. We aimed at describing the prevalence of AAC of patient admitted and treated for suspected IE, identifying factors associated with positive AAC and assessing renal prognosis of positive AAC patients.

Patients and methods: All patients admitted consecutively for suspicion of IE to the cardiology unit of Mondor hospital (APHP, France) between 2017, June, 1st and 2018, June, 1st were included in the present study. IE was defined according to the modified Duke criteria. Urinary samples was collected during the first week of hospitalization and transferred to Tenon hospital (APHP, France) laboratory within 4 h. Outcome principal was the occurrence of AAC defined by the observation of antibiotic crystal formation with combining polarized microscopy and infrared spectroscopy. Secondary outcomes were the occurrence of AKI stage ≥ 2 according the KDIGO classification and the estimated glomerular filtration rate (GFR) estimating by MDRD equation at day-14.

Results: 34 patients were included in the analysis. Most of them were men (71%) and the median age was 70 [62–77] years. Pathogen was identified in 82% of patients. The most frequently identified pathogen was Streptococcus. Modified DUKE criterias defined IE as definite in 21 (64%), possible in 4 (12%) and suspected in 9 (24%) cases. First line antibiotherapy was composed by Amoxicilline (94%), Gentamycine (87%) and Cefazoline (37%). 65 crystallurias were performed within a median time 5[3–6] days. AAC was observed in 18 urinary analysis from 14/34 patients. Multivariate logistic mixed effect analysis identified blood amoxicillin concentration (OR = 1.02[1.01–1.05]) and urinary pH (OR = 0.79[0.62–0.94]) as risk factors to AAC. Severe AKI occurred in 18 (53%) patients. Statistical trend was observed with the occurrence of AAC (p = 0.07). Estimated GFR was below to 60 mL/min/1.73 m² at day 14 in 32% of patients. AAC was not associated with a eGFR < 60 mL/min (p = 0.10). After exclusion of patients with CKD, we observed a statistical trend with the occurrence of AAC (p = 0.08).

Conclusion: AAC is a common complication of patients treated by high dose of antibiotics for suspected IE. Our results suggest of potential renal short term impact of AAC.

Compliance with ethics regulations: Yes.

Fig. 1

Probability of acute kidney injury in patients with (black line) and without crystalluria (red line)

F-030 Cytogenetic Abnormalities and Risk for Acute Kidney Injury in Critically Ill Patients with Acute Myeloid Leukemia (AML)

Inna Mohamadou¹, Michaël Darmon¹, Swann Bredin¹, Emmanuel Canet¹, Lara Zafrani¹, Etienne Lengline¹, Achille Kouatchet², Alexandre Demoule³, Pierre Perez⁴, Frédéric Pene⁵, Djamel Mokart⁶, Christine Lebert¹, Fabrice Bruneel⁷, Emmanuel Raffoux¹, Virginie Lemiale¹, Elie Azoulay¹

¹Hôpital Saint-Louis, Paris, France; ²CHU de Poitiers, Poitiers, France; ³Hôpital de la Pitié Salpêtrière, Paris, France; ⁴CHRU Nancy Brabois, Nancy, France; ⁵Hôpital Cochin, Paris, France; ⁶Institut Paoli-Calmettes, Marseille, France; ⁷Hôpital André Mignot, Versailles, France

Correspondence: Inna Mohamadou (inna.m@hotmail.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-030

Rationale: Acute kidney injury (AKI) is a challenging organ dysfunction in hematology patients. Indeed, optimal management relies on patient’s ability to receive the best standard of care and optimal chemotherapy regimen, which may not be possible in case of severe AKI. Moreover, in ICU patients with high tumoral burden, patients with baseline AKI are at high risk of tumor lysis syndrome (TLS), need for renal replacement therapy (RRT), delayed renal recovery and increased mortality. In patients with aggressive hematologic malignancies, risk factors for AKI have been previously identified. However, in critically ill patients with AML, the risk for AKI across different AML French-American-British (FAB) classes and on cytogenetic abnormalities has never been assessed.

Patients and methods: Patients were classified based on cytogenetic data according to the WHO classification into three groups (favorable, intermediate, or unfavorable). AKI was defined based on KDIGO definitions using creatinine level at ICU admission and need for RRT. A Cox proportional-hazards model was used to identify factors associated with long term survival.

Results: Among the 144 AML patients (age 58y [45–68], 60% men, performance status 1 [0–2]), 54 (37.5%)) were AML4 or 5, 24 (16.7%) were AML3, and 66 (45.8%) had other FAB classes. According to the WHO-cytogenetic classification, 13 (9%) patients were in the favorable group, 77 (54%) in the intermediate group, and 30 (21%) in the unfavorable group. AML3 was found in 24 (16.7%) patients. Patients were untreated in 78.5% of the cases and half of them were hyperleukocytic. SOFA score was 5 (4–9) at admission. Baseline creatinine was 88 (80–97). Based on KDIGO, 68 (47.2%) patients had no AKI, 30 (20.8%) had a stage 1, 17 (12%) a stage 2 and 29 (20.1%) a stage 3 AKI. Throughout the ICU stay, septic shock occurred in 31 (21%) patients, 23 (16%) presented TLS, 67 (46%) received vasopressors, 42 (29.2%) RRT, and 4 (31%) died in the ICU. By multivariable analysis, factors associated with AKI were male gender, poor performance status, hyperleukocytic AML, SOFA score, DIC and the WHO-cytogenetic classification (HR 2.60 [2.17–96.7] for the intermediate class, 1.98 [1.18–69.7] for the unfavorable class). FAB and WHO cytogenetic-based classifications were not significantly associated with mortality.

Conclusion: Half the critically ill AML patients present with AKI, a condition that is independently associated with increased mortality. Patients in the intermediate and unfavorable WHO-cytogenetic groups are at high risk of AKI, independently of leukocytosis and severity. Studies to elucidate the mechanisms pertaining AKI in these specific subgroups are warranted.

Compliance with ethics regulations: NA.

Fig. 1

AKI according to WHO-cytogenetic classification of AML

F-031 Do Renal Resistive Index (RRI) measurement predict renal function evolution in critically ill patients with stage 2 Acute Kidney Injury (AKI)?

Patricia Wiesen, Didier Ledoux, Paul Massion, Benoit Misset

CHU Liège-Soins Intensifs, Liege, Belgium

Correspondence: Patricia Wiesen (p.wiesen@chuliege.be)

Ann. Intensive Care 2020, 10 (Suppl 1):F-031

Rationale: The place of Doppler-derived Renal Resistive Index (RRI) as a potential renal function predictor remains controversial because of its low specificity. Patient’s vascular compliance is known to affect this measurement. Moreover, the optimal timing for the most accurate prediction is unknown.

Patients and methods: This prospective observational study included 57 patients where RRI was measured when they met at least stage 2 Acute Kidney Injury (AKI) according to KDIGO criteria. Risk factors, severity score, presumed AKI aetiology (hypoperfusion, sepsis, congestion, vascular glomerular dysregulation), renal evolution, Renal Replacement Therapy (RRT) requirement, clinical and biological data were recorded. Patients were classified according to their RRI value (cut off: 0.70) for comparison.

Results: Initial RRI measurement was higher than 0.7 for 39 patients (68%).

There were no significant relationship between RRI and past medical history or severity score.

We observed a significant negative correlation between RRI and diastolic arterial pressure (p = 0.004) and heart rate (p = 0.004) as it could be expected by RRI formula. An increased RRI was associated with higher potassium (p = 0.019) and higher creatinine levels (p = 0.042).

Although not significant, we found a higher rate of subsequent RRT in the high RRI group (23% vs 6%, p = 0.146).

Over the first 3 days, fluid balance was significantly different between groups (2217 ml vs –1314 ml respectively for low and high RRI group, p = 0.017). Since standard of care were similar, this suggests different fluid volume status between the two groups. In the low RRI group, the cause of AKI could predominantly be prerenal since positive fluid balance was not explained by more severe AKI with refractory oliguria as shown by the low RRT rate. Nevertheless, we did not observed any relationship between RRI and the evolution of serum urea or creatinine levels, nor with the presumed aetiology of AKI.

Conclusion: When focussing on the first RRI measurement once stage 2 AKI was reached, RRI ≤ 0.7 seems to be in favour of prerenal and transient renal dysfunction even if this is not supported by creatinine serum evolution.

Compliance with ethics regulations: Yes.

Table 1 main characteristics of the population of the study

F-032 Doppler based resistive index measurement in ICU patients and influence of inter-operator variability: Results of a multicenter cohort study

Jean Jacques Tudesq¹, David Schnell², Marie Reynaud³, Stéphane Rouleau², Ferhat Meziani⁴, Alexandra Boivin⁴, Mourad Benyamina⁵, Francois Vincent⁶, Alexandre Lautrette⁷, Christophe Leroy⁷, Yves Cohen⁸, Matthieu Legrand⁵, Jerome Morel³, Jeremy Terreaux³, Aurelie Bourmaud⁹, Michaël Darmon⁵

¹Saint-Louis University Hospital, Paris, France; ²Angouleme Hospital, Angouleme, France; ³Saint-Etienne Hospital, Saint-Etienne, France; ⁴Strasbourg Hospital, Strasbourg, France; ⁵Saint-Louis Hospital, Paris, France; ⁶Montfermeil Hospital, Montfermeil, France; ⁷Clermont-Ferrand Hospital, Clermont-Ferrand, France; ⁸Avicenne Hospital, Bobigny, France; ⁹Robert Debre Hospital, Paris, France

Correspondence: Jean Jacques Tudesq (jean-jacques.tudesq@aphp.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-032

Rationale: Clinical data regarding factors that may influence Doppler-based resistive index (RI) at bedside are limited. Moreover, influence of operator has poorly been assessed. This study aimed at evaluating clinical characteristics associated with RI at bedside and to delineate influence of inter-operator variability as this regard.

Patients and methods: Post-hoc analysis of a multicentre prospectively collected dataset. Adult patients requiring mechanical ventilation were included. Patients with severe chronic renal dysfunction or known renal artery stenosis were excluded. AKI was defined according to KDIGO definition. Renal Doppler was performed at study inclusion. Operators involved in this study were anonymised. Results are reported in n (%) or median (IQR). Adjusted factors associated with AKI development were assessed using mixed linear regression model with the operator as random effect on the intercept.

Results: Overall, 351 patients were included in this study, including 149 patients with AKI stage 1 (42.5%), 33 patients with AKI stage 2 (9.0%) and 51 patients with AKI stage 3 (14.5%). Median age was 62 years (IQR 51–70), and 129 were of female gender (36.8%). Vasopressors was required in 184 patients (52.5%). RI was associated with AKI severity, with a RI of 0.65 (0.59–0.70) in patients without AKI and increasing steadily to 0.72 (0.62–0.78) in patients with AKI stage 3. After adjustment for confounders, factors independently associated with RI were age (estimate per year: 0.001, sd: 0.0003; p < 0.001), case mix (emergency surgery estimate (vs. medical): − 0.06, sd:0.03; p = 0.03), hypovolemia at ICU admission (estimate: 0.03, sd:0.02; p = 0.02), underlying cardiac comorbidity (estimate: 0.05, sd:0.02; p = 0.002), use of norepinephrine (estimate: 0.03, sd:0.0; p = 0.007), and AKI stage 3 (vs. no AKI) (estimate: 0.04, sd:0.02; p = 0.02). When forced in the final model, mean arterial pressure, pulsed pressure, plateau pressure, heart rate, sepsis were not selected. Operator was found to influence significantly RI in the mixed model (Fig. 1a; mean OR 0.03 [0.01–0.05]; p < 0.001). To further depict the influence of inter-operator variability, adjusted changes in RI according to age (1a) and AKI KDIGO stage (1b) are reported for each operator (each line depicting an operator).

Conclusion: Our study suggests that Doppler-based RI in critically ill patients is influenced by patients’ case-mix, age, underlying dehydration or shock, cardiac comorbidities, and AKI severity. Our study confirms and illustrates a strong inter-operator variability of RI measurement in ICU that persists after adjustment. Whether this inter-operator variability impairs RI reliability and usefulness at bedside deserve to be more carefully assessed.

Compliance with ethics regulations: Yes.

Fig. 1

Reporting change in RI according to operators after adjustment for confounders (one line per operator)

F-033 Echocardiographic follow-up of right ventricular dysfunction is the best way to evaluate cardiorenal syndrome (CRS) in ICU

Mario Geneix¹, Sebastien Moschietto², Antoine Frouin², Fanny Depeyre², Thibault Dupont³, Florent Montini²

¹CH Timone, Marseille, France; ²CH Henri Duffaut, Avignon, France; ³CHU Kremlin Bicêtre, Paris, France

Correspondence: Mario Geneix (mario.geneix@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-033

Rationale: Type 1 Cardiorenal syndrome (CRS) is defined by acute decompensated heart failure (AHF) leading to secondary acute kidney injury (AKI) due to persistent congestion. Diagnosis is often difficult and relies on compatible clinical history, symptoms and biology. There are currently no studies evaluating the reliability of transthoracic echocardiography (TTE) as a diagnostic tool in CRS. Therefore, the aim of this study was to assess quantitative and qualitative echocardiographic parameters in patients with CRS in the ICU at baseline and after appropriate care (pharmacologic therapy, with or without additional ultrafiltration).

Patients and methods: We conducted an observational, prospective, single-center study in the ICU department of a general hospital. Patients admitted in the ICU and presenting with type 1 CRS were included. Diagnosis of CRS was made based on clinical context, biology and confirmed by two attending physicians (nephrologist and cardiologist). Patients presenting with other causes of AKI were excluded. Transthoracic echocardiography was performed at baseline and at day 7 after treatment by the same trained operator for the same patients. We report various echocardiographic indices of right and left ventricular function, filling pattern and venous pressure at these two timepoints.

Results: A total of 27 patients were included in this study. At baseline (D0), 96.3% of patients had signs of congestion (IVC dilation > 2 cm), 76% had an altered S-wave (< 11.5 cm/s), 72.73% had an altered TAPSE (< 15 mm), 85.19% had an elevated RV/LV surface ratio (> 0.6). Between baseline and D7, under appropriate management, IVC size significantly decreased ([− 1.8; − 0.3] p < 0.001), the number of patients with an elevated RV/LV diameter ratio (> 0.6) also decreased (OR [IC 95%] = 0.087 [0.02; 0.37] p < 0.001), weight decreased (4.1 kg (5.24) p < 0.001), whereas natriuresis significantly increased (47.11 mM/ml (45.55) p < 0.001), and the amount of vasopressors support decreased (− 0.04 ug/kg/min [− 5.9; 0.46] p < 0.001). Other parameters including creatinine level (− 14.33 uM/L [− 42.02; 13.35]), cardiac index (+ 0.35 L/min/m² [− 0.05; 0.73]), and S-wave velocity (0.32 cm (2.67)) showed non-significant changes.

Conclusion: Main echocardiographic findings at baseline in patients with type 1 CRS consist of a right ventricular dysfunction associated with a state of increased venous pressure and congestion in the form of an IVC dilation. We report that weight, RV/LV diameter ratio, and IVC diameter might constitute good follow-up parameters to monitor treatment response.

Compliance with ethics regulations: Yes.

Fig. 1

Echocardiographic right ventricular parameters at baseline

F-034 Is citrate required for heparin-free intermittent hemodialysis in critically ill patients?

Chloé Medrano, Olivier Cointault, Laurence Lavayssiere, Marie-Béatrice Nogier, Nassim Kamar, Stanislas Faguer

Département de Néphrologie et transplantation d’organes, Toulouse, France

Correspondence: Chloé Medrano (chloemedrano@hotmail.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-034

Rationale: Critically ill patients are at higher risk of bleeding but also dialysis filter clotting (inflammatory state). Intermittent hemodialysis with calcium-free citrate-containing (0.8 mmol/L) dialysate (CafCit-IHD) recently emerged as a new safe and simple alternative to continuous renal replacement therapy allowing heparin-free extended dialysis sessions (> 5 h). In this study, we aimed to answer to two issues still unresolved: (i) can citrate contained in the dialysate accumulate and lead to citrate intoxication in patients with liver disorders, and (ii) can citrate be avoided using citrate- and calcium-free dialysate (CCF-IHD)?

Patients and methods: Monocentric retrospective study. Among the 450 sessions performed with CafCit-IHD, the 20 IHD sessions (18 critically ill patients) with citrate measurement available before and after the dialysis filter were reviewed. Estimation of the liver clearance was performed using the Picco Lemon^® system (Pulsion). In addition, 8 sessions performed using CCF-IHD were reviewed.

Results: All the patients had liver disorders (post-liver transplantation period n = 2; cirrhosis with Child > A6). Among the eighteen CafCit-IHD patients, fifteen (75%) and six (30%) received mechanical ventilation or vasopressive drugs, respectively. The median time of the dialysis session was 5 h [2–5], with hourly ultrafiltration rate of 400 mL (one premature termination not related to dysfunctional catheter). In all patients, ionized calcium (iCa) decreased below 0.40 mmol/L after the filter, whereas post-filter calcium reinjection according to ionic dialysance led to a stable pre-filter (i.e. patient) iCa. Median citrate concentrations were all below 0.8 mmol/L after the filter (minimal concentration to obtain anticoagulation 3 mmol/L) and all except one below the normal value (< 125 µmol/L) before the filter. During all the sessions, ionized to total calcium ratio was below 2.1 and the strong ionized gap decreased. When available (n = 7), no correlation could be identified between serum citrate concentration and liver clearance. Last, in 8 CCF-IHD sessions performed in critically ill patients, no premature termination occurred (median time of the sessions 5 h) and post-filter iCa also decreased below 0.45 mmol/L.

Conclusion: No citrate accumulation could be identified in critically ill patients (even with liver disorders) and receiving extended dialysis sessions (5 h or more) using calcium-free citrate containing-IHD. Interestingly, we demonstrated that citrate is not required to obtain optimal regional anticoagulation (i.e. post-filter iCa < 0.45 mmol/L), and a citrate- and calcium-free dialysate could be a safe alternative.

Compliance with ethics regulations: Yes.

F-035 Assessment of diaphragmatic function in mechanically ventilated children using the neuromuscular efficiency index

Benjamin Crulli, Guillaume Emeriaud

Pediatric intensive care unit, Department of pediatrics, CHU Sainte Justine, Université de Montréal, Montreal, Canada

Correspondence: Benjamin Crulli (benjamin.crulli@umontreal.ca)

Ann. Intensive Care 2020, 10 (Suppl 1):F-035

Rationale: Ventilator induced diaphragmatic dysfunction is highly prevalent in adult critical care and associated with worse outcomes. Specificities in pediatric respiratory physiology suggest that critically ill children may be at high risk of developing this complication, but no study has described the evolution of diaphragmatic function in critically ill children undergoing mechanical ventilation. This study aims to validate a method to quantify diaphragmatic function in mechanically ventilated children.

Patients and methods: In this prospective single-center observational study, 10 children between 1 week and 18 years old intubated for elective ENT surgery and without pre-existing neuromuscular disease or recent muscle paralysis were recruited. Immediately after intubation, diaphragmatic function was evaluated using brief airway occlusion maneuvers during which airway pressure at the endotracheal tube (Paw) and electrical activity of the diaphragm (EAdi) were simultaneously measured for 5 consecutive spontaneous breaths, while the endotracheal tube was occluded with a specific valve. Occlusion maneuvers were repeated 3 times. In order to account for central respiratory drive and sedation use, we recorded the neuromechanical efficiency ratio (NME, Paw/EAdi), in addition to the maximal inspiratory force (MIF). In order to determine the optimal measure of NME during an occlusion, the variability over the three occlusion maneuvers of different variables (first breath, last breath, breath with maximal Paw deflection, breath with maximal NME value, and median NME value) was assessed using coefficients of variation and repeatability coefficients.

Results: Patients had a median age of 4.9 years (interquartile range 3.9–5.5), a median weight of 18 kg (14–23), and 5 were male (50%). The median evolution of Paw, EAdi, and NME ratio over the 5 occluded breaths are represented on Fig. 1. NME values corresponding to the last breath and the breath with maximal Paw deflection were the least variable, with median coefficient of variation of 23% and 23% and repeatability coefficients of 3.44 and 3.44, respectively.

Conclusion: Brief airway occlusions can be used to assess diaphragmatic function in intubated children through both MIF and NME ratio, and the latter should ideally be computed on the last breath or the breath with the largest pressure deflection to improve repeatability and decrease variation.

Compliance with ethics regulations: Yes.

Fig. 1

Median (interquartile) evolution of Paw, eadi, and NME ratio over the 5 breaths of occlusion maneuvers

F-036 Efficacy and safety of dexmedetomidine as sole sedation for children undergoing MRI in comparison to general anesthesia: a single-center retrospective study (DEX-IRM)

Hélène Lepeltier¹, Arnaud Lepetit², Maxime Gauberti³, Clément Escalard³, Anne Lesage², David Brossier⁴, Isabelle Goyer⁵

¹CHU CAEN, Département de pédiatrie, Caen, France; ²CHU CAEN, Département d’anesthésie, Caen, France; ³CHU CAEN, Département de radiologie, Caen, France; ⁴CHU CAEN, Département de réanimation pédiatrique, Caen, France; ⁵CHU CAEN, Département de pharmacie, Caen, France

Correspondence: Hélène Lepeltier (helene.lepeltier@orange.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-036

Rationale: MRI requires prolonged motion-free periods. This may be difficult to obtain in the pediatric patient population. This need for prolonged immobility justifies the use of general anesthesia or procedural sedation in children. Recently, national drug safety agencies have warned physicians about the risk of neurotoxicity of conventional anesthetic agents targeting GABAergic and glutamatergic neurotransmission pathways in children. Dexmedetomidine, a sedative agent with works as an α2 pre-synaptic receptor agonist, does not act on these usual pathways. This molecule has raised great interest in the procedural sedation of children, especially since it preserves respiratory drive. The objective of this study was to evaluate the efficacy and safety of dexmedetomidine sedation in pediatric MRI compared to conventional general halogenated anesthesia.

Patients and methods: This was a retrospective monocentric cohort study performed in the University Hospital of Caen. This study included all patients under 18 years of age who received sedation for MRI between August 1st, 2018 and March 31st, 2019. Patients were retrospectively divided into 2 groups according to the performed sedation modality (DEX and GA).

Results: 78 patients were included (26 in the DEX group and 52 in the AG group). Dexmedetomidine sedation was significantly associated with a decrease in the use of invasive ventilation (p < 0.001) with no difference in image quality. The failure rate of sedation was 42% in the DEX group vs. 0% in the AG group (p < 0.001). None of the patients had any significant adverse reactions to dexmedetomidine.

Conclusion: Dexmedetomidine seems suitable for procedural sedation during MRI in children. It provides an alternative to halogenated general anaesthesia with the aim of limiting children’s exposure to conventional anaesthetic agents and the use of invasive ventilation.

Compliance with ethics regulations: Yes.

F-037 Incidence of withdrawal syndrome after sedation/analgesia in a surgical pediatric intensive care units using the Withdrawal Assessment Tool-1 (WAT-1) score

Pauline Ponsin¹, Guillaume Geslain², Chloé Tridon¹, Charline Riaud¹, Nicolas Robin¹, Alina Lazarescu¹, Gilles Orliaguet¹

¹Department of anaesthesia and intensive care, Assistance Publique-Hôpitaux de Paris, Necker-Enfants Malades Hospital, Paris, France; ²Pediatric Intensive Care Unit, Assistance Publique-Hôpitaux de Paris, Robert Debre University Hospital Université de Paris, Paris, France

Correspondence: Pauline Ponsin (pauline_2014@hotmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-037

Rationale: Withdrawal syndrome (WS) is a known side effect of prolonged sedation/analgesia in pediatric intensive care units (PICU). Epidemiology is poorly understood due to the rare use of validated diagnostic tools. The main objective of the study was to determine, by systematically calculating the WAT-1 score, the incidence of WS in our surgical PICU. The secondary objective was to analyze the risk factors, consequences and management modalities of WS.

Patients and methods: Following Institutional Review Board approval, we conducted a prospective monocentric study between July 2018 and January 2019. All consecutive mechanically ventilated children admitted in our surgical PICU with sedation/analgesia by continuous intra-venous (IV) benzodiazepines (BZD) and/or opioids for at least 48 h were included. As soon as sedation was decreased and during 72 h following their total discontinuation, WAT-1 score was assessed twice a day. WS was defined by a WAT-1 score > 3. The search for risk factors and consequences associated with WS was performed by univariate analysis (Mann–Whitney and Chi2 test). Ethical standards were satisfied and the lack of opposition from patients and their parents was systematically checked.

Results: The incidence of WS was 50% among the 46 patients of our cohort including 54% of children admitted postoperatively and 35% after severe traumatic brain injury (TBI). Significant results are reported in Table 1. Our results show that even for sedation time less than 5 days, children could develop WS (11/23 patients). On the other hand, age, severity (PELOD 2 score), number of previous surgeries and severe TBI were not associated with WS. Our study also demonstrated that cessation of sedation and prevention of WS was not uniform in our unit.

Conclusion: The high incidence of withdrawal syndrome in our study, even in children sedated for less than 5 days, and its consequences require thinking about prevention. We suggest a systematic monitoring of the occurrence of this adverse event using a validated score, from 3 days of continuous IV sedation/analgesia.

Compliance with ethics regulations: Yes.

Table 1 Risk factors and consequences of withdrawal syndrome

F-038 Automatic Real-time Classification of the Validity of Intracranial Pressure Signals Recorded by Ventricular Drain Using a Machine Learning Method

Sally Al Omar¹, Floriane Cannet², Gabriel Masson³, Philippe Jouvet¹, Guillaume Emeriaud¹

¹CHU Sainte-Justine, Montreal, Canada; ²Polytech Marseille, Marseille, France; ³University Hospital of Lille, Lille, France

Correspondence: Sally Al Omar (sally.alomar@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-038

Rationale: Severe traumatic brain injury (TBI) is a major healthcare problem. Amplitude and duration of intracranial hypertension is highly associated with patient outcome. The intracranial pressure (ICP) is therefore one key parameter to monitor in the acute phase. When ICP is monitored with an external ventricular drain, the pressure recorded by the monitor does not always correspond to the real ICP, depending on the status (open/closed) of the 3-way tap. Misleading values could therefore be sent to the patient medical record. Our hypothesis is that a machine-learning algorithm will be able to identify automatically and in real time the reliable and non-reliable values of the ICP signal.

Patients and methods: We retrospectively studied pediatric patients having an external ventricular drain between July 2018 and July 2019, in a single pediatric intensive care unit. The ICP signals were extracted from a high-frequency database (128 Hz) and pre-processed adequately. To train the algorithms, an annotated database was manually created with two classes: reliable ICP vs. non-reliable ICP (drain system opened to allow cerebrospinal fluid removal). Eleven signal characteristics were compared between the two classes (Mann–Whitney test), and significantly differing variables were tested in the algorithms. We compared the performance of two machine-learning algorithms: the K-Nearest Neighbors (KNN) and the Support Vector Machine (SVM). Using 10-fold cross-validation method, 75% of the data was used to train the algorithms and 25% was used for testing. The best classifier was further validated by simulating a real-time ICP analysis, using a 15 s sliding-window approach with 50% overlap. The study was approved by the local research ethics committee.

Results: Sixteen patients were included in the study. The training database created from 14 patients, contained 320 segments (of 15 s duration) per class and per patient. Eight signal variables were identified and kept to define the segments. The KNN algorithm, with k = 3, led to the best performance, with a mean of 98% (mean ± SD: 98% ± 0.29%). The KNN was then visually validated on ICP signals from the remaining two patients (Figure). By simulating a real-time ICP extraction, our algorithm was able to efficiently identify the reliable ICP segments, and to display a mean value only for valid segments.

Conclusion: The proposed machine learning algorithm can help identifying the validity of ICP values recorded using a ventricular drain in real time. After external validation, this algorithm could be implemented in future clinical decision support system to optimize the care of TBI patients.

Compliance with ethics regulations: Yes.

Fig. 1

Results of real-time ICP signal classification with the KNN algorithm using the sliding-window approach. Green: reliable ICP. Red: non-reliable ICP (ventricular drain system opened for fluid removal)

F-039 Epidemiology and prognosis of acute encephalitis in pediatric intensive care unit

Hélène Lienard¹, Jérôme Naudin¹, Florence Renaldo², Fleur Le Bourgeois¹, Anna Deho¹, Maryline Chomton¹, Géraldine Poncelet², Guillaume Geslain², Stéphane Dauger¹, Michael Levy¹

¹Pediatric Intensive Care Unit, Hôpital Robert-Debré, Assistance Publique Hôpitaux de Paris, Paris, France; ²Pediatric Neurology, Hôpital Robert-Debré, Assistance Publique Hôpitaux de Paris, Paris, France

Correspondence: Hélène Lienard (lienardhelene@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-039

Rationale: Acute encephalitis is a rare but potentially severe disease leading to an admission in Pediatric Intensive Care Units (PICU) in 18 to 40% of pediatric patients. However, there are few epidemiological data on these severe forms and little is now about their outcome. The main objective of this study was to describe the etiologies of acute encephalitis in PICU as well as their presentations. Secondary objectives included an evaluation of the outcome (mortality and sequalae) as well as related risk factors of poor outcome.

Patients and methods: A monocentric retrospective study was performed between January 2005 and December 2018 in Robert-Debré University Hospital PICU (Paris). All consecutive children (1 month–18 years) admitted for acute encephalitis were included and diagnosis was confirmed using the 2013 consensus conference criteria’s. Data regarding clinical, biological and radiological presentations were collected as well as data on the therapeutics used and outcomes at discharge and at the last medical consultation.

Results: 106 patients were included with a mean age of 6.2 years (range 0.1 to 17 years old). Infectious causes were identified in 45% (n = 48), autoimmune causes in 8% (n = 8) and acute demyelinating encephalomyelitis in 4% (n = 4) of cases. Etiology remained undetermined in 43% of cases (n = 46). The most common pathogens were, in order of frequency, Influenzae virus, Mycoplasma pneumoniae and Epstein-bar virus. The main clinical features were fever (88% n = 93); epileptic seizures (80% n = 85) and coma (46% n = 49). Regarding therapeutics, 52% of patients required mechanical ventilation and 20% of patients required hemodynamic support. 47% received corticosteroids, 17% intravenous immunoglobulins and 12% plasmatic exchanges. The use of these specific treatments was heterogeneous, especially in infectious and undetermined encephalitis, where respectively 48% and 38% received boluses of corticoids. The mean length of stay in PICU was 10.7 days (range 1–155 days). The mortality rate was 10% and the overall rate of sequelae at discharge was 76% and 61% at distance, with 21% considered as severe (GOSE-PED score > 5). The use of mechanical ventilation and young age at diagnosis were risk factors associated with poor prognosis at discharge.

Conclusion: The etiology of acute encephalitis remains indeterminate in more than 40% cases with a clear predominance of infectious causes when an etiology is found. This is a severe pathology responsible for significant mortality and morbidity requiring long-term follow-up.

Compliance with ethics regulations: Yes.

F-040 Prognostic value of cerebral oxymetry in critically ill children undergoing extracorporeal membrane oxygenation

Meryl Vedrenne-Cloquet, Raphaël Lévy, Judith Chareyre, Manoëlle Kossorotoff, Sylvain Renolleau, Marion Grimaud

Hôpital Necker-Enfants Malades, Paris, France

Correspondence: Meryl Vedrenne-Cloquet (meryl_vedrenne@yahoo.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-040

Rationale: Preserving neurological outcome of children under Extracorporeal Membrane Oxygenation (ECMO) remains challenging. Acute Brain Injury (ABI) is a frequent complication of ECMO that could be prevented by continuous neuromonitoring. Cerebral Near InfraRed Spectroscopy (NIRS) is routinely used for detecting cerebral complications of cardiac surgery. In adults and infants under prolonged ECMO, cerebral hypoxia is associated with poor neurological outcome. The aim of this study was to assess the value of an impaired cerebral oxygenation on mortality and occurrence of an ABI in children under ECMO.

Patients and methods: Children under 18 years old were included in this observational retrospective monocentric study if they needed veno-venous (V–V) or veno-arterial (V-A) ECMO for respiratory and/or circulatory failure and had concomittant NIRS monitoring. Cerebral desaturation was defined as a rScO₂ value under 50% or under 20% from the baseline; cerebral hyperoxia was defined as a rScO₂ value above 80%. Proportion of time in cerebral desaturation and hyperoxia were recorded. Neurological lesions were identified on imaging (MRI or scan) by blinded radiologist and classified as major or minor. ABI was defined as any hemorragic or ischemic lesion on cerebral imaging, including brain death.

Results: 63 patients were included. ECMO duration was 9 [5; 13] days. The mortality rate was 32 (50.8%), and the proportion of ABI was 34 (54%) including 5 brain deaths, 10 (15.9%) major lesions, and 19 (30.2%) minor lesions. Mean rScO₂ was 73 ± 9% in the right hemisphere, and 75 ± 9% in the left hemisphere. There was no significant difference in cerebral hypoxia between survivors and non survivors, and between patients with and without an ABI. Cerebral hyperoxia was associated with a better survival (p = 0.03 in the right hemisphere, and p = 0.02 in the left hemisphere). In V–V ECMO and at the right hemisphere, proportion of patients in hyperoxia was higher in survivors (78 [72; 81.8]% versus 58 [56; 70]%, p = 0.04); proportion of time in hyperoxia was also more important (42 [12; 57] % in survivors versus 4.5 [0; 23.3]% in non survivors, p = 0.049).

Conclusion: In our study, cerebral hypoxia was not associated with poor neurological outcome, but cerebral hyperoxia seems to be protective especially in V–V ECMO. This is the first study assessing the value of cerebral oxymetry in all age ranges pediatric ECMO. In this population, multimodal monitoring might be better than NIRS alone to predict neurological impairment. Further prospective studies are needed to assess first the feasibility, then the impact of such a monitoring.

Compliance with ethics regulations: Yes.

F-041 Cerebral autoregulation impairment is associated with acute neurological events during pediatric extracorporeal membrane oxygenation

Nicolas Joram¹, Erta Beqiri², Stefano Pezzato³, Pierre Bourgoin¹, Alexis Chenouard¹, Jean-Michel Liet¹, Marek Czosnyka², Pierre-Louis Leger⁴, Peter Smielewski²

¹Pediatric Intensive Care Unit, University hospital of Nantes, Nantes, France; ²Division of academic Neurosurgery, Department of clinical neurosciences, Addensbrooke ‘s Hospital, University of Cambridge, Cambridge, UK; ³Pediatric Intensive Care Unit, IRCCS Giannina Gaslini Institute, Genoa, Italy; ⁴Pediatric Intensive Care Unit, Trousseau University Hospital, Paris, France

Correspondence: Nicolas Joram (nicolas.joram@chu-nantes.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-041

Rationale: Children supported by extracorporeal membrane oxygenation (ECMO) present a high risk of adverse neurological complications. As some animal studies have shown, cerebral autoregulation (CA) impairment after exposure to ECMO, may be a key factor. Our main objective was to investigate the feasibility of CA continuous monitoring during ECMO treatment. The second objective was to analyze the relationship between CA impairment and neurological outcome.

Patients and methods: An observational prospective study including children treated by ECMO in 2 centers was conducted. A correlation coefficient between the variations of regional cerebral oxygen saturation (rScO₂) and the variations of mean arterial blood pressure (MAP) was calculated as an index of CA (cerebral oxygenation reactivity index, COx) during ECMO. A COx > 0.3 was considered as indicative for dysautoregulation. COx values were averaged inside 2 mmHg-MAP bins, allowing determining optimal MAP (MAPopt) and lower (LLA) and upper (ULA) limits of autoregulation in 8-h periods. Neurological outcome was assessed by the onset of an acute neurologic event (ANE) defined by occurrence of hemorrhagic or ischemic stroke and/or clinical or electrical seizure and/or brain death during the ECMO treatment.

Results: Twenty-nine patients (31 ECMO runs) treated by veno-arterial (VA, n = 23) or veno-venous ECMO (VV, n = 6) were included (median age 71 days, weight 4.6 kg). COx was always available in all patients and MAPopt, LLA and ULA in 89.8% of time. CA variables were similar during VA and VV ECMO runs. Among children who presented ANE (15/29, 51.8%) versus others, the median time spent with a COx > 0.3 was significantly higher (33% (23.5–64.2) vs 20% (16.6–23.7), p < 0.001). These patients spent also significantly more time with MAP below LLA (10.4% (6.1–26.8) vs 5.2% (1.8–9.2), p = 0.03). After adjustment on the onset of cardiac arrest before or during cannulation and the type of ECMO (VA vs VV), percentage of time spent with a Cox > 0.3 higher than 25% was independently associated with ANE (aOR 8.03, IC 95% 1.02–63.12, p = 0.04).

Conclusion: CA assessment seems to be feasible in pediatric ECMO. The impact of the time below autoregulation threshold on neurological outcome is significant. However, the underlying mechanisms of CA impairment during ECMO need to be explored further.

Compliance with ethics regulations: Yes.

Fig. 3

Autoregulation parameters according to neurological outcome. A. Mean COx. B. Percentage of time with COx > 0.3. C. Percentage of time with MAP < LLA. D. Percentage of time with MAP > ULA. E. Percentage of time with MAP—MAPopt between − 5 and + 5 mmHg. AU, Area Under; MAP, Mean Arterial Pressure; LLA, Lower Limit of Autoregulation; ULA, Upper Limit of Autoregulation; ΔMAPopt = median MAP-MAPopt. Data are shown in box-and-whisker plots, indicating the median, interquartile range, and range. The Mann–Whitney test was used to compare the 2 groups of infants

F-042 Protective mechanisms of pharmacologic preconditioning against myocardial ischemia reperfusion injury: role of Bcl-2 family proteins

Romain Rozier¹, Rachel Paul-Bellon¹, Elodie Villa², Amélie Rolle³, Marc Aimé Raucoules⁴, Jean-Ehrland Ricci¹, Michel Carles³

¹Université Côte d’Azur-CHU Nice-C3M INSERM U1065 Equipe 3, Nice, France; ²Feinberg School of Medicine, Northwestern University, IL, Chicago, USA; ³CHU Point-à-Pitre, Point-À-Pitre, France; ⁴CHU Nice-Université Côte d’Azur, Nice, France

Correspondence: Romain Rozier (romainrozier13@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-042

Rationale: Myocardial ischemia reperfusion (IR) injury is the leading cause of perioperative morbi-mortality. Protective effect of pharmacologic preconditioning such as anesthetic preconditioning (APC) with sevoflurane (SEV) has been widely demonstrated in animal and human models. APC seems to protect myocardial cells from apoptosis, a programmed process of cell death tightly controlled by Bcl-2 family proteins. However, the involved mechanisms in APC have yet to be characterized. We hypothesized that APC protects against myocardial apoptotic cell death by regulating Bcl-2 anti-apoptotic members.

Patients and methods: To study the SEV-induced APC mechanisms against myocardial IR, we used a validated in vitro model reproducing IR injury. Rat cardiomyoblast cells H9c2 were cultivated in 0.1% O2 hypoxia in the presence of ischemia-mimicking medium. After 90 min of ischemia, the reperfusion injuries are induced by replacing the culture medium with a Krebs–Henseleit normoxic medium for 60 min. APC was performed by adding SEV directly into the culture medium at an initial concentration of 20 mM, prior to ischemia, for 60 min. We then used another preconditioning agent, metformin (MET), to explore the same signaling pathways. Apoptotic cell death was measured by caspase activity assay and western blotting (expression of cleaved caspase 3) under IR and APC conditions.

Results: Our model faithfully reproduced the protective effect of APC which results in a significant decreased apoptosis under IR (50% reduction of the Caspase 3 enzymatic activity, correlated with a decrease of Caspase 3 cleavage). We showed that SEV induces overexpression of the anti-apoptotic protein Bcl-xL, which is responsible for the protective effect of APC. Furthermore, these observations were confirmed in vivo in mouse heart lysates. We demonstrated that Bcl-xL overexpression was due to the activation of the protein kinase Akt. Interestingly, we were able to show that preconditioning with MET reproduces the protective effect of SEV by inducing an Akt-dependent Bcl-xL overexpression. Indeed, SEV and MET, which are both complex 1 inhibitors of mitochondrial respiratory chain, seem to share a common Reactive Oxygenated Species-dependent protective mechanism responsible for Bcl-xL protein regulation.

Conclusion: Our results elucidates the molecular mechanisms by which SEV induces APC against IR injuries, i.e. the role of Bcl-xL (Bcl2 family protein). Moreover, this study shows that pharmacologic preconditioning with MET promotes similar protective effect, sharing with SEV the same signaling pathways. Altogether, our results could be of interest to improve the perioperative management of patients at risk of ischemia reperfusion events, such as patients with a high cardiovascular risk.

Compliance with ethics regulations: Yes.

F-043 Early sepsis worsens outcome in experimental ischemic stroke despite recanalization

Thomas Rambaud¹, Lucas Di Meglio¹, Adrien Cogo¹, Veronique Ollivier¹, Mialitiana Solo Nomenjanahary¹, Sébastien Dupont¹, Yacine Boulaftali¹, Nathalie Kubis², Benoit Ho Tin Noe¹, Mikael Mazighi³, Romain Sonneville⁴

¹INSERM U 1148 laboratory for vascular translationnal science, Paris, France; ²INSERM U 1148 laboratory for vascular translationnal science/Service de Physiologie Clinique-Explorations Fonctionnelles, AP-HP, Hôpital Lariboisière, Paris, France; ³INSERM U 1148 laboratory for vascular translationnal science/Department of Interventional Neuroradiology, Rothschild Foundation Hospital, Paris, France; ⁴INSERM U 1148 laboratory for vascular translationnal science/medical and infectious intensive care unit, Bichat hospital, aphp, Paris, France

Correspondence: Thomas Rambaud (rambaud.t@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-043

Rationale: Despite early endovascular treatment with successful recanalization, 50% of acute ischemic stroke (AIS) patients experience a poor functional outcome after a large vessel occlusion. Sepsis is frequent at the acute phase of stroke and is associated with poorer short and long term outcomes. We aimed to investigate the cerebral consequences of sepsis after recanalized AIS and explore possible mechanisms involved.

Patients and methods: Male C57Bl6 mice were randomly assigned to a 2x2 factorial plan to one of the 4 following groups: 1) a 60-minute middle cerebral artery (t-MCAO) transient occlusion under inhaled general anesthesia, followed 5 min after recanalization by intraperitoneal (I.P.) sepsis (LPS, 10 µg/g diluted in 100µL of NaCl0.9%), (tMCAO/LPS group); 2) t-MCAO followed by I.P. placebo (100µL of NaCl0.9%) (tMCAO/placebo group); 3) Sham operation (cervicotomy without carotid catheterization) followed by I.P. LPS. (Sham/LPS group); 4) Sham operation followed by I.P. placebo, (Sham/placebo group). In all groups, animals received subcutaneous fluid resuscitation (200µL NaCl 0.9%) immediately after the procedure and 1 h later. Twenty-four hours after recanalization, animals were scored for sepsis features and neurological deficit (on the modified neurological severity scale), (mNSS) before sacrifice. The primary outcome measurement was a composite of death and hemorrhagic transformation at 24 h. Secondary outcome measurements included neurological deficit, sepsis features, neutrophil activation reflected by plasmatic myeloperoxydase (MPO) levels, stroke volume, and microglial activation in brain parenchyma (infarct core, perilesional area, controlateral hemisphere).

Results: t-MCAO/LPS animals had higher mNSS (1.7 fold, p = 0.02) and sepsis (6 fold, p = 0.0018) scores at 24 h with increased plasma MPO levels at 1 h (2.8 fold, p < 0.01) and 24 h (5.9 fold, p < 0.0001), as well as, lower temperature (3.0 °C reduction, p = 0.01) and glycemia (0.9 g/l reduction, p = 0.01) as compared to tMCAO/placebo animals. t-MCAO/LPS animals had a higher risk of unfavorable outcome at 24 h (4-group comparison: p = 0.03; 2x2 analysis: t-MCAO/LPS, 6/12 − 50%- vs. t-MCAO/placebo 1/17–6%-, p < 0.01), whereas stroke volumes were not significantly different between groups. Detailed results are presented in Table 1. Compared to t-MCAO/placebo group, t-MCAO/LPS animals had 1.3 fold increase (p = 0.05) in the mean number of microglial cells in the hemisphere controlateral to t-MCAO, whereas no significant difference was observed in infarct core or peri-infarct parenchyma.

Conclusion: Early sepsis after experimental AIS worsens outcome and neurological deficit, without impacting stroke volume. Early sepsis-induced systemic activation of neutrophils and increased microglial activation in the hemisphere contralateral to ischemia may have an important role on neurological outcomes observed in this setting.

Compliance with ethics regulations: Yes.

Table 1 detailed results of primary outcome measurement

F-044 Extracellular vesicles as a marker of lung injury after cardiac surgery with cardiopulmonary bypass

William Allali¹, Erwan Dumontet, Murielle Gregoire, Mathieu Lesouhaitier, Arnaud Gacouin, Sebastien Biedermann, Erwan Flecher, Karin Tarte, Jean-Marc Tadié

¹CHU Rennes, Rennes, France

Correspondence: William Allali (William.ALLALI@chu-rennes.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-044

Rationale: Extracellular vesicles (EVs) regulate diverse cellular and biological processes via facilitating intercellular cross-talk. Several studies have suggested an association between lung injury and the generation of EVs derived from platelets, neutrophils, monocytes, lymphocytes, red blood cells, endothelial cells, and epithelial cells. Every year more than 25,000 patients require cardiac surgery with cardiopulmonary bypass (CPB). This CPB allows a substitution of the heart pump function and an oxygenation of the blood permitting a stop of the mechanical ventilation (MV). Stopping MV during CPB is responsible for lung damage, leading to postoperative systemic inflammation while maintaining MV with positive expiratory pressure (PEEP) diminished the occurrence of atelectasis and the postoperative inflammatory response. In addition, this surgery is marked by immune dysfunction, leading to real immunosuppression of patients in postoperative care. A link between pulmonary injury and postoperative immunosuppression has been established, however, the mechanisms underlying this association are not fully known and EVs may have a role in this post-operative immunosuppression. The purpose of this study is to investigate whether lung injury induced during cardiac surgery with CPB lead to the emergence of EVs. The effect of MV during CPB on the production of these EVs has also been studied.

Patients and methods: Patients were prospectively divided into two groups: without MV during CPB and dead space MV with positive end-expiratory pressure during CPB. PaO₂ (arterial oxygen tension)/FIO₂ (inspired oxygen fraction) ratio, biological markers of lung injury (CXCL10, CCL2, TNF-α, IL-1β, IL-10, RAGE, IL-8) and blood cell count were collected before, 24 h and 7 days after surgery. The quantification of plasma EVs was performed using Turnable Resistive Pulse Sensing and characterization of EVS was performed using flow cytometry before, 24 h and 7 days after surgery.

Results: After surgery, patients presented a significant decreased in plasmatic EVs levels. Patients with MV during CPB had a concentration of EVs significantly closer to normal 7 days after surgery than patients without MV. This suggests the existence of a sub-population of EVs with a protective role. In a second set of experiments using flow cytometry, we found that MV during CPB increased EVs derived from platelets compared to patients without MV during CPB.

Conclusion: Since high levels of circulating EVs have been associated with better outcome in ARDS, our data suggest that a significant decrease in plasmatic EVs after cardiac surgery with CPB could be a reliable marker of lung injury that could be prevented maintaining MV during CPB.

Compliance with ethics regulations: Yes.

F-045 Prone positioning decreases macrophage recruitment in ventral lung regions of animals with experimental acute respiratory distress syndrome

Laurent Bitker¹, Nicolas Costes², Didier Le Bars², Maciej Orkisz³, Mehdi Mezidi¹, Nazim Benzerdjeb⁴, Jean Christophe Richard¹

¹Médecine Intensive-Réanimation, hôpital de la Croix Rousse, Hospices Civils de Lyon, Lyon, France; ²CERMEP-Imagerie du vivant, Lyon, France; ³CREATIS, Lyon, France; ⁴Service d’anatomo-pathologie, Centre Hospitalier Lyon-Sud, Hospices Civils de Lyon, Lyon, France

Correspondence: Laurent Bitker (laurent.bitker@chu-lyon.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-045

Rationale: The benefit of prone positioning (PP) during moderate to severe acute respiratory distress syndrome (ARDS) may be related to its impact on the inflammatory response to ventilator-induced lung injuries. [11C]-PK11195 is a positron emission tomography (PET) radiotracer that allows the non-invasive quantification of macrophages. We aimed to evaluate the effects of PP on [11C]-PK11195 lung uptake in animals with experimental ARDS.

Patients and methods: Experimental ARDS (by hydrochloric acid) was induced in 10 pigs in supine position (SP), to obtain a PaO₂/FiO₂ < 300 mmHg. Animals were under general anesthesia, neuromuscular blockade, and ventilated with a 6 ml kg⁻¹ tidal volume, and 5 cmH₂O of positive end-expiratory pressure (PEEP). Immediately after experimental ARDS, animals were randomized to be prone positioned, or to remain in SP. PET and computerized tomography (CT) were acquired 4 h after randomization (H4). [11C]-PK11195 uptake was measured on the whole lungs, and by dividing the lungs into 8 regions or slices-of-interest (SOI) along the ventro-dorsal axis, and was quantified by the standardized uptake value (SUV), corrected for lung tissue density.

Results: PP was performed in 6 animals, and SP in 4. After ARDS induction, PaO₂/FiO₂ was 107 [IQR, 89–163] mmHg, and the elastance-derived transpulmonary pressure (∆PL) increased from 8 [7–10] to 17 [14–18] cmH₂O (P < 0.01). At H4, PP animals had lower ∆PL compared to SP animals (13 [12–13] vs. 20 [17–23] cmH₂O, p = 0.02). CT analysis showed an increase of poorly- and non-aerated lung volumes (− 500 to + 100 Hounsfield units) in ventral SOI of PP animals, compared to SP (most ventral SOI: 16 [8–17] vs. 3 [3–4]% of total SOI volume), but a decrease in dorsal SOI (most dorsal SOI: 76 [65–85] vs. 99 [99–99]%, p < 0.01 between groups on all 8 SOI). At H4, [11C]-PK11195 uptake of whole lungs of PP animals was 1.4 [1.2–1.4], and 1.4 [1.2–1.7] in SP animals (p = 0.61). In PP animals, [11C]-PK11195 SUV was significantly lower in ventral SOI, compared to SP, and significantly increased in dorsal SOI (Fig. 1, *: p < 0.01 between groups in a given SOI). In univariate analysis, [11C]-PK11195 regional SUV was positively associated with regional CT-measured PEEP-related increase in gas volume, and negatively with PEEP-related lung recruitment, but not with regional tidal volume.

Conclusion: During experimental ARDS, PP redistributed lung macrophage recruitment estimated by [11C]-PK11195 uptake from ventral lung regions to dorsal regions, without affecting global macrophage influx. The intensity of macrophage recruitment was associated with PEEP-related lung inflation.

Compliance with ethics regulations: Yes.

Fig. 1

Effect of PP on regional macrophage recruitment during experimental ARDS

F-046 Plasma exosomes as promising biomarkers for Acute Respiratory Distress Syndrome

Gilles Parzibut, Julien Guiot

CHU-Sart Tilman, Liège, Belgium

Correspondence: Gilles Parzibut (g.parzibut@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-046

Rationale: Acute respiratory distress syndrome (ARDS) is a pleiomorphic disease characterized by a severe respiratory failure associated with an increased mortality. Nowadays, predicting clinical outcome of patients suffering from ARDS remains difficult. Therefore, identifying new biomarkers to predict patient outcome, to evaluate response to therapy and to identify new potential pathways of interest are highly needed. Exosomes are extracellular vesicles involved in cell–cell communication by transferring microRNAs (miRNAs) from donor to recipient cells. Thus, exosomal miRNAs can significantly affect biological pathways within recipient cells resulting in alterations of cellular function and the development of a pathological state. As biomarkers are highly needed in the particular field of ARDS, we realized a monocentric and prospective study to identify a new potential biomarker of interest. Therefore, a prospective plasma sampling at the diagnosis of moderate to severe ARDS according to the definition of “Berlin” has been performed. We analysed miRNA content of exosomes from plasma ARDS patients compared to healthy subjects (HS) in order to identify new potential predictive biomarkers in ARDS.

Patients and methods: During one-year period, patients hospitalized in the ICU of CHU Sart Tilman suffering from infectious moderate-to-severe ARDS have been included. The ethical committee review boards of the hospital approved the research protocol (B707201422832, ref: 2014/302), and informed consents were obtained. Exosomes were isolated from plasma samples of 10 ARDS patients and 10 HS with standard ultracentrifugation protocol. Exosomal miRNA content was analyzed using small RNA sequencing method, and diseases/biological processes associated to altered miRs were determined by bioinformatic analysis.

Results: For the first time, exosomal miRNA expression modifications were studied in patients with moderate-to-severe infectious ARDS. We identified a new signature statistically significant composed of three up-regulated miRNAs (miR-122, miR-23a and miR-126) and one down-regulated (miR-Let-7b).

Conclusion: We identified potential biomarkers for ARDS from plasma exosomes. Our findings may thus lead to predict ARDS outcome but also a better understanding about the roles of these miRs in the pathogenesis of ARDS and thus open new avenues for therapeutic approaches. In particular, exploit and develop the pro-fibrotic pathway induced by down-expression of miR-Let-7b. But also confirm in the future the current interest about miR-126 in its ability to restore pulmonary integrity after trauma.

Compliance with ethics regulations: Yes.

Fig. 1

Example of heat map about expression of miRNAs in selection of ARDS population (8) vs healthy subjects (8)

F-047 Assessment of red blood cell deformability in patients with diabetic ketoacidosis

Dorian Leroy¹, Bruno Sirault², Patrick Biston¹, Karim Zouaoui Boudjletia³, Michael Piagnerelli¹

¹Intensive Care. CHU-Charleroi. Université Libre de Bruxelles., 6042-Charleroi, Belgium; ²Internal Medicine. CHU-Charleroi. Université Libre de Bruxelles., 6042-Charleroi, Belgium; ³Experimental Medicine Laboratory. CHU-Charleroi. Université Libre de Bruxelles., 6110-Montigny-Le-Tilleul, Belgium

Correspondence: Dorian Leroy (Dorian.Leroy@ulb.ac.be)

Ann. Intensive Care 2020, 10 (Suppl 1):F-047

Rationale: Diabetic ketoacidosis (DKA) is a life-threatening emergency. Microvascular hyporeactivity was reported in these patients and was completely reversibly when pH was corrected with treatment: aggressive rehydration, electrolyte replacement and insulin therapy (1). Red blood cell (RBC), a component of the microcirculation, showed alterations of their shape in diabetic patients (2) but no data were available concerning the time course of the RBC deformability during treatment for DKA. We aimed to assess the RBC deformability during DKA treatment in ICU patients.

Patients and methods: After approval by the ethics committee, RBCs deformability was assessed, in all ICU patients admitted for DKA and without infection, by ektacytometry technique (Laser-assisted Optical Rotational Red Cell Analyzer—LORRCA): at ICU admission, +8 h, +24 h and at the end of the ICU stay (36–48 h). Elongation index (EI) was defined as (L − W)/(L + W), where L is the length and W is the width. At 37 °C, EI values were determined in the function of shear stress (SS) in a range of 0.5–50 Pa, based upon the laser diffraction pattern changes. A higher EI indicates greater RBC deformation. RBC deformability from patients with DKA was compared at ICU admission to healthy volunteers (V) and to diabetic patients followed in consultation (D). We also studied the evolution of deformability during treatment.

Results: 15 ICU DKA patients compared to 31 D and 20 V were studied. As expected, glycemia and glycated hemoglobin were significantly higher in DKA compared to D (respectively: glycemia: 549 (444–872) vs 139 (92–181) mg/dL and 11.7% (9.4–13.4) vs 7.8 (7.3–8.2); all p < 0.001). DKA patients received 6279 (5482–8308) ml of fluids and 2.24 UI/Kg BW (1.71–2.42) of insulin during their first 24 h of ICU stay. RBCs deformability from DKA patients was significantly more altered at ICU admission compared to others groups (Fig. 1) and these alterations persists despite treatment. No correlations were observed between these alterations and quantity of fluids or insulin received, glycemia, glycated hemoglobin, pH, natremia, age or length of diabetes history.

Conclusion: In contrast of reversible microvascular hyporeactivity, RBC deformability from DKA patients was already altered at ICU admission and remains altered despite treatment. These alterations could contribute to the blood flow abnormalities observed in these patients.

Compliance with ethics regulations: Yes.

Fig. 1

RBC deformability in DKA, diabetic patients and volunteers. *p < 0.001 versus D and V. SD were not shown

F-048 Description of the early hemodynamic profile of septic patients assessed by focused echocardiography in the Emergency Department

Thomas Lafon¹, Pauline Feydeau², Alexandra Appert-Decle³, Mathilde Hadj³, Vincent Bigrat², Vincent Legarcon³, Marine Goudelin⁴, Bruno Evrard⁴, Ana Catalina Hernandez Padilla⁵, Arthur Baisse³, Philippe Vignon⁶

¹Service d’Accueil des Urgences/Inserm CIC 1435, CHU Dupuytren, Limoges, France; ²SAMU/SMUR, CHU Dupuytren, Limoges, France; ³Service d’Accueil des Urgences, CHU Dupuytren, Limoges, France; ⁴Réanimation polyvalente, CHU Dupuytren, Limoges, France; ⁵Inserm CIC 1435, CHU Dupuytren, Limoges, France; ⁶Réanimation polyvalente/Inserm CIC 1435/Inserm UMR 1092, CHU Dupuytren/Université de Limoges, Limoges, France

Correspondence: Thomas Lafon (thomas.lafon@chu-limoges.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-048

Rationale: Sepsis remains the first cause of acute circulatory failure in the Emergency Department (ED). Standardized fluid resuscitation may not be adapted in certain patients, especially those with early sepsis-induced cardiac dysfunction in whom excessive fluid administration could be deleterious. Information on early hemodynamic profile of septic patients in the ED are scarce. Accordingly, we aimed at describing hemodynamic profiles encountered in septic patients assessed shortly after their ED admission using focused echocardiography.

Patients and methods: We prospectively enrolled adult patients with sepsis (qSOFA score ≥ 2) from January 2017 to July 2019 in the ED (NCT02974790). Focused echocardiography were performed by emergency physicians previously trained to ECMU 1 level. Each patient was evaluated according to a standardized protocol based on a limited number of simple binary clinical questions. Investigators interpreted on-line the echocardiographic examination, determined the hemodynamic profile based on simple yet robust criteria (hypovolemia, left ventricular [LV] or right ventricular [RV] failure, vasoplegia with hyperdynamic state, tamponade, severe mitral or aortic regurgitation, or apparently normal profile), and recorded any substantial change in planned therapeutic management (Surviving Sepsis Campaign 2016). Data were digitally stored and validated off-line by an expert in critical care echocardiography.

Results: Focused echocardiography were performed in 81 patients (mean age: 70 ± 15 years; men: 58%; source of infection: pulmonary 37%, urinary 25%, abdominal 25%) after a median fluid loading of 500 mL (IQR: 500–1500 mL). According to Sepsis-3 definition, 44 patients had sepsis and 37 sustained septic shock. Mean SOFA score was 5.3 ± 2.9 (hemodynamic failure 85%, respiratory failure 72%, renal failure 67%), mean lactate reached 4.7 ± 4.3 mmol/L, ICU admission involved 38% of patients and overall 28-day mortality reached 35%. Hemodynamic profile was hypovolemia in 54 patients (67%), vasoplegia in 26 patients (32%), cardiac failure in 21 patients (26%) (LV failure: n = 15; RV failure: n = 6) and without relevant hemodynamic abnormality in 9 patients (11%). Ongoing therapy was altered based on early echocardiographic assessment in 26% of cases. Mortality rate was not significantly different between groups (p = 0.46).

Conclusion: Although hypovolemia was predominantly identified in patients presenting to the ED with sepsis during hemodynamic assessment, early ventricular dysfunction involved one-quarter of patients. These results suggest that early focused echocardiographic assessment promises to help the front-line physician tailoring the therapeutic management of septic patients in ED, especially regarding fluid resuscitation.

Compliance with ethics regulations: Yes.

F-049 Right ventricular failure in septic shock Characterization, incidence and impact on fluid-responsiveness

Guillaume Geri¹, Amélie Prigent², Xavier Repessé², Marine Goudelin³, Gwenael Prat⁴, Bruno Evrard³, Cyril Charron¹, Philippe Vignon³, Antoine Vieillard-Baron¹

¹Ambroise Paré Hospital, Boulogne-Billancourt, France; ²Ambroise Paré Hospital, medical ICU, APHP, Boulogne-Billancourt, France; ³CHU Limoges, Limoges, France; ⁴CHU Brest, Brest, France

Correspondence: Guillaume Geri (guillaume.geri@aphp.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-049

Rationale: Right ventricular (RV) failure was defined by RV dilatation with systemic congestion. Tricuspid annular plane systolic excursion (TAPSE) could be of limited value. We report the incidence of RV failure in patients with septic shock, its potential impact on the response to fluids, as well as TAPSE values.

Patients and methods: Ancillary study of the HEMOPRED prospective multicenter study including patients under mechanical ventilation with circulatory failure. 282 with septic shock were analyzed. Patients were classified in 3 groups based on central venous pressure (CVP) and RV size (RV/LV end-diastolic area, EDA). In group 1, patients had no RV dilatation (RV/LVEDA < 0.6). In group 2, patients had RV dilatation (RV/LVEDA ≥ 0.6) with a CVP < 8 mmHg (no venous congestion). RV failure was defined in group 3 by RV dilatation and a CVP ≥ 8 mmHg. Passive leg raising (PLR) was performed.

Results: 41% of patients were in group 1, 17% in group 2 and 42% in group 3. In group 2 and 3, RV/LV EDA was higher than in group 1, 0.7 [0.7; 0.9] versus 0.5 [0.4; 0.5]. CVP was 12 [10; 14.5] mmHg in group 3. A correlation between RV size and CVP was only observed in group 3. Higher RV size was associated with a lower response to PLR (Figure). A large overlap of TAPSE values was observed between the 3 groups. 36.5% of patients with RV failure had an abnormal TAPSE.

Conclusion: RV failure is frequent in septic shock and alters fluid responsiveness. TAPSE was not accurate enough to diagnose RV failure.

Compliance with ethics regulations: Yes.

Fig. 1

.

F-050 Weaning failure of cardiac origin in high-risk patients: role of the fluid balance

Marine Goudelin¹, Pauline Champy², Jean-Bernard Amiel¹, Bruno Evrard¹, Anne-Laure Fedou¹, Thomas Daix³, Bruno François³, Philippe Vignon³

¹Réanimation polyvalente, CHU Dupuytren, Limoges, France; ²Anesthésie secteur digestif et urologie, CHU Dupuytren, Limoges, France; ³Réanimation polyvalente/Inserm CIC 1435/Inserm UMR 1092, CHU Dupuytren, Limoges, France

Correspondence: Marine Goudelin (marine.goudelin@chu-limoges.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-050

Rationale: Weaning-induced pulmonary oedema (WIPO) is a leading cause of weaning failure in high-risk patients (heart failure, COPD, obesity). We hypothesized that hypervolemia associated with positive fluid balance facilitates WIPO in high-risk patients.

Patients and methods: In this prospective, observational, single-center study, patients with COPD and/or heart failure with reduced ejection fraction (< 40%) were studied. Exclusion criteria were non-sinus rhythm, severe mitral valve disease and inability to obtain adequate echocardiographic views. Echocardiography was performed immediately before and during spontaneous breathing trial (SBT, 30-min T-tube). Patients who failed SBT were treated according to echocardiographic results before undergoing a second SBT. Fluid balance and body weight were collected at each SBT.

Results: Of 304 eligible patients, 114 died before weaning, 114 had non-sinus rhythm, 8 had severe mitral valve disease, 5 had no usable echocardiographic image and 4 underwent self-extubation. Finally, 59 patients were studied (39 men; median age: 62 years [52–73]; SAPS2: 38.5 [30–53]; SOFA: 6 [4–9]; 75% heart failure and 36% COPD), and 12 of them (20%) developed WIPO. When compared to patients who succeeded, body weight (+ 0.75 kg [− 2.95; + 5.57] vs − 2.5 kg [− 4.8; − 1.0]: p = 0.02) and cumulative fluid balance (+ 143 ml [− 2654; + 4434] vs − 2326 ml [− 3715; + 863]: p = 0.007) were increased relative to admission. Doppler mitral profile depicted a higher E wave maximal velocity (101 cm/s [84–144] vs 83 cm/s [62–91]: p = 0.046) and E/A ratio (1.65 [1.28–2.38] vs 0.87 [0.69–1.06]: p = 0.001), and shorter E wave deceleration time (107 ms [94–129] vs 173 ms [128–213]: p = 0.002). After echocardiography-guided treatment, all patients who failed the first SBT could be successfully extubated after a second conclusive SBT. Fluid balance was then negative (− 2224 ml [− 7056; + 100] vs + 146 ml [− 2654; + 4434]: p = 0.005) and left ventricular filling pressures were lower than that measured during the first SBT (E wave velocity: 84 cm/s [67; 127] vs 102 cm/s [85; 149]: p = 0.017; E/E’ ratio: 7.3 [5.0; 10.4] vs 8.9 [5.9; 13.1]: p = 0.028) (Table 1).

Conclusion: In high-risk patients, WIPO is related to hypervolemia and excessive fluid balance. Echocardiography-guided treatment allows further successful weaning.

Compliance with ethics regulations: NA.

Table 1 Sub-group of patients who failed the first SBT

F-051 Measurement site accuracy of inferior vena cava diameters to predict fluid responsiveness in spontaneously breathing patients

Morgan Caplan, Julien Goutay, Perrine Bortolotti, Thierry Onimus, Raphael Favory, Sébastien Préau

CHRU Lille, Lille, France

Correspondence: Morgan Caplan (Morgan.Caplan@gmail.Com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-051

Rationale: The collapsibility index of the inferior vena cava (cIVC) shows interesting performance to predict fluid responsiveness in spontaneously breathing patients. Nevertheless, measurement sites of inferior vena cava (IVC) diameters remain controversial for that purpose. The aim of the study was to test the accuracy of different measurement sites of cIVC to predict fluid responsiveness in spontaneously breathing patients.

Patients and methods: This study is a post hoc analysis of two prospective cohorts. We included spontaneously breathing patients without mechanical ventilation presenting with sepsis-related acute circulatory failure and considered for volume expansion (VE). We assessed hemodynamic status at baseline and after a fluid challenge (FC) induced by a 30 min-infusion of 500 mL-gelatin 4%. The IVC diameters were measured off-line with ultrasonography using the bi-dimensional mode on a subcostal long-axis view. The cIVC was calculated as [ (expiratory-inspiratory)/expiratory] diameters during standardized (cIVC-st) and unstandardized breathing (cIVC-ns) conditions. Breathing standardization consisted of a deep inspiration with concomitant control of buccal pressures and passive exhalation. Patients were referred to be responders to FC (i.e. fluid responsive) when the stroke volume increased by ≥ 10%.

Results: Among the 81 patients included in the study, 41 (51%) were responders to FC. The accuracy of cIVC-st and cIVC-ns before FC to predict fluid responsiveness differed significantly by measurement sites (interaction p value < 0.0001 and < 0.001, respectively). Measuring IVC diameters 4 cm from the junction of the IVC and the right atrium provided the best accuracy to predict fluid responsiveness (Fig. 1). At 4 cm caudal to the right atrium, cIVC-st was significantly better than cIVC-ns to predict fluid responsiveness: area under ROC curve 0.98 (95% CI 0.97–1.0) versus 0.84 (95% CI 0.78–0.94), p < 0.001. At 4 cm, a cIVC-st ≥ 44% and a cIVC-ns ≥ 33% predicted fluid responsiveness with sensitivity of 93% and 66%, and specificity of 98% and 92%, respectively.

Conclusion: Accuracy of cIVC to predict fluid responsiveness in spontaneously breathing patients depends on both measurement sites of IVC diameters and breathing conditions. Measuring IVC diameters during a standardized inspiration maneuver at 4 cm caudal to the right atrium is the most relevant mean to optimize cIVC performance to guide VE.

Compliance with ethics regulations: Yes.

Fig. 1

Area under ROC curves of the collapsibility index of the inferior vena cava (cIVC) assessed during standardized (cIVC-st, panel a) and unstandardized (cIVC-ns, panel b) breathing conditions to predict fluid responsiveness. 0, 1, 3, 4, 5: measurement site of cIVC 0, 1, 3, 4 and 5 cm caudal to the right atrium

F-052 Search for trans-thoracic echocardiography predictive criteria of hemodynamic instability during intermittent hemodialysis in critically ill patients

Maxime Leclerc, Benoit Courteille, Cedric Daubin, Julien Dupeyrat, Suzanne Goursaud, Aurelie Joret, Toufiq Rhanem, Xavier Valette, Damien Cheyron

CHU Caen Côte de Nacre, Caen, France

Correspondence: Maxime Leclerc (leclerc.maxime@hotmail.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-052

Rationale: Intermittent hemodialysis (IHD) is increasingly used in patients admitted to intensive care unit (ICU) with acute kidney injury (AKI) requiring renal replacement therapy (RRT). However, this technique is associated with nearly 20% of episodes of perdialytic hemodynamic instability (HI), a common cause of increased morbidity and mortality. At the same time, trans-thoracic echocardiography (TTE) has become widely used in intensive care units and is now one of the hemodynamic monitoring methods used daily in the ICU setting.

Patients and methods: Search for one or more pre-dialysis TTE criteria predictive of perdialytic HI, defined by a systolic blood pressure (SBP) lesser than 90 mmHg or a suddain decrease in SBP of more than 40 mmHg. Prospective, observational study of standard care in a medical ICU. Collection of demographic, clinical and pre-dialysis echocardiographic data from included patients.

Results: Twenty-five patients with a total of 98 sessions of IHD between November 2017 and November 2018 were included in the study. TTE was performed for each patient before each IHD session. HI occurred in 31 hemodialysis sessions. In univariate analysis, the existence of prior heart disease (38% vs 0%, p = 0.04), a greater diameter of the left atrium (4.2 vs 3.6 cm, p = 0.0001), a lower cardiac output (5.8 vs 6.6 l/min, p = 0.04), a right dysfunction assessed by lowered TAPSE and S-wave (16 vs 24 mm, p < 0.0001 and 12.8 vs 15.7 cm/s, p = 0.002, respectively) and an increase in PAPS (45 vs 34 mmHg, p = 0.003) were significantly associated with the occurrence of perdialytic HI (Fig. 1). In multivariate logistic regression analysis, right cardiac dysfunction assessed by lower TAPSE (OR: 0.6, 95% Confidence intervals (CI)[0.4–0.9]; p = 0.03) and higher PAPS (OR: 1.3, 95% CI[1.1–1.5]; p = 0.03) were predictive of perdialytic HI. This result is confirmed by an analysis using a generalized estimation equation (GEE) model that finds lower TAPSE significantly associated with perdialytic hypotension episodes (adjusted OR: 0.90; 95% CI [0.82–0.99]; p = 0.03).

Conclusion: Right cardiac dysfunction appears to be an easily assessable echocardiographic predictor of the occurrence of hemodynamic instability during IHD in critically ill patients.

Compliance with ethics regulations: Yes.

Fig. 1

Comparison of predialysis TAPSE (a) and PAPS (b) values according to the occurrence or not of hemodynamic instability during the dialysis session

F-053 Global longitudinal strain is the best transthoracic echocardiography parameter of left and right ventricular systolic function in critically ill patients

Pierre Dupland, Ariane Gavaud, Frédéric Pene, Jean-Daniel Chiche, Jean-Paul Mira, Alain Cariou, Mathieu Jozwiak

Service de Médecine Intensive Réanimation, Hôpitaux universitaires Paris-Centre, Hôpital Cochin, Paris, France

Correspondence: Pierre Dupland (pierre.dupland@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-053

Rationale: Several transthoracic echocardiography (TTE) parameters of left (LV) and right ventricular (RV) systolic function are available. We compared the ability of these different parameters to track changes in LV or RV systolic function and to detect LV or RV systolic dysfunction in critically-ill patients.

Patients and methods: In 20 patients (15 mechanically ventilated and 2 with atrial fibrillation), TTE examinations were performed before and after i) infusion of 500-mL of saline (n = 10), ii) changes in norepinephrine (n = 8), iii) or in dobutamine (n = 2) dosage. For the LV systolic function, we compared the mitral annular plane systolic excursion (MAPSE), the systolic (s’) peak velocity of the lateral mitral annulus and the global longitudinal strain (GLSLV) to the LV ejection fraction (LVEF), considered as the gold standard. For the RV systolic function, we compared the tricuspid annular plane systolic excursion (TAPSE), the systolic peak (S) velocity of the tricuspid annulus and the global longitudinal strain (GLSRV) to the RV fractional area change (FAC), considered as the gold standard.

Results: After pooling all values, LVEF (55 ± 12% at baseline) was better correlated to GLSLV (r² = 0.71) than to MAPSE (r² = 0.25) and s’ wave (r² = 0.13) (each p < 0.05). The concordance rate between changes (in %) in LVEF and in the other parameters of LV systolic function was 87% for GLSLV, 72% for MAPSE and 67% for s’ wave. Both MAPSE and s’ wave could not reliably detect moderate (30% ≤ LVEF ≤ 40%) or severe (LVEF < 30%) LV dysfunction. Conversely, a GLSLV > − 12% predicted moderate LV dysfunction with a sensitivity of 100% (95% IC: 29–100%) and a specificity of 89% (95% IC: 71–98%) and a GLSLV > − 9.5% predicted severe LV dysfunction with a sensitivity of 100% (95% IC: 16–100%) and a specificity of 96% (95% IC: 82–100%). After pooling all values, FAC (32 ± 11% at baseline) was better correlated to GLSRV (r² = 0.57) than to TAPSE (r² = 0.31) and S wave (r² = 0.24) (each p < 0.05). The concordance rate between changes (in %) in FAC and in the other parameters of RV systolic function was 81% for GLSRV, 72% for TAPSE and 50% for S wave. Both TAPSE and S wave could detect RV dysfunction (FAC ≤ 35%) with moderate reliability only. Conversely, a GLSRV > − 12% detected RV dysfunction with a sensitivity of 56% (95% IC: 30–80%) and a specificity of 100% (95% IC: 77–100%).

Conclusion: In critically-ill patients, GLSLV and GLSRV seem to be the best TTE parameters of LV and RV systolic function. Enrolments are still ongoing, which may allow further analysis.

Compliance with ethics regulations: Yes.

F-054 The Applicability of Fluid Responsiveness Indices in Circulatory failure (AFRIC study)

Rui Shi, Nello De Vita, Francesco Gavelli, Jean-Louis Teboul, Arthur Pavot, Xavier Monnet

Service de médecine intensive-réanimation, Hôpital de Bicêtre, Hôpitaux universitaires Paris-Saclay, Assistance publique—Hôpitaux de Paris, Inserm UMR S_999, Université Paris-Sud, Le Kremlin-Bicêtre, France

Correspondence: Rui Shi (shiruidingding@hotmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-054

Rationale: Passive leg raising (PLR), pulse pressure variation (PPV), and the 15-second end-expiratory occlusion test (EEXPO) are frequently used to assess preload responsiveness. However, there are conditions in which they are not valid or feasible, which may preclude their applicability in the daily clinical practice. The aim of this study was to estimate the prevalence of such conditions in critically ill patients with acute circulatory failure.

Patients and methods: Between January and April 2019, all patients of a 25-bed medical ICU were daily screened and those with acute circulatory failure, defined by norepinephrine infusion or fluid therapy > 1L during the previous 24 h, were included. In each of them, we screened the criteria of validity/feasibility of PPV, PLR and EEXPO.

Results: Eighty-four patients (78% with septic shock, 8% with cardiogenic shock, 8% with hypovolemic shock, 6% with non-septic vasoplegic shock) were enrolled in the study. Among them, norepinephrine infusion was ongoing at the time of enrolment in 95% of the patients whilst 69% were under mechanical ventilation, and 30% with acute respiratory distress syndrome. PLR was not applicable in 37% of cases. This was mainly due to venous compression stocking (17% of cases), intra-abdominal hypertension (11% of cases), and either an absence of cardiac output monitoring or impossibility to perform echocardiography (9% of cases). Among the 58 intubated patients, PPV was applicable in 16% of cases, including cases with high PPV under conditions generating false negatives (low tidal volume or lung compliance) or low PPV values under conditions generating false positives (spontaneous breathing, cardiac arrythmias). However, PPV was not interpretable in 84% of cases. This was mainly due to low tidal volume ventilation (27% of cases), spontaneous breathing activity (12% of cases), while the remaining non-interpretable cases (51%) had more than one reason. In the 58 intubated patients, EEXPO was not applicable in 21% of cases. This was due to impossibility for patients to sustain a 15-s hold of mechanical ventilation in 58% of cases, and either an absence of cardiac output monitoring or the impossibility to perform echocardiography in 42% of cases. PLR and EEXPO were both valid and feasible in 35% of the patients, and the three tests were all feasible in only 4% of patients.

Conclusion: In patients with acute circulatory failure, among the different indices of preload responsiveness, PLR cannot be performed in 37% of all cases, mainly due to venous compression stocking and intra-abdominal hypertension. PPV cannot be used in 84% of intubated cases. EEXPO is not feasible in 21% of cases.

Compliance with ethics regulations: Yes.

F-055 Sleep Apnea Syndrome in hypercapnic respiratory failure: prevalence and validation of screening scores

Fadoua Houri, Wiem Nouira, Hedia Ben Ahmed, Zeineb Hammouda, Manel Lahmar, Fekri Abroug, Lamia Besbes

CHU F.Bourguiba, Monastir, Tunisia

Correspondence: Fadoua Houri (hourifadwa@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-055

Rationale: Comorbid association between chronic respiratory diseases and sleep apnea syndrome (SAS) revealed frequent with systematic search in ICU following ICU stay. This association carries prognosis impact depending whether specific treatment is implemented or not. NoSAS and Stop Bang scores are proposed for screening of SAS in general population. The aim of the present study is to report the prevalence of SAS in ICU patients admitted for hypercapnic respiratory failure and compare association of NoSAS and Stop Bang score with SAS severity.

Patients and methods: The study was conducted between January 2016 and September 2018. Patients consecutively admitted in the ICU for hypercapnic respiratory failure had calculation of a no SAS and STOP Bang scores at admission. In survivors nocturnal polygraphic records was performed 3 to 4 weeks following ICU discharge. The association between the number of apnea–hypopnea episodes, BMI, and clinical variables suggestive of SAS, was tested by POISSON regression model.

Results: During the study-period, 65 patients (mean age: 69 ± 9 years, pH 7.29 ± 0.03, PaCO₂ 75 ± 16) were admitted for hypercapnic respiratory failure. Non invasive ventilation was used in 85% and death occurred in six patients. Polygraphic records were performed in 45 (9 lost to follow-up) Mean apnea–hypopnea index was 40 ± 15 with a minimum of 7 and a maximum of 62. Poisson logistic regression showed that No SAS (p = 0.006) but not Stop Bang (p = 0.1) was associated with the level of apnea–hypopnea index.

Conclusion: ICU patients admitted for hypercapnic respiratory failure have a very high rate of associated Sleep Apnea Syndrome. Among existing screening scores, No SAS and not Stop Bang are associated with severity.

Compliance with ethics regulations: NA.

F-056 Extracorporeal CO₂ Removal in acute exacerbation of COPD not responding to non-invasive ventilation: a single center experience

Daniel Silva¹, Rita Serbouti², Laurent Laine¹, Nathalie Memain¹, Vincent Ioos¹, Mathilde Lermuzeaux¹, Camille Legouy¹, Morgan Benais¹, Luis Ferreira¹, Jérôme Aboab¹

¹Médecine Intensive Réanimation-Centre Hospitalier de Saint-Denis, Saint-Denis, France; ²Fresenius Medical Care France, Fresnes, France

Correspondence: Daniel Silva (daniel.silva@mac.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-056

Rationale: Extracorporeal carbon dioxide removal (ECCO₂R) has shown a raising interest in the context of acute exacerbation of chronic obstructive pulmonary disease (eCOPD) to reduce length or avoid mechanical invasive ventilation (MIV) when non-invasive ventilation (NIV) has failed.

Patients and methods: We conducted a single center retrospective analysis on successive eCOPD patients for whom NIV failed. ECCO₂R was performed with the iLA Activve (Xenios Novalung, Heilbronn, Germany) device from February 2015 to February 2019. We compared blood gas measurements and complication rates on eCOPD patients treated with MIV after NIV failure from January 2010 to February 2015. 28 patients were treated with ECCO₂R and 26 patients with MIV. Variables are reported as mean ± SD. Quantitative values are compared with t test (95% confidence interval) and qualitative values with Pearson’s Chi squared test withYates’ continuity correction. Population distribution was tested with the Shapiro–Wilk normality test.

Results: In the ECCO₂R group (SAPS 48 ± 15, age 66.1 ± 11.4 years), the ECCO₂R duration was 5.7 ± 4.2 days. The arterial blood pH 6 h before ECCO₂R was 7.24 ± 0.05 and 7.41 ± 0.06 at decannulation (p < 0.001). The PaCO₂ value 6 h before ECCO₂R was 85 ± 21 and 53 ± 9.9 before decannulation (p < 0.001). In the MIV group (SAPS 50 ± 14, age 71.4 ± 11 years), the MIV duration was 28.2 ± 42.5 days. The mean arterial blood pH 6 h before intubation was 7.30 ± 0.15 and 7.40 ± 0.06 before extubation (p = 0.0741). The mean PaCO₂ value 6 h before MIV was 77 ± 32 and 51 ± 13.3 before extubation (p < 0.001). Three patients (10.7%) needed intubation despite the ECCO₂R. Three patients had major bleeding complications in the ECCO₂R group. Three patients had device related thrombosis and one had a venous thrombosis. In the MIV group, 8 patients (29%) had transfusions, five (18%) had extubation failure, 6 patients (21%) had autoextubation and 4 (14%) got a ventilator associated pneumonia. Length of stay in the ICU for the ECCO₂R and the MIV group was 19 ± 15.5 days and 31 ± 42.2 days respectively. Mortality at 28 days and 90 days was 8% and 14% respectively in the ECCO₂R group and 16% and 30% in the MIV group.

Conclusion: Our results show that ECCO₂R is feasible and brings significant improvement on pH and PaCO₂ with relatively low complication rate. Evidence for a mortality benefice of ECCO₂R in eCOPD patients failing NIV therapy has yet to be proven in a larger and randomized study.

Compliance with ethics regulations: Yes.

F-057 High-Flow Nasal nebulization of Salbutamol during severe acute exacerbation of Chronic Obstructive Pulmonary Disease

Clément Beuvon¹, Nicolas Marjanovic¹, Rémi Coudroy¹, Christophe Rault¹, Vanessa Bironneau¹, Xavier Drouot¹, René Robert¹, Arnaud Thille¹, Jean-Pierre Frat¹

¹CHU de Poitiers, Poitiers, France

Correspondence: Clément Beuvon (cbeuvon@hotmail.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-057

Rationale: Patients with severe acute exacerbations of Chronic Obstructive Pulmonary Disease (COPD) may benefit from high-flow nasal oxygen regarding its physiological effects and good tolerance. Bronchodilator vibrating mesh nebulization through high-flow nasal oxygen circuit has been described to induce similar effect to standard facial mask jet nebulization in stable COPD patients. We aim to evaluate whether vibrating mesh nebulization of salbutamol through high-flow nasal oxygen circuit is efficient in unstable patients with COPD.

Patients and methods: We conducted a monocenter non-randomized physiological prospective cross-over study, between January and September 2019, including ICU patients with severe acute exacerbation of COPD and respiratory acidosis treated by salbutamol nebulization. Spirometry and airway resistances records were performed after a 3-h wash-out period without bronchodilator, before and after vibrating mesh nebulization of 5 mg salbutamol through high-flow nasal oxygen circuit. The primary endpoint was Forced Expiratory Volume in 1 s after salbutamol nebulization. Secondary endpoints included other spirometry parameters, clinical parameters, dyspnea assessed by a Borg scale.

Results: Fourteen consecutive patients were included, Forced Expiratory Volume in 1 s increased significantly after salbutamol nebulization through high-flow nasal oxygen (91 ± 111 mL, p = 0.005), as well as Forced Vital Capacity (190 mL ± 193, p = 0.005). Airway resistances were not significantly changed after nebulization (− 0.73 ± 1.67, p = 0.08) as well as Peak Expiratory Flow (+ 231 mL ± 439, p = 0.09). No difference was observed on Borg scale (p = 0.06) and respiratory rate (p = 0.18) after salbutamol nebulization, while heart rate increased significantly (p = 0.01).

Discussion: Salbutamol nebulization using vibrating mesh nebuliser placed on high-flow nasal oxygen circuit induces a significant but moderate bronchodilation in patients with severe acute exacerbation of COPD. Moreover, improvement of Forced Vital Capacity after salbutamol nebulization suggests a reduction of dynamic hyperinflation.

Conclusion: Salbutamol vibrating mesh nebulization through high-flow nasal oxygen circuit increases significantly Forced Expiratory Volume in 1 s.

Compliance with ethics regulations: Yes.

F-058 T-piece versus sub-therapeutic pressure support for weaning from invasive mechanical ventilation in patients with chronic obstructive pulmonary disease: a comparative prospective study

Amira Jamoussi, Fatma Jarraya, Samia Ayed, Takoua Merhabene, Jalila Ben Khelil, Mohamed Besbes

Abderrahmen Mami Hospital, Tunis, Tunisia

Correspondence: Amira Jamoussi (dr.amira.jamoussi@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-058

Rationale: The best weaning strategy for patients with chronic obstructive pulmonary disease (COPD) remains unknown. The spontaneous breathing trial (SBT) represents a crucial step of weaning, but the choice between the T-piece (SV-Tube) or the sub-therapeutic setting of the level of pressure support without positive expiratory pressure (PSV) is still a matter of debate. We aimed to compare the success of extubation between two groups of COPD patients according to the SBT type (VS-Tube vs PSV).

Patients and methods: It was a prospective and comparative study, from April 2017 to March 2019, at the Abderrahmen Mami hospital’s intensive care unit (ICU). COPD patients who underwent invasive mechanical ventilation (MV) for at least 24 h and met the criteria for weaning were included and randomized to SV-Tube or PSV. A multivariate analysis was performed to determine the association between the SBT modality and the success of extubation (no re-intubation during the 48 h following extubation).

Results: During the two years’ study, 32 patients were included. The mean age was 66 ± 10 years, the sex-ratio was 4.33. Weaning process was simple in 16 patients (50%), difficult in 10 patients (31%) and prolonged in 6 patients (19%). Fifteen and 17 patients were respectively randomized to the SV-Tube and PSV groups. The mean duration of MV before randomization was comparable between the 2 groups (SV-tube 6.87 ± 4.3 days vs PSV 6.06 ± 4.8 days, p = 0.622). Mean weaning time (days) was 3.73 [1–16] for the SV-Tube group and 4.35 [1–20] for the PSV group. The mean total MV duration (days) was higher in the SV-Tube group than in the PSV group (15.13 vs 7.65, p = 0.105). The number of re-intubated patients within 48 h following extubation was higher in the PSV group (7/17 vs 1/15, p = 0.024) as well as the overall reintubation rate (64.7% vs 20%, p = 0.011). In multivariate analysis, the SBT’s trial was independently associated to the success of extubation (OR = 0.081, IC [0.008–0.865], p = 0.037) in favor of SV-tube’ modality. The median length of stay in intensive care was 16 days [10; 26]. The mortality was higher in the PSV group (7/17 vs 2/15, p = 0.08). Extubation failure was a factor associated with mortality (OR = 8.333, CI [1.392, 49.872], p = 0.02).

Conclusion: Ventilation weaning was easy in 50% of intubated COPD patients. SV-Tube as SBT modality was associated to success of extubation in patients with COPD. Mortality in intensive care was significantly higher in re-intubated patients.

Compliance with ethics regulations: Yes.

F-059 The incidence of laryngo-tracheal lesions in tracheotomized patients in Post Intensive Care Rehabilitation Unit

Laurence Donetti, Dominique Kaminski, Mohamed Laissi, Jean-Claude Marchal, Pascal Meyer, Karim Nssair, Gérald Choukroun

CH Forcilles, Ferolles Attilly, France

Correspondence: Laurence Donetti (laurdonetti@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-059

Rationale: Post intensive care rehabilitation units (PICRU or SRPR) are dedicated to chronic critically ill patients management: particularly prolonged ventilator weaning, removal of tracheotomy tube (decannulation) and functional recovery. In our unit, more than 80% of patients are tracheotomized. A broncoscopy is systematically performed prior to decannulation either immediately before or to assess laryngeal abnormalities in case of swallowing disorders. The anatomical and functional abnormalities are investigated and the good positioning of the tracheal cannula is checked.

Patients and methods: We conducted a retrospective single-centre observational study from January 1st 2017 to June 30th 2019 in order to evoluate the rates of the different bronchoscopic abnormalitie patterns among the trachotomized patients admitted during the study period.

Results: During the period, 169 patients were admitted in our unit. 109 patients were eligible for analysis. 36 patients (33%) didn’t have lesion vs 73 patients (67%) who had one or more lesions. 30 patients had swallowing disorders. Characteristics of patients without lesion were: mean age of 63 years, lengh of stay of 30.3 days and a 100% success rate of tracheotomy removal after the bronchoscopy. Characteristics of patients with lesions were: mean age of 63 years, length of stay of 49 days. 86% of them underwent successfull decannulation: 47% immediately after the first bronchoscopy and 39% later. Laryngal lesions are listed in the table.

10 patients had both epiglottis and arytenoides edema, 7 patients had both epiglottis, arytenoides and vocal cords edema. Tracheal disorders patterns were: tracheal ring fractures (15 patients), granulomas (11 patients), tracheomalacia (7 patients), subglottic edema (4 patients), tracheal stenosis (2 patients).

Tracheal cannula malposition was found in 9 patients.

Conclusion: Laryngeal or tracheal abnormalities are frequent among chronic critically ill patients admitted in PICRU and probably associated with delayed ventilator weaning and decannulation. Further studies are needed to understand better causes and consequences.

Compliance with ethics regulations: Yes.

Table 1 Patterns of laryngeal lesions

F-060 Which score to predict NIV failure in Hypercapnic Respiratory Failure: HACOR score vs ROX index

Wiem Nouira, Zeineb Hammouda, Syrine Maatouk, Hedia Ben Ahmed, Amir Bedhaifi, Yed Maatouk, Fadoua Houri, Fekri Abroug, Lamia Besbes

CHU F.Bourguiba, Monastir, Tunisia

Correspondence: Wiem Nouira (wiemnouira1@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-060

Rationale: Non-invasive ventilation has become the mainstay in hypercapnic respiratory failure. Delaying intubation and invasive ventilation is associated with a worse outcome in these patients. Although a predictive score of NIV failure has been validated for hypoxemic respiratory failure no such score exists in hypercapnic respiratory failure. The aim of our study is to compare the performance of two scores in the predictive NIV failure hypercapnic respiratory failure.

Patients and methods: Consecutive patients admitted between January 2017 and July 2019 for hypercapnic respiratory failure, were included. HACOR score and ROX score were calculated in each patient at admission. In patients ventilated non-invasively, the outcome (NIV success or failure) was noted. The area under curve (AUC) and operative characteristics were computed for both scores.

Results: During the study-period, 107 out of 133 patients admitted for hypercapnic respiratory failure received NIV as the primary ventilatory mode. These patients were mainly men (77/30), had a mean age of 66.7 ± 10 years and had the following pulmonary disease: COPD exacerbation 50.5%, obesity-hypoventilation syndrome 18.7%, bronchiectasis 15.9%, and other diseases: 8.4%. NIV failure occurred in 39 patients (36.4%) and ICU mortality in 18.7%. Mean HACOR score and ROX score were 5.8 ± 3.6 and 18.3 ± 2, respectively. The AUC under ROC was higher for HACOR than ROX (0.91 and 0.76 respectively) (Fig. 1). The HACOR score (cut-off 6) had a sensitivity of 0.93 and specificity of 0.85.

Conclusion: HACOR score seems more accurate in predicting NIV failure in hypercapnic respiratory failure. Further prospective validation is needed.

Compliance with ethics regulations: NA.

F-061 Characteristics and outcome of patients managed for difficult weaning in a weaning and rehabilitation care unit

Sarah Makoudi, Julie Delemazure, Martin Dres, Alexandre Demoule, Elise Morawiec

Unité de soins de rééducation post réanimation Service de Pneumologie, Médecine Intensive, Paris, France

Correspondence: Sarah Makoudi (smmakoudi@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-061

Rationale: Published data on outcomes in respiratory weaning centers are limited and seem to depend on the organisation of healthcare systems and patient case-mix. The weaning center of our university hospital (Post intensive care rehabilitation unit) admits for weaning and rehabilitation patients from medical and surgical intensive care units without severe neurological pathologies. The aim of this study was to describe patient’s characteristics and outcome (weaning outcomes and survival) and to compare in subgroups according to the initial medical, surgical or cardiac surgical context.

Patients and methods: We conducted a monocentric retrospective observational study between 01/02/2016 and 30/06/2018. «Successful outcome» was defined by the association of survival and weaning from invasive ventilation. Factors associated with evolution were investigated by uni- and multivariate analysis. Survival after discharge was analysed according to the initial context and according to the type of ventilation at discharge.

Results: Among 215 patients included, 167 (77.6%) had a successful outcome with high use of non-invasive ventilation (NIV) (37%). Respiratory history (p = 0.02), female gender (p < 0.001), IGS2 score at admission to the SRPR (p = 0.001) and non-cardiac surgical setting (p < 0.012) were associated with an adverse course. The 6-month survival rate was 84% in discharged patients. The outcome was not different in the tree subgroups. NIV rate at discharge was high in the subgroup of cardiac surgery patients.

Conclusion: A multidisciplinary and personalised approach by a specialized weaning unit can provide a successful service model for patients who require liberation from prolonged invasive mechanical ventilation.

Compliance with ethics regulations: Yes.

F-062 Hemodynamic impact of High-dose Insulin Euglycemic Therapy in calcium channel blockers poisoning

Julien Goutay¹, Erika Parmentier-Decrucq¹, Olivier Nigeon², Julien Poissy¹, Daniel Mathieu¹, Thibault Duburcq¹

¹CHU de Lille, Pôle de Réanimation, Hôpital Salengro, Lille, France; ²CH de Lens, Service de Réanimation, Lens, France

Correspondence: Julien Goutay (julien.goutay@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-062

Rationale: High-dose Insulin Euglycemic Therapy (HIET) is recommended as first line therapy for calcium channel blockers (CCBs) poisoning because of its inotropic effect. Our first objective was to study its hemodynamic impact.

Patients and methods: We performed a retrospective cohort study of all consecutive patients admitted for CCBs poisoning treated with HIET, in one ICU at the University Hospital of Lille between January 2013 and July 2018. The hemodynamic impact was studied through mean arterial pressure (MAP), Vasoactive-Inotropic Score (VIS) and MAP/VIS ratio during the 24 h following HIET initiation. Metabolic parameters were also collected.

Results: 104 patients admitted for CCBs poisoning. 43 patients treated with HIET in ICU (7 patients without circulatory shock, 5 patients with shock after HIET and 31 patients with shock at baseline before HIET). Among shocked patients at baseline (n = 31), no hemodynamic improvement was found except an increased MAP/VIS ratio at H24 (p < 0.05). On the contrary, an initial worsening of VIS (137 [26–267] at H0 vs. 143.4 [40.1–330] at H2, p = 0.03) was observed. 22 patients (51.2%) presented a severe hypokalemia and 10 (23.2%) a severe hypoglycemia. The existence of severe conduction abnormality and the value of Simplified Acute Physiology Score II were associated with mortality in the multivariate analysis.

Conclusion: No hemodynamic improvement was observed in the first hours following HIET. Moreover, severe metabolic disorders occurred. A randomized prospective study aiming to assess the HIET risk–benefit ratio is needed.

Compliance with ethics regulations: Yes.

Fig. 1

Evolution of MAP/VIS ratio among shocked patients at baseline

F-063 Liver Dysfunction and Ketamine use in critically ill patients with ARDS: a Post Hoc analysis of the ACURASYS study

Florence Daviet¹, Benjamin Coiffard¹, Arnaud Gacouin², Gilles Perrin³, Anderson Loundou⁴, Samir Jaber⁵, Jean-Michel Arnal⁶, Dider Perez⁷, Jean-Marie Seghboyan⁸, Jean-Michel Constantin⁹, Jean-Yves Lefrant¹⁰, Claude Guérin¹¹, Gwenael Prat¹², Jean-François Timsit¹³, Antoine Roch¹, Laurent Papazian¹, Jean-Marie Forel¹

¹Médecine Intensive Réanimation, Service du Pr PAPAZIAN, Hôpital Nord, Marseille, France; ²Service des maladies infectieuses et réanimation médicale, hôpital Pontchaillou, centre hospitalier universitaire de Rennes, Rennes, France; ³Réanimation des Urgences et Medicale, CHU la Timone 2, Marseille, France; ⁴Unité d’Aide Méthodologique à la Recherche Clinique-Assistance Publique-Hôpitaux de Marseille, Marseille, France; ⁵Département d’Anesthésie-Réanimation, hôpital Saint-Éloi, Montpellier, France; ⁶Service de Réanimation Polyvalente, Hôpital Sainte Musse, Toulon, France; ⁷Service de Réanimation-USC, Hôpital Lons-le-Saunier, Lons-Le-Saunier, France; ⁸Réanimation Hopital Européen, Marseille, France; ⁹Department of Perioperative Medicine, University Hospital of Clermont-Ferrand, Clermont-Ferrand, France; ¹⁰Department of Anaesthesiology, Critical Care and Emergency Medicine, CHU de Nîmes, Nimes, France; ¹¹Réanimation Médicale Hôpital de la Croix Rousse, Lyon, France; ¹²Medical Intensive Care Unit, Cavale Blanche University Hospital, Brest, France; ¹³Medical and Infectious Diseases ICU (MI2), Bichat Hospital, Paris, France

Correspondence: Florence Daviet (florence.daviet@ap-hm.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-063

Rationale: Ketamine is used in the induction and maintenance of general anesthesia. Recently, there were concerns regarding its liver toxicity. We conducted a study to investigate the link between Ketamine use and Liver Dysfunction (LD) in Intensive Care Unit (ICU) patients.

Patients and methods: Data were extracted from the [anonymized] study, a randomized controlled trial designed to evaluate the effect of Cisatracurium on 90-day mortality rate in moderate and severe Acute Respiratory Distress Syndrome (ARDS) patients. The main endpoint was the occurrence of a LD defined as a total serum bilirubin superior or equal to 33 micromol/l. A matched case–control cohort was created: cases, receiving at least 1 day of continuous Ketamine infusion, were paired 1 for 1 with controls according to treatment with Cisatracurium, hepatic and cardiovascular SOFA sub-score, total serum bilirubin level at the time of inclusion, age, sex, ARDS from septic origin, shock any-time after inclusion. An analysis was also made on the whole cohort comparing the patients receiving at least 1 day of continuous Ketamine infusion to all patients who did not fulfill this criterion.

Results: 66 cases were identified and matched to 66 controls. In the Ketamine group, the median Ketamine duration was 5 (3–8) days, and median total cumulative dose 18.6 (9.6–42.5) g. The occurrence of LD was higher in the Ketamine group than in the matched control group (53.8% versus 28.6%, p = 0.002, Fig. 1). The Hazard ratio (HR) for LD in the Ketamine group was 2.03 (95% CI 1.15–3.59, p = 0.015). There was an increased risk of LD of 5.9% per day of exposure to Ketamine (HR 1.06, 95% CI 1.02–1.1 p = 0.003) and of 0.8% per gram of Ketamine infused (HR 1.01, 95% CI 1.00–1.01, p = 0.006), with a risk starting to be statistically significant after 4 days and 18gr. In multivariate analysis on the whole cohort, Ketamine exposure (HR 2.49, 95% CI 1.40–4.44, p = 0.02), cumulative dose in gram (HR: 1.02, 95% IC: 1.01–1.04, p = 0.002) and ketamine exposure in days (HR: 1.16, 95% IC: 1.07–1.25, p < 0.001) remained independent risk factors for LD occurrence.

Conclusion: Ketamine use in critically ill patients treated for ARDS is associated to a higher risk of liver dysfunction, assessed by total serum bilirubin. This risk is dose-dependent and increases with duration of treatment. The prescription of high doses or prolonged treatment with Ketamine should probably be avoided in critically ill patients.

Compliance with ethics regulations: Yes.

Fig. 1

Kaplan–Meier Curves of Probability of Liver Dysfunction (LD) free days according to Ketamine exposure in the matched cohort

F-064 Cardiovascular complications following Ciguatera fish poisoning in the French West Indies—A case series

Dabor Resiere¹, Florentin Jonathan², Mahi Zacharia², Papa Gueye², Hatem Kallel³, Hossein Mehdaoui¹

¹Department of Critical Care, University Hospital of Martinique, Fort-De-France, France; ²University Hospital of Martinique, Fort-De-France, France; ³Intensive Care Unit, Cayenne General hospital, Cayenne, 97300, French Guiana, France

Correspondence: Dabor Resiere (dabor.resiere@chu-martinique.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-064

Rationale: Ciguatera is one of the most common cases of marine poisoning associated with fish consumption in the world. The incidence of this intoxication is largely unreported. In Martinique, the incidence of this intoxication seems constantly increasing. During the last 3 years, numerous cases of large collective poisonings have been reported in Martinique, especially during summer. The spectrum of clinical manifestations is large including gastrointestinal, neurological and cardiovascular symptoms. Ciguatoxin, the toxin responsible for ciguatera fish poisoning is considered as a sodium channel agonist with cholinergic and adrenergic activity. It is rarely fatal and management of poisoned patients is essentially based on supportive care. The objective of this study was to describe the clinical characteristics and complications of ciguatera poisoning in Martinique, focusing on the cardiovascular ones.

Patients and methods: Observational, retrospective, single—center study covering six-year period from October 2012 to September 2018, including all patients admitted to the Emergency Department of the University Hospital of Martinique (CHU), and all patients who were declared to the Regional Health agency (ARS) for ciguatera intoxication.

Results: One hundred and forty-nine patients (149) who were ciguatera-affected were included. The incidence rate found was to be 0.67 cases per 10.000 patient-years in Martinique over the period. About 90% of patients had gastrointestinal symptoms such as nausea, vomiting, diarrhea, or abdominal pain; 83% neurological disorders and 42% cardiovascular symptoms including, bradycardia, hypotension and interventricular block. Ingestion of Carangue fish was related to a major risk of chronic signs.

Conclusion: The incidence of ciguatera in Martinique is increasing, with 0.67 cases/10.000 patient-years. The clinical presentation is defined mainly by digestive signs, followed by peripheral neurological disorders and cardiovascular symptoms. Ciguatera fish poisoning in Martinique presents similar clinical presentation to that of the other Caribbean Islands. There is no specific treatment. Acute ciguatera poisoning is responsible for significant cardiovascular complications. Physicians should be aware of the potential cardiovascular risk of ciguatera poisoning.

Compliance with ethics regulations: Yes.

Reference

• 1. Friedman MA, Fleming LE, et al. Ciguatera fish poisoning treatment, prevention and management. Mar Drugs 2008; 6:456

F-065 Acute organophosphate poisoning

Fahd Moussaid, Reda Elhadrami, Hamza Elhamzaoui, Taoufik Abouelhassan

Emergency department, University Hospital Mohamed VI, Marrakec, MOROCCO

Correspondence: Fahd Moussaid (fahdmoussaid5@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-065

Rationale: Pesticides have represented the most incriminated products in severe acute poisonings, in the developing countries, due to the availability of these products. Organophosphate poisoning accounts for 3 million poisonings/year worldwide. Organophosphate (OP) pesticides are used mainly as insecticides in agriculture. the Moroccan anti-poison and pharmacovigilance centrer shows that OP poisoning are responsible for 13% of all poisonings combined. The aim of our study: epidemiological, clinical, management and prognostic factors.

Patients and methods: A retrospective study was conducted on patients with OP poisoning admitted to our nine-bed medical intensive care unit between January 2018 and December 2018. inclusion criteria were: all patients over 16 years of age and the exlusion criteria were: pesticide poisoning other than OP, alcohol poisoning, drug poisoning, scorpionic poisoning and snake bites. statistical analysis was performed with SPSS software.

Results: Forty patients were admitted for acute OP poisoning. In Morocco, organophosphores are available over-the-counter in several forms: Rodentocides, Malathion, Cockroach trap, Baygon Insecticide (Fig. 1). The average age was 26 years with a female prévalence of 74.1%. The intoxications were mostly intentional (85%). The symptomatology was determined by the three syndromes: central syndrome in 50%, muscarinic syndrome in 69%, nicotinic syndrome in 10%. Rhythm disorders in 8%, and cardiovascular collapse in 9%. The symptomatic treatment was applied to all patients, antidotic treatment was administered in 80% of patients. The average length of hospitalization was 06 days.

Conclusion: Acute OP poisoning is a real public health problem. Its associated symptomatic treatment (respiratory and neurological resuscitation) and antidotic treatment. The mortality remains high in our context, therefore, we must attach great importance to the prevention.

Compliance with ethics regulations: Yes.

Table 1 The distribution of the acute organophosphate poisonings according to the responsible agent

F-066 Long-term exposure to Sargassum-seaweed pollution in the French Caribbean Islands: clinical consequences and outcome

Dabor Resiere¹, Ruddy Valentino¹, Rishika Banydeen², Jonathan Florentin¹, Papa Gueye¹, Alain Blateau³, Hatem Kallel⁴, Bruno Megarbane⁵, Remi Neviere¹, Hossein Mehdaoui¹

¹Department of Critical Care, University Hospital of Martinique, F-97200 Fort-de-France, France; ²Department of Epidemiology & Methodology, University Hospital of Martinique, F-97200 Fort-de-France, France; ³ARS Martinique, Fort-De-France, France; ⁴Intensive Care Unit, Cayenne General Hospital, Cayenne, 97300 French Guiana, France; ⁵Department of Medical and Toxicological Critical Care, Lariboisière Hospital, Paris-Diderot University, INSERM UMRS1144, Paris, France

Correspondence: Dabor Resiere (dabor.resiere@chu-martinique.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-066

Rationale: Since the beginning of 2018, there has been an unexplained increasing invasion of Sargassum on the coast of Caribbean countries, including Martinique, Guadeloupe (1). Over an 8-month period, health officials in Guadeloupe and Martinique reported more than 12.000 such cases. Assault of these brown algae represents not only an environmental and economic disaster, but also a threat for human health. After 48 h on seashore, large amounts of toxic gas are produced by matter decomposition, including hydrogen sulfide (H2S) and ammoniac (NH3). The acute effects on humans after exposure to high concentrations of H2S are well described and of increasing severity with concentration, leading to potentially fatal hypoxic pulmonary, neurological and cardiovascular injuries (Table 1); however, the association of long-term exposure to Sargassum and health events is unknown. Although less documented, long term exposures may result in conjunctiva and upper airways irritation, headaches, vestibular syndrome, memory loss, and modification of learning abilities. In the absence of any available antidote, management of H2S intoxication relies on supportive care and prevention using individual protection. The objective of this study was to evaluate the clinical characteristics and consequences of long-term exposure to Sargassum among the local population.

Patients and methods: We conducted a prospective observational cohort study including all patients admitted to the emergency department at the University hospital of Martinique from March 2018 to December 2018 due to exposure to sargassum. Patients were managed according to the protocol established by the Research Group on Sargassum in Martinique. We assessed the patients exposure to Sargassum and air pollutants using 14 monitor located near of the patient’s residence. Demographics and clinical data (including cardiovascular, neurological and respiratory events) were collected. Data are presented as mean ± SD or %. Comparisons were performed using univariate analysis.

Results: In 8 months, 160 patients were included (age: 48 ± 20 years, 54 M/146 W, past history: hypertension (N = 25), diabetes (N = 29), asthma (14). Patients arrived with referral letter from their general practitioner (80%) and presented headaches (76%), developed gastrointestinal disturbances (79%), dizziness (54%), skin lesions (30%), cough (44%) and conjunctivitis (33%). Not all patients were clinically symptomatic. In the patients presented in June (14%), symptoms more frequently occurred in the workplace or at home (p < 0.05). Initial lung function tests were normal (50%). Three patients were admitted in intensive care unit.

Conclusion: Our study indicates that the magnitude of health effects following long-term exposure to Sargassum may be larger than previously recognized. Efforts to limit long-term exposure are mandatory.

Compliance with ethics regulations: Yes.

Table 1 Clinical Characteristics of acute hydrogen sulfide poisoning

F-067 Prognosis value of acute liver dysfunction after out-of-hospital cardiac arrest

Marie-Charlotte Delignette¹, Thomas Baudry¹, Marie Simon¹, Flavie Lavigne¹, Eric Bonnefoy-Cudraz², Laurent Argaud¹, Martin Cour¹

¹Médecine Intensive Réanimation-Hôpital Edouard Herriot, Lyon, France; ²USIC-Hôpital Louis Pradel, Lyon, France

Correspondence: Marie-Charlotte Delignette (marie-charlotte.delignette@chu-lyon.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-067

Rationale: Liver consequences of out-of-hospital cardiac arrest (OHCA) have been poorly studied. The aim of this study was to describe the characteristics of OHCA-induced acute liver dysfunction and its association with outcomes.

Patients and methods: We analyzed all consecutive OHCA patients admitted to two academic centers between 2007 and 2017. Patients treated with vitamin K antagonist were not included. Acute hepatocellular insufficiency (AHI), liver failure (LF) and hypoxic hepatitis (HH) were defined as a prothrombin (PT) ratio < 50%, a hepatic SOFA sub-score > 2 and an increase in transaminases > 20 times the normal values, respectively. Indocyanine green (ICG) clearance was used as the reference measure of liver function in a subset of patients. Multivariate logistic regression was used to identify potential risk factors for day 28 mortality.

Results: Of the 418 patients included (median age: 64 years; sex ratio: 1.4; nonshockable OHCA: 73%), 67 (16%), 3 (0.7%) and 61 (14.6%) presented AHI, LF and HH at admission, respectively. Death occurred before day 28 in 337 (80.6%) patients. Among the conventional static function liver tests, only bilirubin did not significantly differ between patients who survived or died (p = 0.91). ICG clearance, available in 23 patients, was abnormal in 17 (73.9%) cases. PT ratio was the only static test of liver function correlating significantly with ICG clearance (r = − 0.66, p < 0.01). After multivariate analysis, PT < 50% at admission (OR 10.63; 95% IC [1.92–200.9]), alkaline phosphatase at admission ≥ 100 UI/L (OR 5.26; 95% CI [2.29–13.23]), OHCA occurring at home (OR 3.83; 95% IC [1.85 8.11]), absence of bystander cardiopulmonary resuscitation (OR 2.94; 95% IC [1.40–6.39]), non-cardiac cause of OHCA (OR 5.22; 95% IC [2.48–11.43]) and low flow duration ≥ 20 min (OR 7.26; 95% IC [3.28–17.45]) were independently associated with day 28 mortality. As shown in Fig. 1, survival significantly differed according to the degree of decrease in PT ratio (p < 0.0001, log rank test).

Conclusion: Early AHI, as defined by PT < 50% at ICU admission, occurred frequently after OHCA and was independently associated with day 28 mortality.

Compliance with ethics regulations: Yes.

Fig. 1

Survival according to PT ratio at ICU admission

F-068 Using neuron-specific-enolase (NSE) for prediction of severe brain damage after cardiac arrest: insights from the Parisian Registry

Guillaume Savary¹, Antonin Ginguay², Phuong-Nhi Bories², Paul Jaubert¹, Ariane Gavaud¹, Elodie Baron¹, Pierre Dupland¹, Jean-Paul Mira¹, Florence Dumas³, Alain Cariou¹

¹Medical Intensive Care Unit, Cochin Hospital, APHP, Centre—Université de Paris, Paris, France; ²Biochimie, Cochin Hospital, APHP, Centre—Université de Paris, Paris, France; ³Emergency Department, Cochin University Hospital, APHP, Paris, France

Correspondence: Guillaume Savary (guillaume.savary@aphp.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-068

Rationale: Neuron-specific-enolase (NSE) is commonly used as a biomarker reflecting the extent of brain injury in different settings. In post-cardiac arrest patients, previous clinical studies reported that an increase in NSE was predictive of a poor outcome but did not specifically focused on neurological outcome. In this prospective study, we aimed to determine the NSE performance for prediction of severe brain damage in post-cardiac arrest patients.

Patients and methods: All consecutive patients admitted in our ICU after cardiac arrest between January 2017 and February 2019 that were still comatose at H48 and had at least one measurement of serum NSE were included. Blood samples for NSE measurement were serially collected at 48 (H48) and 72 h (H72) after cardiac arrest and serum NSE levels were measured within 4 h. We used the following criteria for the definition of severe brain damage (primary endpoint): Cerebral Performance Categories (CPC) 3 or 4 level at discharge, brain death or withdrawal of life-sustaining treatments (WLST) based on neurological status. We also assessed the predictive value of serum NSE using all-cause mortality as a secondary endpoint.

Results: During the study period, 155 patients were available for the analysis. They were mostly male (66.9%), with an age of 60.6 years. Among these patients, 83 (50.3%) had a good neurologic outcome (CPC 1–2) and 68 patients were classified as having a severe brain damage (52 WLST based on neurological status, 11 brain deaths and 5 survivors with CPC 3–4). In univariate analysis, patients with severe brain damage less frequently received bystander CPR, had longer duration of no-flow, less initial shockable rhythm, more post-resuscitation shock and higher NSE values: mean at H48 were 217.3 versus 26.2; and 270.1 versus 21.4 at H72 (p < 0.001). NSE levels at H48 and H72 were strong predictors of severe brain damage (AUC of 0.948 and 0.972 respectively, Figure 1) and also predicted all-cause mortality (AUC of 0.896 and 0.931 respectively). To predict severe brain damage with 100% specificity, best NSE cutoff values at H48 and H72 were 52.3 and 53.2 µg/l, with a sensitivity of 77.6 and 82.9% respectively.

Conclusion: A high serum NSE measured at H48 and H72 after cardiac arrest accurately predicted severe brain damage with a high specificity. Our results support the use of NSE for neuroprognostication after cardiac arrest, in combination with other predictors.

Compliance with ethics regulations: Yes.

Fig. 1

NSE Receiver-Operating curves

F-069 Availability of psychologists in French-speaking intensive care units

Jean-Baptiste Lascarrou¹, Saber Davide Barbar², Florence Boissier³, Guillaume Decormeille⁴, Nicholas Heming⁵, Bertrand Hermann⁶, Sami Hraiech⁷, Gwenaëlle Jacq⁸, Jean-FrançOis Llitjos⁶, Lamia Besbes⁹, Laurent Poiroux¹⁰, Xavier Monnet¹¹, Gaël Piton¹²

¹Medical Intensive Care Unit, Nantes, France; ²Medical Intensive Care Unit, Nîmes, France; ³Medical Intensive Care Unit, Poitiers, France; ⁴Medical Surgical Intensive Care Unit, Toulouse, France; ⁵Medical Intensive Care Unit, Garches, France; ⁶Medical Intensive Care Unit, Paris, France; ⁷Medical Intensive Care Unit, Marseilles, France; ⁸Medical Surgical Intensive Care Unit, Versailles, France; ⁹Medical Intensive Care Unit, Monastir, Tunisia; ¹⁰Medical Intensive Care Unit, Angers, France; ¹¹Medical Intensive Care Unit, Le Kremlin-Bicêtre, Francel; ¹²Medical Intensive Care Unit, Besançon, France

Correspondence: Jean-Baptiste Lascarrou (jeanbaptiste.lascarrou@chu-nantes.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-069

Rationale: The psychological care of patients, their relatives and of healthcare workers is a major issue in the intensive care unit (ICU). Psychologists may provide emotional support during trying times. The intervention of a psychologist may alleviate long term mental health issues such as post-traumatic stress disorder. The main objective of our study was to describe the availability of psychologists in French-speaking ICUs.

Patients and methods: Internet survey conducted between March and May 2019 using SurveyMonkey (San Mateo, USA). Survey consisting of 20 questions sent to subscribers of the SRLF mailing list via MailChimp software (Atlanta, USA). Frequencies and percentages were determined for categorical variables and median and interquartile range for continuous variables. The ICUs with or without psychologist were compared using nonparametric Fisher exact test. STATA 14 used (Lakeway Drive, TE, USA).

Results: 263 responses were obtained from 197 unique ICUs in France (n = 182), Belgium (n = 6), Switzerland (n = 5), Algeria (n = 2), Morocco (n = 1) and Tunisia (n = 1). 164 (82%) ICUs were part of public hospitals, 33 (18%) of private facilities. 187 (94%) ICUs cared for adult patients, 10 (6%) for children. The median number of beds was 20 [15–26]. 79 (40%) ICUs were open to visitors 24/7, 46 (23%), to visitors > 8 h/day and 72 (37%) to visitors < 8 h/day. Psychological consults were established in 89 (45%) wards (2 ICUs did not answer). Pediatric ICUs employed more psychologists than adult ICUs (P = 0.006). Comparison of ICUs based on the presence or not of a psychologist appears in Table 1. In ICUs where a consulting psychologist is available, their effective availability is 0.5 [0.2–1] full time equivalent. Consults are delivered to: patients (100%), families (100%) or healthcare workers (60%). Out of the 106 ICUs without a psychological consult, responders from 5 (2.5%) ICUs believe that a psychological consult is undesirable. Out of the 106 ICUs without psychological consult, 24 (22%) responders cannot obtain a psychological consult, whatever the circumstances, 41 (39%) can require an outside psychological consult when needed, while 41 (39%) can require assistance from a psychologist working in another unit (several answers possible for each respondent).

Conclusion: Psychologists consult in only half of adult ICUs but in almost all pediatric ICUs. 22% of ICUs are unable to provide a psychological consult. Psychological consults are delivered in similar proportions to patients, their family and to a lesser extent to healthcare workers. Responders from 2.5% ICUs without an established psychological consult believe that the availability of a psychologist is undesirable.

Compliance with ethics regulations: NA.

Table 1 .

F-070 Should Intermediate care units and intensive care units be 2 separate units?

Juliette Audibert, Adel Ben Salah, Alexandre Conia, Olivier Gontier, Mouldi Hamrouni, Cecile Jourdain, Benedicte Mauchien, Pierre Kalfon

Hôpitaux de Chartres, Chartres, France

Correspondence: Juliette Audibert (juliettesouhaid@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-070

Rationale: Comfort of patients in Intensive Care Unit (ICU) is now a real concern for the healthcare teams. Perceived patient discomfort assessment is a daily practice for our staff. The primary objective of our study was to assess whether the overall discomfort score reported by patients hospitalized in a separate intermediate care unit differs from that reported by patients hospitalized in ICU.

Patients and methods: A tailored multicomponent program consisting of assessment of ICU-related self-perceived discomforts with a 18-item questionnaire, immediate and monthly feedback to healthcare teams and site-specific tailored interventions, was applied in our department, located in a general hospital, and comprising a 12-bed ICU and a separate 6-bed intermediate care unit. From May 1, 2018 to June 30, 2019, all patients, who were hospitalized in the ICU or in the intermediate care unit, aged 18 years or older, who survived a stay of 3 calendar days or more, were eligible for inclusion and were asked to answer the 18-item questionnaire.

Results: During this period of 13-months period, 173 ICU patients whereas 163 intermediate care unit patients were included in the study with complete questionnaire. Age and sex ratio were similar between the 2 groups whereas the length of stay and IGSII were higher in ICU. The overall discomfort score was higher in the ICU group than in the intermediate care unit group (23 ± 15 vs. 18 ± 11, p = 0.005) as well as scores for thirst (3.43 ± 3.61 vs. 2.53 ± 3.53, p = 0.016), feeling of cold (2.31 ± 3.18 vs 0.70 ± 1.81, p < 10–3), dyspnea (3.43 ± 3.52 vs. 2.07 ± 2.76, p = 0.001), lack of phone (1.50 ± 2.86 vs. 0.43 ± 1.57, p < 10–3), isolation (1.70 ± 2.80 vs. 0.99 ± 2.18, p = 0.020) and depression (2.39 ± 3.23 vs. 1.34 ± 2.49, p = 0.002). In the separate intermediate care unit, the only 2 following items reported with higher scores than in the ICU were feeling of heat (2.01 ± 3.13 vs. 0.96 ± 2.18, p = 0.001), probably because of the absence of air conditioner, and bed-related discomfort (2.63 ± 3.18 vs. 1.90 ± 2.97, p = 0.009).

Conclusion: Overall discomfort score was lower in our separate intermediate care unit than in our ICU. In the future, intermediate care units and intensive care units should be 2 separate units to ensure maximum comfort for our patients.

Compliance with ethics regulations: Yes.

F-071 Evaluation of an early post-ICU discharge consult program on handoff quality

Camille Vissac¹, Cécilia Tabra¹, Rusel Leon¹, Marina Axus¹, Morgane Commereuc¹, Frédérique Schortgen¹, Jerome Cecchini¹

¹Réanimation et surveillance continue adulte, Hôpital intercommunal de Créteil, Créteil, France

Correspondence: Camille Vissac (camille.vissac@chicreteil.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-071

Rationale: The transition period surrounding the discharge from ICU to hospital ward is a critical period in the course of the patient. Handoff of complex patients is at high risk for communication failures between providers, inaccurate cares and ICU readmission. A transition program including a post ICU follow-up has been proposed to improve handoff quality. Post ICU consults by ICU team represent, also, an opportunity for improving feedback on the quality of ICU cares. The goal of the present study is to assess the feasibility and the impact of a systematic early post-ICU consult (EPICUC) program on handoff quality in a 14 bed mixed ICU.

Patients and methods: Before the development of the EPICUC program, standardized handoffs were already applied including identified day and hour of discharge and both verbally communicate and written medical and nurse information for receiving team. From 1st March to 30th October 2019, all patients who were discharged to the ward of our hospital were candidates for EPICUC. EPICUC were performed by ICU staff (at least one ICU physician) within the 3 days following discharge. The EPICUC consisted of a face-to-face discussion with the receiver team to assess the accuracy, completeness and understanding of passing information and of a patient visit. A standardized form was used for collecting data. The impact of EPICUC on handoff quality was assessed by the number of communication failures and the number of patients in whom EPICUC resulted in a management change. Personal feeling of EPICUC providers on its usefulness was assessed by a 0–10 rating scale.

Results: Among the 209 candidates for EPICUC, 4 were dead and 74 already discharged alive from hospital at EPICUC time. EPICUC were performed in 131 patients (63%) within 4 ± 3 days after ICU discharge. 85 EPICUC (65%) were performed by both, nurse and ICU physician. 111 (85%) patients and 84 receiver teams (64%) were available at EPICUC time. EPICUC duration was 11 ± 4 min. A communication failure was identified in 38 EPICUC (29%), either a rectification of passing information (n = 30; 23%) and/or a change in patient management (n = 19; 15%). The usefulness of the EPICUC was rated at 4 ± 3 and 5 ± 3 by ICU physicians and nurses, respectively.

Conclusion: The time spent for EPICUC appears reasonable. EPICUC identified a communication failure in one-third of handoffs and allowed care readjustment in one quarter of patients. Factors associated with handoff failures will be presented during the congress.

Compliance with ethics regulations: Yes.

F-072 Patients’ perceptions of quality of life after ICU discharge

Danielle Prevedello, Claire Steckelmacher, Marianne Devroey, Jean-Charles Preiser

Hopital Erasme, Bruxelles, Belgium

Correspondence: Danielle Prevedello (danielle.prevedello@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-072

Rationale: Surviving a critical illness is a challenging condition for patients and relatives. The psychological aspects are directly affected by physical status and performance. Patients can feel depressed or anxious facing difficulties during recovery time. The aim of this study was to correlate patients’ perceptions of his health status and his clinical performance measured after ICU discharge.

Patients and methods: This is a prospective pilot study of an ICU follow-up clinic conducted in a single center from January 2018 to July 2019. This clinic is multidisciplinary and includes two visits at 3 and 6 months after ICU discharge. Patients with more than 5 days of ICU LOS were eligible. All patients at 3 and 6-m visit were evaluated with SF-36, 6MWT, MRC and time-up-and-go test. We conducted an analysis comparing clinical performance data and qualitative data between 3 and 6 months after ICU discharge.

Results: The investigation included 90 patients who had at least 5 days of ICU length of stay. 50 patients attended the consult at 3-m and 28 patients attended the consult both times. The median age (IQR) was 63 (54–71) and 56% were men. 48%, 30% and 22% of patients had medical, scheduled surgical and emergency surgical admission causes respectively, with median (IQR) SAPS III score 53 (44–73). 38%, 14% and 80% of patients had sepsis, delirium and mechanical ventilation as a support. The physical status was progressively increased overtime likewise the physical capacity assessed by SF-36 score with p-value 0.007 between 3 and 6-m. However, no significant difference between the subjective dimension of SF-36, which analyses the perception of the patient about his physical capacity, assessed at 3-m and at 6-m was demonstrated (p 0.35).

Conclusion: In this pilot-phase of following a cohort of critically ill patients, the natural physical improvement does not seem to change the patient’s perception of their performances. This paradigm rouses a different perspective that should take into account when setting up rehabilitation programs.

Compliance with ethics regulations: Yes.

F-073 Post-traumatic stress disorder after discharge from an acute medical unit

Basma lahmer¹, naoufel madani^1,2, jihane belayachi^1,2, redouane abouqal^1,2

¹Acute Medical Unit, Ibn Sina University Hospital, Rabat, Morocco; ²Laboratory of Biostatistics, Clinical, and Epidemiological Research, Faculty of medicine and pharmacy—university Mohammed V, 10000, Rabat, Morocco

Correspondence: Basma lahmer (basma.lahmer.93@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-073

Rationale: Post-Traumatic Stress Disorder (PTSD) occurs after exposure to a traumatic event and comprises of symptoms of repeated re-experiencing of the said event, avoidance of reminders, emotional numbing and persistent hyperarousal. In individuals exposed to “medical stress”, various studies found evidence of PTSD occurring after the onset, diagnosis, or treatment of physical illness. Our study aims to determine PTSD’s risk factors in patients of an acute medical unit (AMU) after their discharge.

Patients and methods: It was a prospective, analytical study conducted over a period of 2 months at an acute medical unit. We collected sociodemographic and clinical data, patients’ medical history, and evaluated the symptoms of anxiety and depression during their stay using the Hospital Anxiety and Depression Scale (HADS). The prevalence of severe PTSD symptoms was assessed with the Impact of Events Scale-Revised (IES-R) at 6 weeks and 3 months using a cutoff of 33. Associations between PTSD as evaluated by IES-R at 3 months and patients’ characteristics, including HADS scores at admission were investigated using unadjusted linear regression, for univariate and multivariate regression analysis. Statistical analyses were carried out using SPSS for Windows (SPSS, Inc., Chicago, IL, USA).

Results: We included 141 patients in our study with a mean age of 49.74 ± 17.8. In our population, 20.8% of patients scored higher than a 33 IES-R cutoff at 6 weeks compared to 17.6% at 3 months. The mean HADS-Anxiety score is 7.8 ± 6 and that of the HADS-Depression score is 7.1 ± 6.3. On one hand, higher HADS-anxiety Score during the stay in the AMU was linked to higher IES-R scores at 3 months β: 1.47 IC95% (0.61; 2.26) p < 0.001. On the other hand, living alone but with family in close proximity β: − 14.1 IC95% (-25.22; − 2.97) p = 0.01 and a longer stay in the AMU β: − 0.37 IC95% (− 0.71; − 0.03) p = 0.03 were found to be predictive of less severe PTSD symptoms.

Conclusion: We identified that patients who had shown less significant anxiety symptoms during their stay, lived alone with family nearby and stayed longer in the AMU were less at risk of developing PTSD after their discharge.

Compliance with ethics regulations: Yes.

F-074 Quadriceps strength of critically ill patients: how weak are they?

Anne Meunier, Jean-Louis Croisier, Isabelle Kellens, Benoit Misset, Jean Joris, Anne-Françoise Rousseau

University Hospital of Liège, Liège, Belgium

Correspondence: Anne Meunier (anne.meunier@student.uliege.be)

Ann. Intensive Care 2020, 10 (Suppl 1):F-074

Rationale: Objective of critical care includes restoration of functional capacities. Prompt identification of muscle acquired weakness (ICU-AW) is crucial to target efficient rehabilitation. In published literature, data of quadriceps strength (QS) cannot be compared because of insufficient standardization of measurement protocols. We recently validated a highly standardized protocol of QS measurement. In order to build basic and comparable knowledge and to identify the weakest patients, this study aimed to describe QS of critically ill (CI) patients during their short-term evolution, and to compare them to surgical (S) and healthy (H) subjects.

Patients and methods: This observational study included CI patients who spent at least 2 days in ICU, patients scheduled for elective colorectal surgery (S) and young healthy volunteers (H). Maximal isometric QS was assessed using a handheld dynamometer (MicroFet2^®) and expressed in Newton/kg (N/kg). Dominant leg was tested in supine position using a highly standardized procedure. CI and S patients were tested at T0 (as soon as collaborative in ICU) and 1 month after discharge (M1). Medical Research Council (MRC) test and Barthel questionnaire were used at same time points. Data are expressed as median (min–max).

Results: 38, 32 and 34 subjects were included in CI, S and H groups, respectively. Demographic data (age, sex, body mass index) were similar in CI and S groups. At T0, QS was lower in CI group than in S and H groups: 3.01 (0.77–7.61), 3.6 (2.26–5.98) and 5.5 (3.19–7.96) N/kg, respectively. At T0, only 13 CI patients (34%) had a MRC < 48 (definition of ICU-AW) but their QS was not lower than those with MRC ≥ 48. QS did not improve at M1 in CI patients, especially if previously sedentary (Fig): QS was 2.96 (1.32–5.33) N/kg. CI patients with QS < 2.1 N/kg (i.e. the weakest QS in S patients) remained more dependent at M1 than those with QS ≥ 2.1 N/kg: Barthel was 87.5 (30–100) vs 100 (40–100), respectively (p = 0.0103). QS at M1 in CI group was significantly associated with QS at T0 (r = 0.67, p < 0.0001) and inversely correlated with age (r = − 0.39, p = 0.0173).

Conclusion: Muscle function is a key component of post-ICU recovery. A huge QS heterogeneity was observed among CI patients. QS in ICU was correlated with QS at M1, but was not consistent with ICU-AW diagnosis based on MRC. Special attention should probably be paid to CI patients previously sedentary and to patients with QS < 2.1 N/kg: they were weaker or more dependent at M1.

Compliance with ethics regulations: Yes.

Fig. 1

QS change at M1 in CI patients according to previous physical activity

F-075 Healthcare trajectories before and after critical illness: population-based insight on diverse patients clusters

Youenn Jouan¹, Leslie Grammatico-Guillon², Antoine Guillon¹, Stephan Ehrmann¹

¹Service de Médecine Intensive Réanimation, CHU de Tours, Tours, France; ²Service d’Information Médicale, d’Epidémiologie et d’Economie de la Santé, Tours, France

Correspondence: Youenn Jouan (youenn.jouan@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-075

Rationale: The post intensive care syndrome (PICS) gathers various disabilities, associated with a substantial healthcare use. However, patients’ comorbidities and active medical conditions prior to intensive care unit (ICU) admission may partly drive healthcare use after ICU discharge. To delineate the relative contribution of critical illness and PICS per se to post-critical illness increased healthcare use, as opposed to pre-existing comorbidities, we conducted a population-based evaluation of patients’ healthcare use trajectories.

Patients and methods: Using discharge databases in a 2.5-million-people region in France, we retrieved, over three years, all adult patients admitted in ICU for septic shock or acute respiratory distress syndrome (ARDS), intubated at least 5 days and discharged alive from hospital. Healthcare use (days spent in healthcare facilities) was analyzed two years before and two years after ICU admission. Healthcare trajectories were next explored at individual level: patients were assembled according to their individual pre-ICU healthcare use trajectory by clusterization with the K-Means method.

Results: Eight-hundred and eighty-two (882) patients were included. Median duration of mechanical ventilation was 11 days (interquartile ranges [IQR] 8; 20), mean SAPS2 was 49, and median hospital length of stay was 42 days (IQR 29; 64). Prior to ICU admission, we observed, at the scale of the whole study population, a progressive increase in healthcare use. However, clusterization of individual according to pre-ICU healthcare trajectories identified patients with elevated and increasing healthcare use (n = 126), and two main groups with low (n = 476) or no (n = 251) pre-ICU healthcare use. Patients with high healthcare use had significantly more comorbidities than those with low healthcare use. In ICU, however, SAPS2, duration of mechanical ventilation and length of stay were not different across the groups. Interestingly, analysis of post-ICU healthcare trajectories for each group revealed that patients with low or no pre-ICU healthcare (which represented 83% of the population) switched to a persistent and elevated healthcare use during the two years post-ICU.

Conclusion: For 83% of ARDS/septic shock survivors, critical illness appears to have a pivotal role in healthcare trajectories, with a switch from a low and stable healthcare use prior to ICU, to a sustained higher healthcare recourse two-years after ICU discharge. This underpins the hypothesis of long-term critical illness and PICS-related quantifiable consequences in healthcare use, measurable at a population level.

Compliance with ethics regulations: Yes.

Fig. 1

Healthcare use days (% of total days per quarter spent in acute care settings) during the pre-ICU period and the post-ICU period (B) after clustering based on total pre-ICU healthcare use

F-076 Prehospital triage in severely traumatized children: experience of a French regional trauma system

Jordan Porteaud¹, Guillaume Mortamet², Francois-Xavier Ageron³, Pierre Bouzat¹

¹Département d’anesthésie, CHU de Grenoble, Grenoble, France; ²Service de réanimation pédiatrique, CHU de Grenoble, Grenoble, France; ³CH Annecy, Annecy, France

Correspondence: Jordan Porteaud (jporteaud@chu-grenoble.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-076

Rationale: Traumatic injuries are the leading cause of death and a major cause of disability in children. Prehospital triage is a key element in the management of trauma patients. This study aims (1) to describe the pre-hospital grading protocol developed by the Northern French Alps Emergency Network (TRENAU) for children, (2) to evaluate its quality to detect the most severe trauma patients and (3) to assess the accuracy of this procedure to perform an adequate triage.

Patients and methods: Our regional trauma system included 13 hospitals categorized as Level I, II or III pediatric trauma centers. Each patient was graded A, B or C by an emergency physician, according to the seriousness of their injuries at presentation on scene. The triage was performed according to this grading and the categorization of centers. This study is a registry analysis of an 8-year period (2009 to 2017).

Results: A total of 1142 children (mean age 10 years, 65% were boys) with severe trauma were included in the cohort. Fifty-seven, 22% and 21% of patients were admitted to a level I, II and III, respectively. Road accident was the main mechanism of injury (35% of patients). Thirty-six percent of patients had a severe trauma, defined as an Injury Severity Score (ISS) higher than 15. One quarter of patients had at least 2 severe lesions and one-third of patients had a trauma brain injury. The pre-hospital gradation was closely related with injury severity score (ISS) and intra-hospital mortality rate. The triage protocol had a sensitivity of 78% and a specificity of 33% to predict adequate admission of patients with ISS more than 15. Using a specific trauma score (including occurrence of death, an admission in Intensive Care Unit and the need for urgent surgery), sensitivity and specificity reached 91 and 41%, respectively. Fourty-six percent of patients were not graded at the scene (Non-graded group). Undertriage rate was significantly reduced in the Graded group compared with the Non-graded group, (22% versus 45%), without significant modification of the overtriage rate (43% versus 45%). Overall, mortality at discharge from hospital was 2%, but 33% in Grade A patients.

Conclusion: Implementation of a regional pediatric trauma system with a specific pre-hospital triage procedure was effective in detecting severe pediatric trauma patients and in lowering the rate of pre-hospital undertriage.

Compliance with ethics regulations: Yes.

F-077 Glomerular filtration rate estimation formulas for assessing piperacillin clearance in critically ill children

Agathe Béranger¹, Sihem Benaboud², Saïk Urien³, Thao Nguyen Khoa⁴, Inès Gana², Julie Toubiana⁵, Yi Zheng², Fabrice Lesage⁶, Sylvain Renolleau⁸, Déborah Hirt², Jean-Marc Treluyer², Mehdi Oualha¹

¹Réanimation pédiatrique, Necker Enfants Malades, Paris, France; ²Pharmacologie clinique, Cochin, Paris, France; ³URC, Cochin-Tarnier, Paris, France; ⁴Biochimie générale, Necker Enfants Malades, Paris, France; ⁵Pédiatrie générale, Necker Enfants Malades, Paris, France; ⁶Réanimation pédiatrique, Necker, Paris, France

Correspondence: Agathe Béranger (agathe.beranger@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-077

Rationale: Critically ill children suffer from pathophysiological changes, leading to large between-subject variability in drug clearance. Since piperacillin is eliminated mainly via the kidney, changes in renal function go along with a modified elimination, and possible sub-therapeutic or toxic drug concentrations. We aimed to determine the most accurate glomerular filtration rate (GFR) estimation formula for assessing piperacillin clearance in critically-ill children.

Patients and methods: All children hospitalized in pediatric intensive care unit and receiving piperacillin were included. Piperacillin was quantified by high performance liquid chromatography. Pharmacokinetics were described using the non-linear mixed effect modeling software MONOLIX. In the initial pharmacokinetics model, GFR was estimated according to the Schwartz 1976 formula. In the study, GFR was estimated with 11 additional formulas, developed with plasma creatinine and/or cystatin C. Biases, precisions, Spearman’s rank correlation coefficient and normalized prediction distribution error (NPDE) were used to assess the models.

Results: We included 20 children with a median (range) postnatal age of 1.9 (0.1–19) years, body weight of 12.5 (3.5–69) kg and estimated GFR according to the Schwartz 1976 formula of 160.5 (38–315) mL min–1.1.73 m². Piperacillin concentrations were best predicted with the model using the creatinine clearance. The correlations were most accurate: r2 = 0.7 between the population-predicted and the observed concentrations, r2 = 0.0004 and r2 = 0.04 for the NPDE versus population-predicted concentrations and time, respectively. Concerning the individual predicted concentrations, bias and precision were respectively − 4.1 mg L⁻¹ and 14.7 mg L⁻¹. GFR estimations based on serum creatinine were higher than those based on cystatin C (p = 0.02).

Conclusion: In summary, the 12-h creatinine clearance is the best predictor of piperacillin clearance and this could be investigated for drugs with renal elimination. As a whole, literature and our findings strongly suggest using creatinine clearance to also estimate GFR in critically ill children. The gap between the GFR estimations is large depending on the formulas, with higher estimations with equations based on serum creatinine.

Compliance with ethics regulations: Yes.

Fig. 1

Boxplots of the glomerular filtration rate estimates in mL min⁻¹ 1.73 m⁻² according to 12 formulas. Boxplots of the glomerular filtration rate estimates in mL min⁻¹ 1.73 m⁻² according to 12 formulas. In blue, yellow and green, formulas depending respectively on creatinine, cystatin C and both. In dotted red line, the limit defining augmented renal clearance (ARC), at 130 mL min⁻¹ 1.73 m⁻²

F-078 Acute pancreatitis in the pediatric intensive care unit: French retrospective study

Elsa Amouyal¹, Julie Sommet¹, Jerome Rambaud², Anna Deho¹, Jérôme Naudin¹, Michael Levy¹, Géraldine Poncelet¹, Fleur Le Bourgeois¹, Guillaume Geslain¹, Arielle Maroni¹, Stéphane Dauger¹, Maryline Chomton¹

¹Robert Debré Hospital, Paris, France; ²Trousseau Hospital, Paris, France

Correspondence: Elsa Amouyal (elsa_amouyal@hotmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-078

Rationale: Acute pancreatitis (AP) incidence have increased dramatically over the past years. New guidelines in 2018 were recently published in order to standardize the definition and management of AP. The aim of this study is to describe the management of children that were diagnosed with AP from the pediatric intensive care unit (PICU) in two French hospitals.

Patients and methods: This retrospective cohort study included children aged under 18 years old, who were admitted to the PICU of Robert-Debré hospital and Trousseau from 2006 to 2018 with a discharge diagnosis of AP. Data collected included management, severity and outcomes. We have also obtained data on clinical, biological and radiological presentation.

Results: Sixty patients were included, the median age was 8 years (5–14) and 75% had a co-morbidity mainly hematologic (26/60). Most of the AP were moderate (52%) or severe (45%). Hemodynamic failure was the main reason for PICU admission requiring a median fluid resuscitation 56 ml/kg complemented by a median intravenous fluid therapy of 4 ml/kg/h (2–5) during the first 24 h. Twenty patients (37%) required mechanical ventilation. Fasting has been instituted in 59 patients (98%) for a median of 4 days (1–6), whereas 54 patients (90%) received parenteral nutrition, only 18 patients (30%) received enteral nutrition. Antibiotic therapy was given to 53 patients (88%) including 13% for curative therapy. The median length of stay in PICU was 4 days (2–6). The mortality rate was 13%.

Conclusion: This is the first French study which precisely described the management of patients with AP in PICU. It highlighted the differences with the new international guidelines. This study could improve the management of PA in PICU and open research perspectives.

Compliance with ethics regulations: Yes.

F-079 Therapeutic plasma exchange for immunologic disorders: a well-tolerated and safe process for pediatric life-threatening conditions

Jean-Sebastien Diana, Sandra Manceau, Tioka Rabeony, Caroline Elie, Anne-Marine Lenzotti, Valerie Jolaine, Melodie Aubart, Pierre Tremolieres, Nadege Salvi, Christine Bodemer, Brigitte Bader-Meunier, Christine Barnerias, Franck Iserin, Christophe Chardot, Florence Lacaille, Sylvain Renolleau, Remi Salomon, Francois Lefrere, Mehdi Oualha, Marianne Delville, Laurent Dupic

APHP, Paris, France

Correspondence: Jean-Sebastien Diana (jean-sebastien.diana@aphp.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-079

Rationale: Apheresis and Therapeutic plasma exchange (TPE) for children diseases has been poorly investigated in mostly small-uncontrolled studies. The purpose of this study is to describe indications and safety of TPE in children.

Patients and methods: In this single center and retrospective study, we included 94 patients who underwent TPE with an age < 18 years old in the pediatric center of Necker-Enfants-Malades hospital from January 2005 to December 2014. Data were retrospectively collected in an electronic case report form via a web-based data collection system.

Results: 78 patients with a median age of 9.8 years [range 0.53; 17.93] were selected. They achieved a total number of 731 procedures. Indications were antibody-mediated rejection (n = 33; 42%) or desensitization therapy (n = 5; 6%) for solid organ or hematopoietic transplantations; microangiopathy (n = 17; 22%); renal diseases (n = 6; 8%) and pediatric inflammatory diseases (n = 16; 21%); or hyperviscosity syndrome (n = 1; 1%). Each patient had an average of 6 procedures for the first session [range 1; 19] with a median volume of 1834 ml [range 500; 5000 ml] corresponding to a median (rang) total plasma volume (TPV) equivalent of 1.39 l/m2 [0.58–2.1]. Within 15 days since the beginning of sessions, 72 patients (92%) present a total of 311 Adverse Events (AEs) potentially related to TPE. There was a median (range) of 5 AEs/patients [0–24]. There was no association between AEs and diseases, severity of patients, venous access, plasma substitute and body weight. Few of AEs (n = 23 for 72 patients) were potentially life-threatening and concerned mostly critically ill children. Allergic reactions represented only 20 AEs for 14 patients (grade I n = 18; grade II n = 1; grade III n = 1). At the 12 months endpoint, 15 (19%) patients died and 10 (13%) patients had severe persistent disease. No death had been related to the TPE process.

Conclusion: We describe one of the largest retrospective pediatric cohort updated to the last international recommendations. TPE in children is performed for specific and potentially refractory disease. It is feasible without a major risk of life threatening adverse events.

Compliance with ethics regulations: Yes.

F-080 Epidemiology and risk factors of infections on catheters with implantable chamber in the pediatric population

Yacine Benhocine

University hospital nedir mohamed, Tizi-Ouzou, ALGERIA

Correspondence: Yacine Benhocine (yacine001@yahoo.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-080

Rationale: Although analysis of literature data shows that implantable chamber catheters (ICCs) are less at risk of infectious complications than other central venous catheters, these complications can be serious, which may differ from ongoing treatments such as chemotherapy, and may lead to the removal of the implanted device. The literature on preventing these infections is quite disparate, as practices. Purpose: to evaluate the incidence of infections, to identify responsible germs and to measure the impact of preventive measures.

Patients and methods: Prospective, descriptive, mono-centric study, from January 2012 to January 2019. All patients under the age of 15 who have benefited from an implantable chamber catheter, whose insertion procedure is as follows: local anesthesia, surgical asepsis (polyvidone iodine) in an operating room, double disinfection, no antibiotic prophylaxis, routes used: subclavian (89%), internal jugular (11%) by anatomic registration. The main criteria of judgment are: the incidence of local and general infections, their time of onset, responsible microorganisms. Statistical analysis used the Statistical Package for the Social Sciences software.

Results: 4480 patients were included, the average incidence density of early infection is 0.61/1000 day-catheters. The time of onset of infection is essentially between the 2nd and 3rd week post-exposure, of which 20% is general infection. Ablation involved 52% of infected catheters. The causative organisms are mainly Gram-positive cocci (67.99%), Gram-negative bacilli are less involved (23.33%), with a significant number of candida infections (8%).

Discussion: Higher incidence of data from the literature. To remedy this requires the implementation of additional hygiene measures: antiseptic showers preoperatively, Chlorhexidine??, and practice changes: echo guidance, antibiotic prophylaxis or locks? Second generation catheters? Our practices are disparate especially since the recommendations specifically concerning the prevention of infectious risk associated with internationally published ICCs are rare.

Conclusion: At the end of this work, our perspectives are to: update the procedure, highlight risk factors on which it is possible to act, the adhesion of the different staff to the protocols.

Compliance with ethics regulations: Yes.

F-081 Effect of a compliant bundled care on severity and mortality in children with severe sepsis hospitalized in PICU. Retrospective analysis of the DIABACT III study

Camille Beaucourt¹, Fleur Lorton², Elise Launay², Florent Baudin³, Christèle Gras-Leguen², Etienne Javouhey⁴

¹CHRU Jean Minjoz, Besancon, France; ²Hôpital Mère Enfant, Nantes, France; ³Hôpital Femme Mère Enfant, Bron, France; ⁴Hôpital Femme Mère Enfant, Lyon, France

Correspondence: Camille Beaucourt (camille.beaucourt91@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-081

Rationale: The 2014 sepsis and septic shock pediatric guidelines advise to treat patients using care bundles. In the first hour, the «Resuscitation bundle» contains an appropriate fluid resuscitation, a broad-spectrum antibiotics administration after blood cultures, and initiation of inotrope if needed. The objectives were to evaluate the Resuscitation bundle compliance in a cohort of septic children with cardiovascular dysfunction, and to analyze the effect on severity and outcome in pediatric intensive care unit (PICU).

Patients and methods: Retrospective analysis of the DIABACT III study. This study analyzed the care course of children with severe community-acquired bacterial infection, hospitalized in PICUs in France’s west departments, between August 2009 and January 2014. Children with severe sepsis and cardiovascular dysfunction were retrospectively included.

Results: We included 92 children of whom 6 (6.5%) had compliant bundled care. The severity scores at PICU’s admission were similar between groups (p = 0.55 for the PRISM score and 0.58 for the PELOD 2). There was the same proportion of fluid-refractory shock (p = 0.65), mechanical ventilation (p = 1.0), neurological dysfunction (p = 1.0) and cardiac arrest (p = 0.39). In the «Resuscitation bundle compliant» group, 33.3% died versus 15.1% in the other group (p = 0.46). We highlighted a severity bias: the sickest patients were more likely to receive compliant bundled care.

Conclusion: In our cohort, the Resuscitation bundle’s compliance was low. We did not show some effect on morbidity nor mortality. However, this study helps understand the factors associated with Resuscitation bundle’s compliance.

Compliance with ethics regulations: Yes.

F-082 Epidemiology of nosocomial infections with extended-spectrum β-lactamase producing gram-negative bacilli in a neonatal and paediatric Tunisian intensive care unit

Ahmed Ayari¹, Asma Bouziri¹, Ahmed Hajji¹, Shatila Hadj Hassine¹, Assaad Louati¹, Khaled Menif¹, Aïda Borgi¹, Nejla Ben Jaballah¹

¹Service de Réanimation Pédiatrique Polyvalente/Bechir Hamza Children’s Hospital of Tunis, Tunis, Tunisia

Correspondence: Ahmed Ayari (Ahmed.alayari@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-082

Rationale: Nosocomial infections with extended-spectrum β-lactamase (ESBL) producing gram-negative bacilli (GNB) are an important cause of hospital morbidity and mortality. The objective of this study was to determine the incidence and risk factors of nosocomial ESBL-producing GNB infections in a paediatric intensive care unit (PICU).

Patients and methods: A prospective surveillance study was performed from January 2015 through March 2019 in a PICU. All patients hospitalized for more than 48 h were included. Centers for Disease Control and Prevention criteria were applied for the diagnosis of nosocomial infection.

Results: During the study period, 1783 patients (median age: 12 ± 434 days) were included. The average length of stay was 9 ± 11 days with a total of 15,980 days of hospitalization. Newborns accounted for 61.8% of patients. Sixty-two per cent of patients were colonized with multi drug resistant Gram-negative rods, on admission or during their stay in the PICU. One hundred and nineteen bacterial infectious episodes were registered (7.4/1000 patient days). One hundred infectious episodes were caused by a GNB and 44 (36.9%) by ESBLs producing GNB with an incidence of 2.7/1000 patient days (bloodstream infections: 21 episodes, ventilator acquired pneumonia: 16 episodes). ESBLs producing GNB infection had a specific incidence of 10.9 per 1000 catheter-days, and 2.3 per 1000 mechanical ventilation-days. Fifty-nine percent of patients infected with ESBLs producing GNB had a prior digestive colonization with a multidrug-resistant GNB. Forty-one episodes (93%) occurred in patients with central venous catheters. Klebsiella Pneumoniae was the most frequently isolated bacteria (45.4%). Mortality in the ESBLs producing GNB group was high (29.5%). Associated factors of nosocomial ESBLs producing GNB infection were mechanical vrntilation (p < 0.001), central venous catheterization (p < 0.001) and colonization with multiple drug-resistant Gram-negative bacteria (p < 0.001).

Conclusion: Nosocomial ESBL-producing GNB infection had an incidence of 2.7 per 1000 patient days in our unit and seems to increase the mortality rate. Factors associated with this infection were identified.

Compliance with ethics regulations: Yes.

F-083 Serum phosphate variation precedes acute kidney injury in Tumor Lysis Syndrome

Marie Lemerle¹, Aline Schmidt², Valérie Thepot-Seegers³, Achille Kouatchet¹, Valérie Moal⁴, Mélina Raimbault⁵, Corentin Orvain², Jean-Francois Augusto⁶, Julien Demiselle¹

¹CHU Angers, Médecine intensive réanimation, Angers, France; ²CHU Angers, Maladie du sang, Angers, France; ³CHU Angers-ICO, Angers, France; ⁴CHU Angers, Pharmacie, Angers, France; ⁵CHU Angers, LABORATOIRE de Biochimie, Angers, France; ⁶CHU Angers, néphrologie dialyse transplantation, Angers, France

Correspondence: Marie Lemerle (marielemerle@yahoo.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-083

Rationale: Acute kidney injury (AKI) is associated with high morbidity and mortality in the setting of tumor lysis syndrome (TLS). Thus, strategies aimed at preventing AKI occurrence represent a major goal to improve prognosis of patients with TLS. The role of hyperphosphatemia as a risk factor of TLS has been poorly analyzed. The aim of this study was to study the association between hyperphosphatemia and AKI, and to determine whether a cut-off value of phosphatemia or phosphatemia’s variation was associated with AKI development during TLS.

Patients and methods: In this retrospective and monocentric study, we included all patients with TLS and whithout AKI at admission, admitted to hematology, nephrology and intensive care units of the University Hospital of Angers between 01/01/2007 and 31/12/2017.

Results: One hundred and thirty TLS episodes were identified in 120 patients. AKI developed during 56 episodes of TLS (43%). Hospital mortality was much higher in AKI patients (26.8% versus 10.8%, p = 0.018). Phosphate maximal values (2.2 ± 0.7 versus 1.9 ± 0.3) and LDH maximal values (4337.5 ± 4511.9 versus 2437.7 ± 2937.0) were higher in TLS with AKI, before AKI occurrence (p = 0.006 and p = 0.009, respectively). We found no association between the other biological parameters of TLS and AKI (serum calcium, uric acid and potassium). After adjustment for cofounders, there was a strong association between a rise in phosphate level of 0.1 mmol/L (HR 1.31 IC 95% [1.19–1.42], p < 0.0001), exposure to platinum salts (HR 3.66 IC 95% [1.74–7.71], p = 0.0006) and increasing maximal LDH value (HR per 1000UI/L increase 1.10 IC 95% [1.03–1.17], p = 0.030) with AKI.

Conclusion: This study highlights the utmost importance of serum phosphate in the setting of TLS: phosphate is an early relevant biomarker for the risk of AKI development. Further studies are needed to assess whether aggressive prophylactic treatment to control serum phosphate concentration, such as renal replacement therapy before AKI onset, constitutes a valuable approach.

Compliance with ethics regulations: Yes.

F-084 Chimeric Antigen Receptor T Cell-related toxicities requiring ICU admission: a single-centre experience

Clara Sortais¹, Amélie Seguin², Thomas Gastinne¹, Amandine Le Bourgeois¹, Steven Le Gouill¹, Jean-Baptiste Lascarrou³, Charlotte Garret³, Maelle Martin³, Arnaud-Felix Miailhe³, Helene Migueres³, Laura Crosby³, Olivier Zambon³, Jean Reignier³, Emmanuel Canet³

¹Service d’Hématologie Clinique/CHU de Nantes, Nantes, France; ²Service de Médecine Intensive Et Réanimation/CHU de Nantes, Nantes, France; ³Service de Médecine Intensive Et Réanimation/CHU de Nantes, Nantes, France

Correspondence: Clara Sortais (clara.sortais@chu-nantes.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-084

Rationale: Chimeric Antigen Receptor T cells (CARTs) are genetically modified T lymphocytes with promising results in refractory or relapsed (R/R) lymphoid hematologic malignancies. Severe forms of CARTs-related toxicities may require ICU admission and management. Our aim was to describe the clinical features and outcome of patients admitted to the ICU after CARTs infusion.

Patients and methods: Retrospective cohort of patients admitted to the medical ICU of university affiliated hospital after CARTs treatment between August 2018 and August 2019.

Results: Of the 20 patients treated by CARTs in the haematology department, 7 (35%) were subsequently admitted to ICU. Median age was 66 [50.5–68.5] years, and 5 (71.4%) were female. CARTs were indicated for R/R lymphoma. The median time between CARTs injection and ICU admission was 5 [3.5–5.75] days. All patients had Cytokine Release Syndrome (CRS), and 6 (85.7%) developed CAR-related encephalopathy syndrome (CRES). Median SOFA score and SAPS 2 were 4 [2–4.5] and 45 [40.5–49.5], respectively. Four (57.1%) patients had hypotension treated by fluid bolus (n = 3) or vasopressors (n = 2), and 2 (28.6%) had acute respiratory failure requiring oxygen therapy (n = 1) or mechanical ventilation (n = 1). Six (85.7%) patients had neurological symptoms (impaired consciousness n = 4, confusion n = 2, transient aphasia n = 1), of whom one developed refractory convulsive status epilepticus afterwards. All patients received broad spectrum antibiotics, of whom 2 (28.6%) had documented infections. Six (85.7%) patients received interleukin-6 inhibitor (single dose n = 2, multiple doses n = 4), and 5 (71.4%) received intravenous dexamethasone. One patient died in the ICU from septic shock. Median ICU and hospital length of stays were 4 [4–12] and 24 [21.5–28.75] days, respectively. Two (28.6%) patients died from relapsing malignancy before hospital discharge. Three months after ICU admission, four (57.1%) patients were alive in complete remission.

Conclusion: More than 30% of patients treated with CARTs required ICU admission for the management of a CRS or a CRES. Early ICU admission, close collaboration between haematologists and intensivists, and prompt administration of appropriate therapy (IL-6 inhibitor and/or dexamethasone) and supportive care resulted in a good prognosis.

Compliance with ethics regulations: Yes.

F-085 Pediatric Intensive Care Unit (PICU) management after Tisagenlecleucel (CTL019 Chimeric Antigen Receptor T Cell Therapy) infusion for B-Cell Acute Lymphoblastic Leukemia (ALL): experience of the first pediatric French patients

Jérôme Naudin¹, Arthur Felix², Maryline Chomton¹, Michael Levy³, Marie-Emilie Dourthe⁴, Delphine Chaillou⁴, Karima Yakouben⁴, André Baruchel⁵, Stéphane Dauger³

¹Service de réanimation pédiatrique-CHU Robert Debré-APHP, Paris, France; ²Institut Gustave Roussy, Villejuif, France; ³Service de réanimation pédiatrique-CHU Robert Debré-APHP-Université de Paris, Paris, France; ⁴Service d’hématologie pédiatrique-CHU Robert Debré-APHP, Paris, France; ⁵Service d’hématologie pédiatrique-CHU Robert Debré-APHP-Université de Paris, Paris, France

Correspondence: Jérôme Naudin (jerome.naudin@aphp.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-085

Rationale: Tisagenlecleucel (CTL019) is a chimeric antigen receptor T cell therapy that reprograms autologous T cells to target CD19 + leukemia cells, approved in the US since August 2017 and in the EU since August 2018 for children and young adult (< 25 years old) with relapsed/refractory B-cell acute lymphoblastic leukemia (B-ALL). This study reports the experience of PICU management of CTL019 toxicity in patients treated in Robert-Debré University Hospitals.

Patients and methods: All patients (age < 18 years old) treated by Tisagenlecleucel infusions between March 1, 2016 and September 15, 2019, included in sponsored-clinical trials or treated within the French compassionate program or with the commercial product, were retrospectively analyzed.

Results: Twenty-four patients were infused and 8 patients (33%) were managed in PICU for 11 stays. (2 stays: n = 1 and 3 stays: n = 1). Median age at PICU admission was 11.5 years old [10.9; 13.2] with a median delay after CAR-T cells infusions of 5 days [4.5; 6]. The median length of stay in PICU was 2 days [1.5; 3] with a max at 25 days. Cytokine release syndrome (CRS) was the main indication of PICU hospitalization (37.5%, n = 9) with grade 2 (n = 5) and grade 3 (n = 4) according to American Society for Transplantation and Cellular Therapy (ASTCT) consensus grading system and treated by corticosteroid (n = .9) and Tocilizumab (n = 5, only one infusion). Norepinephrine was the only vasopressor used. The median Vaso-Inotrope Score (VIS) for grade 3 was 20 [17.5; 22.5] with a maximum at 30. Neurologic toxicity was observed in 2 patients with a grade 4 (status epilepticus) and grade 3 (focal edema on neuroimaging with depressed level of consciousness) according to Immune effector Cell-associated Neurotoxicity Syndrome (ICANS) grading system from ASTCT consensus. The status epilepticus was managed with anti-epileptic drugs without mechanical ventilation. The focal edema was related to HHV6 and toxoplasmosis encephalitis. Evolution was positive with foscavir and ganciclovir and 14 days of mechanical ventilation. One patient was hospitalized for septic shock secondary to gram-negative central line bloodstream infection in aplasia, with a VIS score at 20. Evolution was favorable with antibiotics and central line removal. No death in PICU from severe Tisagenlecleucel toxicity was observed since the beginning of the CAR-T cells program.

Conclusion: Toxicity profile of Tisagenlecleucel required frequent and early PICU hospitalization after infusions for severe CRS and ICANS management.

Compliance with ethics regulations: Yes.

F-086 CAR-T cell therapy in ICU patients: a single-center experience

Sandrine Valade¹, Eric Mariotte¹, Virginie Lemiale¹, Lara Zafrani¹, Adrien Mirouse¹, Yannick Hourmant¹, Jean Jacques Tudesq¹, Asma Mabrouki¹, Igor Theodose¹, Catherine Thieblemont², Nicolas Boissel², Michaël Darmon¹, Elie Azoulay¹

¹APHP, Saint Louis Hospital, Medical ICU, Paris, France; ²APHP, Saint Louis Hospital, Hematology, Paris, France

Correspondence: Sandrine Valade (sandrine.valade@aphp.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-086

Rationale: CAR-T cell (chimeric antigen receptor T) therapy is a promising treatment in refractory acute lymphoid leukemia (ALL) and diffuse large B cell lymphoma (DLBCL). The main complication consists in a cytokine release syndrome (CRS) leading to an inflammatory state that can be very severe with life-threatening organ failure. Neurological toxicity is also reported. We aim to describe CAR-T cells-related complications in ICU patients.

Patients and methods: this is a single-center prospective study conducted between July 2017 and August 2019. All the patients who have received CAR-T cells and who required ICU admission were included. CRS grading was defined according to the most recent classification of the ASBMT and neurological toxicity was assessed with the CARTOX scale. Each admission is considered independent and therefore corresponds to one patient.

Results: 48 admissions, representing 41 patients (27 men and 14 women), were considered. The median age was 56 years [27–65]. Two-thirds of the patients have been diagnosed with DLBCL (n = 34, 71%) and one-third with ALL (n = 14, 29%), 31 months [23–59] ago. They had received 4 lines [3–4] of chemotherapy and had a high tumor burden (65% of lymphomas classified stage IV). The majority of the patients was admitted because of hemodynamic failure (n = 23, 48%) or respiratory failure (n = 9, 19%), 5 days [3–7] after CAR-T cells infusion. SOFA at admission was 4 [2–6]. All the patients presented at least one complication (Figure), the most common being CRS (n = 39, 81%) with a median grade of 2 [1–2]. Neurological toxicity was reported in 17 (35%) patients (worst grade at 3 [2–4]). Documented bacterial infection involved 29% of the patients and consisted in catheter-related infections for half of the cases. In the ICU patients were managed with fluid resuscitation (n = 27, 56%) during the first day, vasopressors (n = 14, 29%) and broad spectrum antibiotics (98%). A single patient required mechanical ventilation and two patients underwent dialysis. Tocilizumab (anti-IL6 receptor) was given to 34 patients (87% of CRS) in a median time of 5.5 h [2.1–21.5] after ICU admission. 30 patients (61%) received corticosteroids. The median ICU length of stay was 4.5 days [3–6]. 3 patients (6%) died in the ICU and hospital mortality was 14%.

Conclusion: CAR-T cells-induced complications occur in more than 50% of the patients. They are mostly represented by CRS. Infections are common and should always be considered and empirically treated.

Compliance with ethics regulations: Yes.

F-087 5-Fluorouracil-induced encephalopathy: a French national survey

Alice Boilève¹, Laure Thomas², Coralie Coccini³, Agnès Le-Lillo Louët⁴, Louise Gaboriau⁵, Laurent Chouchana⁶, Michel Ducreux¹, David Malka¹, Valérie Boige¹, Antoine Hollebecque¹, Dominique Hillaire-Buys³, Mathieu Jozwiak⁷

¹Département de Médecine Oncologique, Institut Gustave Roussy, Villejuif, France; ²Centre Régional de PharmacoVigilance, Assistance Publique des Hôpitaux de Paris, Hôpitaux Universitaires Paris-Est, Hôpital Henri Mondor, Créteil, France; ³Centre Régional de PharmacoVigilance, Hôpital Universitaire de Montpellier, Montpellier, France; ⁴Centre Régional de PharmacoVigilance, Hôpital Européen George Pompidou, Paris, France; ⁵Centre Régional de PharmacoVigilance, Assistance Publique des Hôpitaux de Paris, Hôpitaux Universitaires Paris-Centre, Hôpital Européen Georges Pompidou, Lille, France; ⁶Centre Régional de PharmacoVigilance, Assistance Publique des Hôpitaux de Paris, Hôpitaux Universitaires Paris-Centre, Hôpital Cochin, Paris, France; ⁷Service de Médecine Intensive Réanimation, Assistance Publique des Hôpitaux de Paris, Hôpitaux Universitaires Paris-Centre, Hôpital Cochin, Paris, France

Correspondence: Alice Boilève (alice.boileve@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-087

Rationale: The 5-Fluorouracil (5-FU)-induced hyperammonemic encephalopathy is a rare but serious 5-FU adverse drug reaction, which could require the admission of patients in intensive care unit (ICU). Given the paucity of data regarding this 5-FU adverse drug reaction, we performed a retrospective national survey from the French pharmacovigilance database to better characterize 5-FU-induced hyperammonemic encephalopathy and its management.

Patients and methods: Since the inception of the French pharmacovigilance database, we identified all patients that experienced 5-FU-induced encephalopathy. Variables regarding epidemiology, characteristics, management and prognosis of these patients were collected and analyzed.

Results: From 1985 to 2018, 30 patients (60[55–66] years-old, 43% of women) were included. Overall mortality was 17% (n = 5) and 57% (n = 17) of patients were admitted in ICU. The 5-FU-induced hyperammonemic encephalopathy started 2[1–4] days after the onset of 5-FU infusion. The most common neurological disorders were consciousness impairment, confusion and seizures. Abnormalities in CT scan, MRI, electroencephalogram and lumbar puncture were found in 9%, 65%, 77% and 16% of the whole population respectively, similar in ICU and non-ICU patients. Ammonemia was dosed in 50% of the whole population and in 65% of ICU patients. Hyperammonemia tended to be higher in ICU than in non-ICU patients (250[133–522] vs. 139[68–220] µmol/L, respectively, p = NS) and in patients with the lowest Glasgow outcome scale, but was not different between survivors and non-survivors. Among ICU patients, 70% required mechanical ventilation and 47% anti-epileptic drugs administration. Besides 5-FU discontinuation, lactulose intake, renal replacement therapy or ammonium chelators were used to decrease hyperammonemia in 17%, 27% and 7% of patients respectively. A complete neurological recovery was observed in up to 70% of ICU and non-ICU patients within a delay of 5[2–10] days. A dihydropyrimidine deshydrogenase (DPD) deficiency was found in 21% of tested patients. A 5-FU rechallenge was considered in 47% (n = 14) of patients with complete neurological recovery, including a patient with a partial DPD deficiency, within a delay of 19[18–28] days after recovery. A 5-FU-induced hyperammonemic encephalopathy relapse was observed in 57% of patients with 5-FU rechallenge. No relapse was observed when 5-FU rechallenge was performed with a decreased 5-FU dosage.

Conclusion: We report the first national survey and the largest cohort of patients with 5-FU-induced hyperammonemic encephalopathy so far. This serious 5-FU adverse drug reaction must be known by intensivists, since more than half of patients are admitted in ICU and specific treatments are available.

Compliance with ethics regulations: Yes.

F-088 Immune related adverse events: a retrospective look into the future of oncology in the intensive care unit

Adrien Joseph¹, Annabelle Stoclin², Antoine Vieillard-Baron³, Guillaume Geri³, Jean-Marie Michot⁴, Olivier Lambotte⁵, Elie Azoulay⁶, Virginie Lemiale⁶

¹U1138, INSERM, Equipe 11 labellisée Ligue Nationale Contre le Cancer, Centre de Recherche des Cordeliers, Paris, France; ²Département d’anesthésie-réanimation, Institut Gustave Roussy, Villejuif, France; ³Service de réanimation médico-chirurgicale, Hôpital Ambroise Paré, Assistance Publique Hôpitaux de Paris, Boulogne Billancourt, France; ⁴Département d’innovations thérapeutiques et d’essais précoces (DITEP), Institut Gustave Roussy, Villejuif, France; ⁵Service de médecine interne et immunologie clinique, Hôpital Bicêtre, Assistance Publique Hôpitaux de Paris, Le Kremlin-Bicêtre, France; ⁶Service de réanimation médical, Hôpital Saint-Louis, Assistance Publique Hôpitaux de Paris, Paris, France

Correspondence: Adrien Joseph (adrien.joseph@hotmail.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-088

Rationale: Immune checkpoint inhibitors (ICI) represent a paradigmatic shift in oncology. With their new position as a mainstay in cancer treatment, new toxicities called immune related adverse events (irAEs) have emerged.

Patients and methods: Retrospective study including patients admitted in the ICU within 60 days after treatment with an ICI in 3 French hospitals. Patients were classified into 3 groups according to the reason for admission: irAE, intercurrent adverse event (intAE) or event related to tumor progression (TumProg).

Results: 84 patients were admitted during the course of an ICI treatment, including 21 irAE, 25 intAE and 38 TumProg, with a significant increase between 2013 (n = 1) and 2018 (n = 24 patients, p for trend < 0.001). irAE included 5 pneumonitis, 4 colitis, 4 diabetes complications, 2 hypophysitis, nephritis, myocarditis and cardiac disorders, hepatitis or allergic reaction and 1 meningitis. The immune related nature of the complication was known before admission in only 4 (18%) cases. Mean age was 61 (± 14) years and 73% had a performance status of 0–1. Primary tumors were melanomas (14, 67%), non-small cell lung cancers (4, 19%), urothelial carcinomas (2, 10%) and Hodgkin lymphomas (1, 5%). ICI at the time of admission included anti-CTLA4 (5, 24%), anti-PD1/PDL1 (10, 48%) and anti-CTLA4/anti-PD1 combination in 6 (29%) patients. Mean duration of stay in the ICU was 6.5 (± 9) days. Three patients required vasopressor therapy alone, 2 with mechanical ventilation and one with extracorporeal membrane oxygenation. Three patients required non-invasive ventilation and 2 renal replacement therapy alone. Six required only endocrine or electrolytic equilibration and 4 others did not receive any form of organ support. ICU mortality was 14%. Compared with other admissions, anti-CTLA4 or anti-CTLA4/anti-PD1 combination treatments were associated with irAE diagnosis (OR = 6.1 [1.1–43.4], p = 0.021 for anti-CTLA4 and 5.7 [1.2–31.4] for anti-CTLA4/anti-PD1, p = 0.014) and so was the diagnosis of melanoma (5.7 [1.8–20.1], p = 0.001). There was no difference in terms of ICU and post-ICU survival between irAE (median post-ICU survival 14 months [7-NA]), intAE (20.9 [9.5-NA]) and TumProg (8.2 [7.3-NA]). Six patients admitted for an irAE were rechallenged with the same ICI after ICU discharge and 3 achieved complete response.

Conclusion: We conducted the first study describing patients admitted in the ICU for irAEs. Their specific and heterogeneous profile, along with the expected increase in the number of admissions, underlines the need for an in-depth knowledge for ICU physicians in order to take part in the multidisciplinary care required by these patients.

Compliance with ethics regulations: Yes.

Fig. 1

Swimmer plots of patients admitted in the intensive care unit for an immune related adverse event

F-089 Tyrosin kinase inhibitor: an effective tool against lung cancer involvement responsible for acute respiratory failure in ICU

Yanis Akrour¹, Yacine Tandjaoui-Lambiotte¹, Boris Duchemann², Frederic Gonzalez³, Annabelle Stoclin⁴, Fadwa El Kouari⁵, Aude Gibelin⁶, Paul Jaubert⁷, Anne-Sophie Moreau⁸, Florent Wallet⁹, Etienne De Montmollin¹⁰, Damien Contou¹¹, Nicolas Dufour¹², Romain Persichini¹³, Marc Pineton De Chambrun⁵, Alexandre Lautrette¹⁴, Anne Oppenheimer¹⁵, Yves Cohen¹⁶, Stéphane Gaudry¹⁶

¹Réanimation médico-chirurgicale, CHU Avicenne, APHP, Bobigny, France; ²Oncologie, CHU Avicenne, APHP, Bobigny, France; ³Réanimation, Institut Paoli Calmette, Marseille, France; ⁴Réanimation, Institut Gustave Roussy, Villejuif, France; ⁵Réanimation, CHU Pitié-Salpêtrière, APHP, Paris, France; ⁶Réanimation, CHU Tenon, APHP, Paris, France; ⁷Réanimation, CHU Cochin, APHP, Paris, France; ⁸Réanimation, CHRU, Lille, France; ⁹Réanimation, CHU Lyon, Lyon, France; ¹⁰Réanimation, CHU Bichat, APHP, Paris, France; ¹¹Réanimation, Argenteuil, France; ¹²Réanimation, CH Réné Dubos, Pontoise, France; ¹³Réanimation, CHU de la Réunion, Saint Denis De La Réunion, France; ¹⁴Réanimation, CHU, Clermont-Ferrand, France; ¹⁵EA 7285 Research Unit ‘Risk and Safety in Clinical Medicine for Women and Perinatal Health’, Versailles-Saint-Quentin University, Clamart, France; ¹⁶Réanimation, CHU Avicenne, APHP, Bobigny, France

Correspondence: Yanis Akrour (akrour.yanis92@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-089

Rationale: Patients with advanced-stage non-small-cell lung cancer have high mortality rates in the intensive care unit (ICU). In this context, acute respiratory failure due to cancer involvement is the worst situation. In the last two decades, targeted therapies have changed the prognostic of patients with lung cancer outside the ICU. Unlike cytotoxic chemotherapy, the fast efficacy of targeted therapies led some intensivists to use them as rescue therapy for ICU patients. We sought to investigate the outcomes of patients with lung cancer involvement responsible for acute respiratory failure and who received Tyrosine Kinase inhibitor during ICU stay.

Patients and methods: We performed a national multicentric retrospective study with the participation of the GRRROH (Groupe de Recherche en Réanimation Respiratoire en Onco-Hématologie). All patients with non-small-cell lung cancer admitted to the ICU for acute respiratory failure between 2009 and 2019 were included in the study if a Tyrosine Kinase Inhibitor was initiated during ICU stay. Cases were identified using hospital-pharmacies records. We collected demographic and clinical data in ICU charts. Vital status was assessed at the time of study completion (August 2019). The primary outcome was overall survival 90 days after ICU admission.

Results: Twenty-nine patients (age: 60 ± 14 years old) admitted to a total of 14 ICUs throughout France were included. Seventeen patients (59%) were nonsmoker. The most frequent histological type was adenocarcinoma (n = 20, 69%) and a majority had metastatic cancer (n = 21, 72%). Epithelial Growth Factor Receptor mutation was the most common oncologic driver identified (n = 15, 52%). During the ICU stay, 16 (55%) patients required invasive mechanical ventilation, 12 (41%) catecholamine infusion, 3 (10%) renal replacement therapy and one (3%) extracorporeal membrane oxygenation. In addition to Tyrosine Kinase Inhibitor, 20 (69%) patients received steroids (beyond 0.5 mg/Kg/day) and 3 (10%) cytotoxic chemotherapy during ICU stay. Seventeen patients (59%) were discharged alive from ICU and 9 (31%) were still alive after 90 days (see Kaplan–Meier curve Figure). Moreover, 5 patients (17%) were alive one year after ICU discharge.

Conclusion: Despite a small sample size this study showed that, in the context of lung cancer involvement responsible for acute respiratory failure, the use of Tyrosine Kinase Inhibitor should not be refrained in patients with severe condition in ICU.

Compliance with ethics regulations: Yes.

Fig. 1

probability of survival after ICU admission

F-090 Center effect in intubation risk in critically-ill immunocompromised patients with acute hypoxemic respiratory failure

Guillaume Dumas

¹Service de Médecine intensive-réanimation, Hôpital Saint-Louis, Paris APHP, Inserm UMR 1153, Paris, France

Correspondence: Guillaume Dumas (dumas.guillaume1@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-090

Rationale: Acute respiratory failure is the leading reason for intensive care unit (ICU) admission in immunocompromised patients and the need for invasive mechanical ventilation has become a major clinical end-point in randomized controlled trials (RCT). However, data are lacking on whether intubation is an objective criteria that is used unbiasedly across centers. This study explores how this outcome varies across ICUs.

Patients and methods: Hierarchical models and permutation procedures for testing multiple random effects were applied on both data from observational cohort (the TRIAL-OH study: 703 patients, 17 ICUs) and randomized controlled trial (the HIGH trial: 776 patients, 31 ICUs) to characterize ICU variation in intubation risk across centers.

Results: The crude intubation rate varied across ICUs from 29% to 80% in the observational cohort and from 0 to 86% in the RCT. This center effect on the mean ICU intubation rate was statistically significant, even after adjustment on individual patient characteristics (observational cohort: p-value = 0.013, Median OR 1.48 [1.30–1.72]; RCT: p-value:0.004, Median OR 1.51 [1.36–1.68]). Two ICU-level characteristics were associated with intubation risk (the annual rate of intubation procedure per center and the time from respiratory symptoms to ICU admission) and could partly explain this center effect. In the RCT that controlled for the use of high-flow oxygen therapy, we did not find significant variation in the effect of oxygenation strategy on intubation risk across centers, despite a significant variation in the need for invasive mechanical ventilation.

Conclusion: Invasive mechanical ventilation has become an important endpoint in immunocompromised patients with acute respiratory failure. However, we found significant variation in intubation risk across ICU in both an observational cohort and a randomized controlled trial. Our results highlight the need to take into account center effect in analysis because it could be an important confounder. Reasons for heterogeneity are various (case-mix differences, center practices). This gives opportunities to future improvement in care management and study design.

Compliance with ethics regulations: Yes.

F-091 The impact of Influenza on 28-day survival in ARDS non-trauma patients

Mathieu Lesouhaitier, Jean-Marc Tadié, Aurélien Frérou, Sonia Rafi, Claire Lhommet, Adel Maamar, Yves Le Tulzo, Arnaud Gacouin

CHU Rennes, Rennes, France

Correspondence: Mathieu Lesouhaitier (mathieu.lesouhaitier@chu-rennes.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-091

Rationale: Influenza virus (IV) infection is a major cause of ARDS that has been the focus of attention since the pandemic 2009 H1N1 (H1N1pdm2009) IV. Although IV-mediated damage of the airway has been extensively studied emphasizing specificity compared to other causes of ARDS, the impact of IV infection on the prognosis of ARDS patients, compared to the other causes of ARDS, has been few assessed.

Patients and methods: Systematic detection of IV in times of epidemic using RT-PCR in respiratory specimen is routine practice in our ICU along with prospective data collection of patients admitted to our ICU for ARDS with PaO₂/FiO₂ ratio ≤ 150 mmHg. All patients received lung-protective ventilation, the Sequential Organ Failure Assessment (SOFA) score was calculated on the first 3 days of mechanical ventilation. The primary endpoint compared the 28-day survival from the diagnosis of ARDS between patients with and without IV infection.

Results: From October, 2009 to May, 2019, 509 patients (pts) [median SAPS II score = 59 (33–67); age 58 years (47–68); PaO₂/FiO₂ ≤ 100 mmHg, n = 308 (61%)] were admitted to our ICU for ARDS with PaO₂/FiO₂ ratio ≤ 150 mm/Hg, including 100 pts (20%) with IV infection (H1N1pdm2009 IV A, n = 45; H3N2 A virus, n = 45; B virus, n = 10; associated bacteria, n = 46). Other main causes of ARDS were bacterial pneumonia without IV (39%), aspiration (15%), non-pulmonary sepsis (18%). 221 (42%) received prone positioning, and 47 (9%) extra-corporeal membrane oxygenation. The overall mortality rate at day-28 for the entire population was 33% (22 pts (22%) with IV infection versus 148 pts (36%) without IV infection, p = 0.007). Kaplan-Meier survival curves showed that survival was significantly higher in patients with IV infection than in those without IV infection. IV infection remained independently associated with a better prognosis at day-28 when entered as dichotomous variable (IV infection, yes/no) (adjusted hazard ratio (HR) = 0.60, 95% CI 0.37–0.98, p = 0.04) and when IV infection only was distinguished from other causes of ARDS including mixed infection IV plus bacteria (adjusted HR = 0.37, 95% CI 0.19–0.72, p = 0.003). Of note, within the first 3 days of mechanical ventilation, non-pulmonary SOFA scores were significantly lower in IV patients although similar pulmonary SOFA scores.

Conclusion: Our results suggest that patients with IV related ARDS have less severe non-pulmonary organ dysfunctions than those with ARDS from other and a lower mortality at day-28 despite similar ARDS severity.

Compliance with ethics regulations: Yes.

F-092 Acute Respiratory Distress Syndrome: Acute Respiratory Distress Syndrome among French Army. Retrospective analysis about war casualties on fifteen years study

Johan Schmitt¹, Mathieu Boutonnet², Goutorbe Philippe¹, Raynaud Laurent³, Cyril Carfantan⁴, Antoine Luft⁵, Pierre Pasquier², Eric Meaudre¹, Julien Bordes¹

¹Military Teaching Hospital Saint Anne, Toulon, France; ²Military Teaching Hospital Percy, Paris, France; ³Tropical Medecine Institute’s French Health Army Service, Marseille, France; ⁴Military Medical Service, Ventiseri, France; ⁵French Army Health Service Direction, Paris, France

Correspondence: Johan Schmitt (schmitt.johan83@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-092

Rationale: Acute Respiratory Distress Syndrome (ARDS) remains frequent in intensive care unit (ICU) with 20% to 40% mortality. According to Joint Theater Trauma System, ARDS occurs among 30% of war casualties: direct lung trauma, blast lesions, burn, massive transfusion and systemic inflammatory response syndrome lead to ARDS development. However, there is no data reporting ARDS among French evacuated casualties from forward environment. Our study’s aim is to describe ARDS incidence and its severity concerning medical evacuations from War Theater.

Patients and methods: This is an observational retrospective multicentric study analyzing all evacuated patient from War Theater and admitted in ICU. All patients developing ARDS according to Berlin definition have been included. Study has been approved by local ethic committee. Primary study endpoint was ARDS developing. Second study endpoints were ARDS severity, duration of invasive ventilation, ARDS treatments, ICU length of stay and mortality.

Results: 141 patients have been admitted in ICU between 2003 and 2018. 5 have been excluded. A total of 136 patients have been analyzed. 84% (n = 48) were military aged 30 (25–36) years. 42% (n = 57) developed ARDS. We found 57% (n = 32) war casualties, 30% (n = 17) trauma not related to war and 14% (n = 8) medical patients. Among severe trauma, median ISS was 34 (27–44), AIS thorax 3 (2–4), and 71% benefited from surgery on forward environment and 37% (n = 18) received massive transfusion. 22% (n = 13) suffered from mild ARDS, 42% (n = 24) moderate ARDS and 36% (n = 20) severe ARDS. Evacuation time was 26 (24–48) h. At admission in ICU, PaO₂/FiO₂ ratio was 241 (144–296) (Fig. 1). All patients were intubated. ARDS treatments used were curarization (76%, n = 43), prone position (16%, n = 9), inhaled nitric oxide (NOi) (10%, n = 6), almitrine (7%, n = 7) and extracorporeal life support (ECLS) (4%, n = 2). Invasive ventilation duration was 13 (7–27) days, length of stay 18 (9–33) days, and 3-month mortality 21% (n = 12).

Conclusion: According to our study, ARDS among French evacuated patients from war theaters remains frequent: it occurs on 42% among ICU admitted patients. 36% suffer from severe ARDS with 21% global mortality. Those datas are consistent with US studies. Also, we wonder if we must adapt our treatment capacities on forward environment for the most severe patients. In US army, a specialized team (Acute Lung Rescue Team) is trained to care the most hypoxemic war casualties with more treatment options as NOi, ECLS.

Compliance with ethics regulations: Yes.

Fig. 1

7-day evolution in PaO₂/FiO₂ ratio

F-093 Title: Acute respiratory failure due to Pneumocystis with or without acute respiratory distress syndrome requiring ICU admission

Laure Folliet, Jean-Christophe Richard, Damien Dupont, Meja Rabodonirina, Laurent Argaud, Vincent Piriou, Thomas Rimmele, Claude Guérin

HCL, Lyon, France

Correspondence: Laure Folliet (laure.folliet@chu-lyon.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-093

Rationale: Pneumocystis jirovecii pneumonia (PCP) is an opportunistic severe respiratory infection occurring in immunocompromised individuals. The main objective of present study was to evaluate the mortality of PCP patients with acute respiratory failure hospitalized in ICU, according to the presence of acute respiratory distress syndrome (ARDS) within the first 24 h of ICU stay.

Patients and methods: We performed a multicenter retrospective study in 4 adult ICUs in Lyon, France. Patients were included if they were admitted to ICU for acute respiratory failure with a respiratory sample tested positive for Pneumocystis jirovecii. ARDS was defined according to the Berlin definition. The primary endpoint was the mortality at ICU discharge. Patients were compared according to survival status at ICU discharge and ARDS within the first 24 h of ICU stay for demographic and anthropometric variables, underlying chronic medical condition, severity of the acute illness, PCP characteristics, ventilatory management and adjunct therapies during the ICU stay. Univariate comparisons were made by using non parametric tests. A multivariate logistic regression analysis was performed by using as covariates the variables which differed between survivors and not survivors in the univariate analysis, the ICU and the ARDS. No propensity score was used.

Results: From January 1st 2014 to September 30th 2018, 245 patients were included. The mortality at ICU discharge was 45.7% (112/245). In univariate analysis, non-survivor subjects were older, had higher SOFA score and SAPS2 score (69 vs. 66; p = 0.012), had more solid tumor (50.9% vs. 36.1%; p = 0.020), interstitial lung disease (21 vs. 10; p = 0.008) and less HIV (1.8% vs 10.5%;p = 0.006). Ninety-one patients (37.1%) met ARDS criteria. ARDS was more frequent in the non-survivor group in comparison to the survivor group (49.1% vs 27.1% p < 0.001). The results of the multivariate analysis indicated that the factors associated with mortality in PCP patients with respiratory failure were: chronic underlying interstitial lung disease, solid tumor. The factors associated with mortality reduction were oxygen therapy alone, high flow oxygen therapy alone, and adjunction of steroids for PCP.

Discussion: PCP is a severe respiratory infection that can complicate the evolution of many diseases accompanied by immunosuppression. We found that some risk factors are modifiable, and hence would allow an adapted and personalized care in patients with PCP admitted to ICU for acute respiratory failure.

Conclusion: This study identifies risk factors of mortality in patients with PCP and admission to ICU for acute respiratory failure.

Compliance with ethics regulations: Yes.

F-094 Impact of bronchoalveolar lavage in mechanically ventilated patients: an electrical impedance tomography—based study

Guillaume Franchineau, Paul Masi, Nicolas Brechot, Guillaume Hekimian, Loïc Le Guennec, Simon Bourcier, Charles-Edouard Luyt, Alain Combes, Matthieu Schmidt

Médecine Intensive-Réanimation, CHU Pitié-Salpêtrière, Paris, France

Correspondence: Guillaume Franchineau (gfranchineau@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-094

Rationale: Impact of bronchoalveolar lavage (BAL) on regional ventilation during mechanical ventilation patients has not been thoroughly studied. Electrical impedance tomography (EIT) allows monitoring tidal and end-expiratory lung volumes, and their respective regional distribution on mechanical ventilation. Objective: To describe, using EIT, the effects of BAL on tidal volume distribution and end-expiratory lung volume in mechanically ventilated patients.

Patients and methods: Monocentric prospective observational study including mechanically ventilated patients requiring a BAL between December 2018 and May 2019. Tidal volume distribution (i.e. tidal impedance (Δz)), variation of end-expiratory lung volume (i.e. End-Expiratory Lung Impedance (EELI)), and global static compliance were recorded before, immediately after, and every hour until 6 h after BAL in two distinct sub-groups, namely PaO₂/FiO₂ < 200 or ≥ 200. Results are expressed as median (1st–3rd quartiles) and number (percentage).

Results: Twenty-one patients (62% male, 62% with a PaO₂/FiO₂ < 200) were included. Median BAL duration was 6:37 (5:28–8:48) min and was mostly performed in the right lung (18/21 patients). Median volume injected was 120 mL, resulting in 53 (40–67) mL collected. Global compliance decreased after BAL from 33.5 (22.8–42.2) to 29.3 (20.4–36.2) mL/cmH₂O, before and 30 min after, respectively (P = 0.09). In patients with PaO₂/FiO₂ ≥ 200, Δz in the BAL-side during the following 6 h was significantly lower than before BAL. The lowest values were obtained at 30 min and 1 h (P = 0.01). Meanwhile, percentage of total EELI shifted from 55 (49–64) to 49 (46–56) % at 30 min in the BAL-side (P = 0.09) (Fig. 1). Same trends were observed in patients with PaO₂/FiO₂ < 200 though not reaching significance.

Conclusion: BAL in mechanically ventilated patients induces a significant decrease of 1) the global and regional static compliances, and 2) the end-expiratory lung volume in the BAL-side, which was still present 6 h after the procedure. These results reached significance only in patients with PaO₂/FiO₂ ≥ 200. The severe ventilation inhomogeneity before the BAL in patients with PaO₂/FiO₂ < 200 could have attenuated the BAL effect and may therefore explains these findings.

Compliance with ethics regulations: Yes.

Fig. 1

Distribution of A) median tidal impedance variation and B) median End-Expiratory Lung Impedance, in the side of broncho-alveolar lavage on patients without Acute Respiratory Distress Syndrome

F-095 Effect of transesophageal echocardiography on tracheal microaspiration and ventilator-associated pneumonia in intubated critically-ill patients: a multicenter prospective observational study

François Bagate¹, Anahita Rouze², Farid Zerimech³, Florence Boissier⁴, Vincent Labbe⁵, Keyvan Razazi¹, Guillaume Carteaux¹, Saad Nseir², Armand Mekontso Dessap¹

¹CHU Henri Mondor, Service de Médecine Intensive Réanimation, Créteil, France; ²CHU Lille, Service de Médecine Intensive Réanimation, Lille, France; ³CHU Lille, Centre de Biologie Pathologie, Lille, France; ⁴CHU Poitiers, Service de Médecine Intensive Réanimation, Poitiers, France; ⁵CHU Tenon, Service de réanimation polyvalente, Paris, France

Correspondence: François Bagate (francois.bagate@aphp.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-095

Rationale: Microaspiration of gastric and oropharyngeal secretions is the main mechanism of ventilator-associated pneumonia (VAP) pathogenesis. Transesophageal echocardiography (TEE) is performed routinely in intensive care unit and could potentiate microaspiration. The study aimed at evaluating the impact of TEE on microaspiration and VAP in adult intubated critically ill patients.

Patients and methods: Four-center prospective observational study. Quantitative measurement of microaspiration biomarkers (pepsin and salivary amylase) concentrations were performed in tracheal aspirates before and after TEE. The primary endpoint was the proportion of patients with TEE-associated microaspiration, as defined by: i) ≥ 50% increase in biomarker concentration between pre TEE and post TEE samples, and ii) a significant post TEE biomarker concentration (> 200 µg/L for pepsin and > 1685 IU/L for salivary amylase). Secondary endpoints included the occurrence of VAP within 3 days after TEE and the evolution of tracheal cuff pressure during TEE.

Results: Among 100 patients enrolled, VAP occurred in 19 patients (19%) within the 3 days following TEE. Patients with VAP had a larger tracheal tube size, had more attempts of TEE probe introduction, and were more often under anticoagulation treatment, as compared to their counterparts. We identified 17 patients with TEE-associated microaspiration (23%) among the 74 patients analyzed for biomarkers, but overall, pepsin and salivary amylase levels were not significantly different before and after TEE, with a wide interindividual variability. TEE induced an increase in tracheal cuff pressure, especially during insertion and removal of the probe.

Conclusion: Our results suggest no significant impact of TEE on microaspiration markers of gastric contents and oropharyngeal secretions and on VAP in intubated critically ill patients.

Compliance with ethics regulations: Yes.

F-096 Severe Chest Trauma as a Risk Factor of Post-Traumatic Pulmonary Embolism in the ICU

Mariem Dlela¹, Yassmine Kammoun², Olfa Turki¹, Kamilia Chtara¹, Mounir Bouaziz¹

¹Habib Bourguiba hospital, Sfax, Tunisia; ²Hedi Chaker hospital, Sfax, Tunisia

Correspondence: Mariem Dlela (mariem241090@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-096

Rationale: It is wellestablished nowadays that trauma patients bear a significantly increased risk of venous thrombo-embolic (VTE) events. Recent literature suggests that severe chest injury could constitute a risk factor for post-traumatic pulmonary embolism. The aim of this study was to determine predictive risk factors of pulmonary embolism (PE) in ICU trauma patients and evaluate whether severe chest trauma is independently associated to post-traumatic PE.

Patients and methods: We conducted a prospective cohort including trauma patients admitted to our ICU over a 20-month period between January 1^st, 2017 and August 31^st 2018. Patients were screened for PE at least once during ICU stay. PE was diagnosed based on computed tomography pulmonary angiogram (CTPA) results. Severe chest trauma was defined based on an Abbreviated Injury Scale (AIS) over three. Both univariate and multivariate analysis were used to determine level of significance.

Results: A total of 365 trauma patients were admitted to our ICU during the study period and 66 patients (18.08%) were diagnosed with PE. Patients had a mean age of 40 ± 14 years and a mean Injury Severity Score (ISS) of 34 ± 10. Thirty-eight patients (10.4%) were diagnosed with severe blunt chest injury (AIS ≥ 3). Univariate analysis identified the following variables as predictive of post-traumatic PE, including older age, obesity (BMI > 30), severe head injury (AIS head ≥ 3), severe chest trauma (AIS chest ≥ 3), higher SOFA scoring, lower PaO₂/FiO₂ ratio on admission and post-traumatic disseminated intravascular coagulation. On multivariate analysis, we found that severe chest trauma was an independent factor predictive of post-traumatic PE (p = 0.008; OR = 8.7; CI = 7.76–43.4), in addition to severe head injury (p = 0.035; OR = 2.3; CI = 1.16–5.15) and obesity (p < 0.001; OR = 4.8; CI = 2.03–12.3).

Conclusion: This cohort study confirms that severe chest trauma should be recognized as a newly identified risk factor of post-traumatic PE. This knowledge should lead clinicians to maintain a high index of suspicion in injured patients.

Compliance with ethics regulations: Yes.

F-097 Impact of pulmonary and non-pulmonary sepsis on susceptibility to secondary bacterial pneumonia: a pivotal role for the alveolar macrophage?

Jean-FrançOis Llitjos¹, Zakaria Ait Hamou¹, Christophe Rousseau¹, Matthieu Benard¹, Hugues Vicaire¹, Clémence Martin², Pierre-Régis Burgel², Jean-Paul Mira³, Jean-Daniel Chiche³, Maha Zohra Ladjemi¹, Frédéric Pene³

¹INSERM U1016, Institut Cochin, 22 rue Méchain, 75014, Paris, France; ²Service de pneumologie, hôpital Cochin, Paris and INSERM U1016, Institut Cochin, 22 rue Méchain, 75014, Paris, France; ³Médecine Intensive-Réanimation, hôpital Cochin, Paris and INSERM U1016, Institut Cochin, 22 rue Méchain, 75014, Paris, France

Correspondence: Jean-FrançOis Llitjos (jllitjos@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-097

Rationale: We recently reported that septic shock patients with pneumonia exhibit a high risk of ICU-acquired pneumonia, suggesting that a primary pulmonary insult may drive profound alterations in lung defence towards secondary infections (1). Given their importance in lung immune surveillance, alveolar macrophages (AM) are likely to play a pivotal role in this setting. The objective of this experimental study is to address the impact of primary pulmonary or non-pulmonary infectious insults on lung immunity.

Patients and methods: We established relevant double-hit experimental models that mimic common clinical situations. C57BL/6 J mice were first subjected either to polymicrobial peritonitis induced by caecal ligation and puncture (CLP), or to bacterial pneumonia induced by intra-tracheal instillation of Staphylococcus aureus or Escherichia coli. Respective control mice were subjected to sham laparotomy or intra-tracheal instillation of phosphate-buffered saline. Seven days later, mice that survived the primary insult were subjected to intra-tracheal instillation of Pseudomonas aeruginosa (PAO1 strain). We assessed survival and pulmonary bacterial clearance of post-septic animals subjected to P. aeruginosa pneumonia, as well as the distribution and functional changes in alveolar macrophages.

Results: When compared to sham-operated mice, post-CLP animals exhibited increased susceptibility to secondary P. aeruginosa pneumonia as demonstrated by defective lung bacterial clearance and increased mortality rate (50% vs. 0%, p < 0.05). In contrast, all post-pneumonia mice survived and even exhibited improved bacterial clearance as compared to their control counterparts. When addressing whole-lung immune cell distribution at the time of second hit (day 7), amounts of AM were decreased in post-CLP mice while preserved or even increased in post-pneumonia mice. Antigen-presenting functions of AM appeared similar in all conditions. Percentages of apoptotic (AnnexinV⁺) and necrotic (7-AAD⁺) AM were comparable at day 1 and day 7 after the first hit. Interestingly, both Ly6C^high and Ly6C^low monocytes were sustainably increased in the lungs of post-CLP mice, while only transiently expanded following pneumonia, suggesting that differences in AM counts could be related to modulated turnover from precursor monocytes.

Conclusion: Using clinically relevant double-hit experimental models, a primary pulmonary infection conferred resistance to secondary bacterial pneumonia. Ongoing investigations are aimed at addressing the antibacterial AM functions, as well as the turnover-driving mechanisms.Compliance with ethics regulations: Yes.

Fig. 1

.

Reference

• 1. Llitjos JF et al., Ann Intensive Care 2019.

F-098 Local signs at insertion site and prediction of catheter-related infections in short-term central venous and arterial catheters in the ICU. Individual findings from 4 multicenter randomized controlled trials

Niccolo Buetti¹, Stephane Ruckly¹, Jean-Christophe Lucet¹, Olivier Mimoz², Bertrand Souweine³, Jean-François Timsit¹

¹INSERM IAME DESCID, Paris, France; ²Services des Urgences Adultes and SAMU 86, Centre Hospitalier Universitaire de Poitiers, Poitiers, France; ³Medical ICU, Gabriel-Montpied University Hospital, Clermont-Ferrand, France

Correspondence: Niccolo Buetti (niccolo.buetti@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-098

Rationale: Little is known on the role of exit-site signs in predicting intravascular catheter infections. The current study aimed to describe the association between local signs at the exit-site and catheter-related bloodstream infection (CRBSI), which factors substantially influenced local signs and which clinical conditions may predict CRBSIs if inflammation at insertion site is present.

Patients and methods: We used individual data from 4 multicenter randomized-controlled trials in intensive care units (ICUs) that evaluated various prevention strategies regarding colonization and CRBSI in central venous and arterial catheters. We used univariate and multivariate logistic regression stratifying by center in order to identify variables associated with redness, pain, non-purulent discharge, purulent discharge and ≥ 1 local sign and subsequently evaluate the association between CRBSI and local signs. Moreover, we evaluated the role of the different local signs for developing CRBSI in subgroups of clinically relevant conditions.

Results: A total of 6976 patients, 14,590 catheters (101,182 catheter-days) and 114 CRBSI (0.8%) from 25 ICUs with described local signs were included. Redness, pain, non-purulent discharge, purulent discharge and ≥ 1 local signs at removal were observed in 1633 (11.2%), 59 (0.4%), 251 (1.7%), 102 (0.7%) and 1938 (13.3%) episodes, respectively. The sensitivity of ≥ 1 local sign for CRBSI was by 40.4%, whereas the highest specificities were observed for pain (99.6%) and purulent discharge (98.4%). Positive predictive value (PPV) was low for redness (2%), pain (3%), non-purulent discharge (3%) and ≥ 1 local sign (2%), but increased for purulent discharge (12.7%). Negative predictive values were high for all local signs. After adjusting on confounders, CRBSI was associated with redness, non-purulent discharge, purulent discharge and ≥ 1 local sign (Fig. 1). Conditions independently associated with ≥ 1 local sign were age ≤ 75 years old (OR 1.23, 95% CI 1.07–1.40, p < 0.01), SOFA score (SOFA < 8 OR 1.45, 95% CI 1.23–1.71, p < 0.01), non-immunosuppression (OR 1.38, 95% CI 1.12–1.68, p < 0.01), catheter maintenance > 7 days (OR 3.36, 95% CI 3.03–3.73, p < 0.01) and insertion site (OR for subclavian site 1.63, 95% CI 1.37–1.93, p < 0.01). However, the presence of ≥ 1 local sign was more predictive for CRBSI in the first 7 days of catheter maintenance (OR 6.30, 95% CI 3.53–11.24 vs. > 7 catheter-days OR 2.61, 95% CI 1.58–4.32, p heterogeneity = 0.02).

Conclusion: This post hoc analysis showed that local signs were related to CRBSIs in the ICU. Local signs were independently associated with specific patient’s and catheter’s conditions. In the first 7 days of catheter maintenance, local signs were predictive for CRBSI.

Compliance with ethics regulations: Yes.

Fig. 1

Unadjusted and adjusted CRBSI-risk for the different local signs

F-099 Epidemiology and Outcomes of Pneumococcal meningitis with Sepsis in France

Claire Dupuis¹, Michael Thy¹, Bruno Mourvillier¹, Lila Bouadma¹, Stephane Ruckly², Anne Perozziello³, Damien Van-Gysel⁴, Etienne de Montmollin¹, Romain Sonneville¹, Jean-François Timsit¹

¹Réanimation médicale et infectieuse, CHU Bichat, AP-HP, Paris, France; ²UMR 1137 IAME Team Descide, Paris, France; ³UMR 1137-IAME Team 5–DeSCID, Paris, France; ⁴Département d’Informations Médicales, CHU Bichat, APHP Paris, Paris, France

Correspondence: Claire Dupuis (cdup83@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-099

Rationale: Pneumococcal meningitis (PM) is the leading cause of bacterial meningitis in adult patients requiring ICU admission and is associated with a high case fatality rate (CFR), ranging from 15 to more than 33% (1–3). Patients with PM may develop sepsis or septic shock that may impact management and outcomes. We aim to describe the epidemiology and outcomes of PM associated with sepsis in adult patients in France.

Patients and methods: We analysed the occurrence of PM with sepsis from 2010 to 2015 in adult patients, using the national French hospital Database PMSI (Programme de Médicalisation des Systèmes d’Information). For all analyses, only the first hospital admission was considered. Cases were identified using a combination of a diagnosis code for PM plus a diagnosis code for sepsis (either a code for organ failure or a procedure code for organ support). Data recorded included comorbidities (4), characteristics of the hospital stay, severity of the patients including major intracranial complications and characteristics of the infection. Costs and endpoints were determined at the end of all the hospital stays related to the first admission for PM with sepsis. Standardized incidence, hospital mortality, and CFR were estimated. Temporal trends were assessed using Cochran Armitage tests of trends and linear trend analyses.

Results: A total of 1236 PM with sepsis aged ≥ 18 years were hospitalized in France during 2010–2015. The incidence of PM decreased from 3.7 to 2.6 per 1 M inhabitants (p < 0.02) (Fig. 1). Most of them came from home (94%), were admitted in an academic institution (93%) and benefited from ICU (93%). Their median age was 62 [51; 73] years. Two-third of them had at least one comorbidity. The initial neurological presentations included coma (58%), focal signs (15%), seizures (12%) and brain stem involvements (16%). The SAPS II score was 57 [44; 69] points. The main neurological complications were cerebrovascular complications (6%), cerebral abscess (4%) and hydrocephaly (2%). PM was associated with pneumococcal septicaemia or pneumococcal pneumonia in 30% and 23% of cases respectively. The length of ICU and hospital stays were 10 [4; 21] and 22 [10; 41] days respectively and only ICU length of stay decreased over time (p < 0.01). The prognosis was poor since only 27.6% of the patients were discharged to home. Indeed, 42.6% of them died and 18% were transferred to rehabilitation units. No temporal trends could be observed for these outcomes. The average hospital costs per case were 21,717€ [13.198; 34.232].

Conclusion: PM with sepsis in adult in France remained a real burden associated with a high mortality rate, and disability.

References

• 1. Auburtin, Crit Care Med. 2006.

• 2. Muralidharan, Eur J Neurol. 2014.

• 3. Tsai, Clin Infect Dis. 2008.

• 4. Quan H, Med Care. 2005.

Compliance with ethics regulations: NA.

Fig. 1

Incidence, mortality and case fatality of pneumoccocal meningitis in France from 2010 to 2015

F-100 Poor outcome associated with Mucormycosis in critically-ill hematological patients: Results of a multicenter study

Matthieu Jestin¹, Adrien Mirouse¹, Jean-Jacques Tudesq¹, Yannick Hourmant¹, Fabrice Bruneel², Julien Mayaux³, Martin Murgier⁴, Frédéric Pene⁵, Eric Mariotte⁶, Virginie Lemiale⁶, Lara Zafrani⁶, Elie Azoulay⁶, Michaël Darmon⁶, Sandrine Valade⁶

¹Service de Médecine Intensive et Réanimation, Hôpital Saint-Louis, Paris, France; ²Service de réanimation médico-chirurgicale, Centre hospitalier de Versailles, Le Chesnay, France; ³Service de Pneumologie, Médecine Intensive et Réanimation, Hôpital Universitaire Pitié-Salpêtrière, Paris, France; ⁴Service de Réanimation polyvalente, Centre hospitalo-universitaire de Saint-Etienne, Saint-Etienne, France; ⁵Service de Médecine Intensive et Réanimation, Hôpital Cochin, AP-HP, Paris, France; ⁶Service de Médecine Intensive et Réanimation, Hôpital Saint-Louis, Paris, France

Correspondence: Matthieu Jestin ⁽m_jestin56@hotmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-100

Rationale: Mucormycosis is an emerging fungal infection, especially in patients with hematological malignancies. Although this infection may lead to multi organ failure, no study has been dedicated to critically ill patients with hematological malignancy. The primary objective was to assess outcome in this setting. The secondary objective was to assess prognostic factors.

Patients and methods: This retrospective cohort study was performed in 5 ICUs. Critically ill adult patients with hematological malignancies and mucormycosis were included between 2002 and 2018. Mucormycosis was classified as “probable” or “proven” regarding EORTC criteria. Variables are reported as median [IQR] or number (%). Adjusted analysis was performed using Cox Model.

Results: Twenty-six patients were included with a median age of 38 years [IQR, 26–57]. Acute leukemia was the most frequent underlying disease (n = 13, 50%). Nine patients (35%) were allogeneic stem cell transplantation (SCT) recipients. Nineteen patients (73%) had neutropenia and 16 patients (62%) had received steroids. The main reason for admission was acute respiratory failure (n = 14, 54%) followed by shock (n = 5, 19%). The median SOFA score at admission was 7 [IQR, 5–8] points. Only 3 patients (11%) had received prior anti-fungal prophylaxis effective against mucorales. Mucormycosis was “proven” in 14 patients and “probable” in 12 patients. Diagnosis was made by histopathologic examination in 14 patients, direct microscopy or culture in 5, and polymerase chain reaction in 7. Rhizopus and Mucor were the most frequent documented species. Seven patients (27%) had concurrent Aspergillus infection. Mucormycosis was diagnosed 1 day [− 4 to + 6] after ICU admission. Ten patients (38%) had pulmonary involvement whereas five patients (19%) had rhino-cerebral involvement. Infection was disseminated in eight patients (31%). Twenty-two patients (85%) were treated with liposomal amphotericin B. Twelve patients (46%) received antifungal combination including posaconazole in 7. Eight patients (31%) underwent curative surgery. Multiple organ failure was frequent, 21 patients (81%) requiring invasive mechanical ventilation (IMV), 19 (69%) vasopressors, and 9 (35%) renal replacement therapy. ICU and hospital mortality rates were 77% and 88%, respectively. Only two patients were alive at day 90. Three variables were associated with mortality in a Cox model including allogeneic SCT (HR 4.84 [95% IC 1.64–14.32]; figure), SOFA score (HR 1.19 [95% IC 1.02–1.39]) and dual therapy (HR 3.02 [95% IC 1.18–7.72]) (Fig. 1).

Conclusion: Mucormycosis is associated with a high mortality rate in patients with hematological malignancies, especially in allogeneic SCT recipients. Futility of ICU management in these patients is to be considered and strategies aiming to improve these patients’ outcome are urgently needed.

Compliance with ethics regulations: Yes.

Fig. 1

Kaplan-Meier curve reporting unadjusted influence of allogenic stem cell transplantation (alloSCT)

F-101 Prediction of clinical phenotypes with biomarkers during presumed sepsis: a non-supervised analysis of the Captain cohort

Benoit Misset¹, François Philippart², Virginie Moucadel³, Mariana Parlato⁴, Catherine Fitting⁵, Jean-Pierre Bedos⁶, Jean-Luc Diehl⁷, Olfa Hamzaoui⁸, Djillali Annane⁹, Didier Journois⁷, Jean-Marc Cavaillon⁵, Joël Coste⁷

¹CHU Sart-Timan, Liège, Belgium; ²GHPSJ, Paris, France; ³Biomérieux, Lyon, France; ⁴Inserm, Paris, France; ⁵Pasteur Institute, Paris, France; ⁶CH, Versailles, France; ⁷APHP, Paris, France; ⁸APHP, Clamart, France; ⁹APHP, Garches, France

Correspondence: Benoit Misset (benoit.misset@chuliege.be)

Ann. Intensive Care 2020, 10 (Suppl 1):F-101

Rationale: Sepsis is a life-threatening organ dysfunction caused by a dysregulated host response to infection. Several mediators, alone or in combination, were proposed to characterize individual response, but none was proven to have good external validity. The aim of this work was to establish whether some combinations are linked to clinical phenotypes in patients with presumed sepsis, using the data collected in the Captain multicenter cohort which methods and first results were previously published (Parlato, ICM 2018).

Patients and methods: 279 patients were prospectively included at the time of sepsis criteria, 188 (67%) of whom with a secondary confirmed infection. Community acquired pneumonia was causal in 70% of infections. SAPS2 score = 55 points [50–61], age = 65 years [57–78], male sex = 64%. 203 patients were followed for more than 3 days, in whom usual ICU clinical and biological parameters were collected, as well as 29 plasma biomarkers and 10 leucocyte associated RNAs. Patients were clinically classified according to their acute severity (SOFA score, serum lactate), confirmed initial infection, outcome (secondary infection occurrence, ICU survival). Non-supervised principal component analysis of the maximal values of biomarkers assessed on first 2 days of sepsis, and Varimax rotation technique of the selected components using SAS software.

Results: 203 patients, med SOFA day1 = 5 pts, med serum lactates day 1 = 1.5 mEq/L, bacterial infection = 141 (69%), Enterobacteriaceae infection = 45 (22%), VAP and/or bacteremia after day 5 = 30 (15%), alive at ICU d/c = 146 (72%). Five components explain 57% of the variance of the biomarkers. The first component (26% of the variance) was not linked to the clinical predetermined phenotypes. The second component (11% of the variance) was principally made of HLA-DR RNA, CD74 RNA and CX3CR1 RNA, and linked to a lower initial severity (r = − 0.32, p = 0.0001), a less frequent confirmation of initial infection (p = 0.0001), a lower occurrence of pneumonia or bacteremia (p = 0.02) or death (p = 0.008).

Conclusion: In our cohort, using non supervised analysis, we could separate a biomarker association linked to lower initial severity, lower rate of a bacterial cause to sepsis, and better outcome. The markers found are among those which are regularly considered as describers of the peripheral alteration of the immune system observed during sepsis (Pachot, CCM 2005; Friggeri, CC 2016; Peronnet ICM 2017).

Compliance with ethics regulations: Yes.

F-102 Pro-adrenomedullin and Procalcitonin serial levels in predicting infection relapse or superinfection and outcome in ICU—a post hoc analysis of the PRORATA biobank

Lila Bouadma¹, Tiphaine Robert¹, Charles-Edouard Luyt², Philippe Montravers³, Carole Schwebel⁴, Stephane Ruckly¹, Jean-François Timsit¹

¹Hôpital Bichat Claude Bernard, Paris, France; ²Hôpital de la Pitié Salpêtrière, Paris, France; ³Hôpital Michallon, Paris, France; ⁴Hôpital Michallon, Grenoble, France

Correspondence: Lila Bouadma (lila.bouadma@aphp.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-102

Rationale: Prorata trial (1) compared a standard of care to a procalcitonin (PCT) oriented use of antimicrobials for sepsis in 8 iCUs. Serial blood samples were biobanked in 4/87 ICUs (455/621 patients enrolled for pro-adrenomedullin (ProADM) and PCT concentrations).

Patients and methods: The aim of the study was to evaluate the respective impact of serial PCT and proADM measurements in predicting relapse or superinfection and death on day 28*. Relapse was defined as the growth of one or more of the initial causative bacterial strains (i.e., same genus, species) from a second sample taken from the same infection site at 48 h or more after stopping of antibiotics, combined with clinical signs or symptoms of infection. Superinfection was defined as the isolation from the same or another site of one or more pathogens different from that identified during the first infectious episode, together with clinical signs or symptoms of infection. Methods: Outcomes of interest were recurrence or reinfection, D28 mortality, or both. Data are presented as median [interquartile ranges] or n (%) as appropriate. ProADM and PCT were entered in a multivariate Cox model as time dependent variables. Best thresholds were estimated by maximizing the HR of death for both biomarkers and from available literature (2.3).

Results: Of the 455 patients included (age: 63 [54–74] years, SAPS II: 47 [35; 60] points and SOFA: 8 [4–11] points at amission, medical admission: 427 (95%), immunocompromised: 81 (18%), on mechanical ventilation 306 (67%), PCT and ProADM at inclusion were 2 [0.6–9.3] ng/mL and 1.5 [0.4–2.7] nM/L respectively. 173 (38%) patients developed a first episode of recurrence or supereinfection after a median delay of 11 days [8–16] and 98 (22%) died before d28. The HR maximization process proposed an optimal cut point of 1 ng/mL for PCT and 2 nM/L for Pro ADM to predict d28 death. In the multivariate Cox model, both PCT and ProADM were associated with death but not with relapse or superinfection (Table 1).

Conclusion: Conclusion: Both serial measurements of PCT and ProADM are independent predictors of death in patients treated for sepsis in ICU. Our study confirmed the use of 2 nM/L as a good prognosis cut point for ProADM. 1.

Compliance with ethics regulations: Yes.

Table 1 Multivariate Cox model

References

• 1. Bouadma L et al.—Lancet 2010; 375: 463–74 2.

• 2. Maisel A et al. J Am Coll Cardiol 2011;58:1057–67. 3.

• 3. Ara-Somohano C et al.—Minerva Anesthesiol 2017 Aug;83 (8):824–835.

F-103 Performance of repeated (1-3)-β-d-glucan measures, a post hoc analysis of the EMPIRICUS Randomized Clinical Trial

Claire Dupuis¹, Clément Le Bihan¹, Muriel Cornet², Stephane Ruckly³, Carole Schwebel², Lila Bouadma¹, Elie Azoulay⁴, Jean-François Timsit¹

¹Réanimation médicale et infectieuse, CHU Bichat, APHP, Paris, France; ²Mycologie-Parasitologie, CHU Grenoble, Grenoble, France; ³UMR 1137, IAME, équipe Descide, Université de Paris, Paris, France; ⁴Réanimation médicale, CHU Saint Louis, APHP, Paris, France

Correspondence: Claire Dupuis (cdup83@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-103

Rationale: The performance of serum (1-3)-β-d-glucan (BDG) and its evolution to predict the occurrence of invasive fungal infection (IFI) in a high risk non immunocompromized population remains to be determined (1).

In a post hoc analysis of the EMPIRICUS Randomized Clinical Trial (2), we aimed to assess the prognostic value of repeated measures of BDG on the occurrence of invasive fungal infections.

Patients and methods: Non-neutropenic, non-transplanted, critically ill patients with ICU-acquired sepsis, multiple Candida colonization, multiple organ failure, exposed to broad-spectrum antibacterial agents, and enrolled between July 2012 and February 2015 in 19 French ICUs were included. BDG were collected in ICU at day 0, 3, 7, 14 and 28 after inclusion. A value Time 0 of more than 80 pg/mL, 250 pg/mL and an increase by more than 24% from the previous measurement (threshold of measurement error) were assessed at baseline and overtime. For that purpose, we conducted cause specific hazard models with death as a competing risk. We also planned subgroup analyses on the placebo and the micafungin groups. Cumulative risk (CumRisk) of IFI at day 28 were derived from models.

Results: 234 patients were included: age: 64.2 years [53.4; 73.7], SAPS II: 48 points [38; 57]. 52 (22.2%) were surgical patients. The delay between ICU admission and trial inclusion and the corresponding first measurement of BDG was 10 days [7; 16] for a result of 95.6 pg/mL [41.9; 200.7]. 114 patients received micafungin. 28 patients presented an IFI and 68 (29.1%) patients died at day 28. Finally, only BDG at baseline over 80 pg/mL was associated with an increased risk of IFI. We found those results in the whole cohort (CSHR (IFI) = 4.98 [1.73; 14.35]; CSHR (Death) = 1.28 [0.75; 2.17], with a CumRisk of IFI (CumRisk IFI) of 19.8% [19.5; 20.0] (BDG > 80 pg/ml) versus 5.24% [5.1; 5.37] (BDG < 80 pg/ml)) (Fig. 1) and in the placebo subgroup (CSHR (IFI) = 6.41 [1.47; 28.06]; CSHR (Death) = 1.09 [0.52; 2.27], CumRisk IFI (BDG > 80) = 25.9% [25.1; 26.6] versus CumRisk IFI (BDG < 80) = 5.4% [5.1; 5.7]), but not in the micafungin sub group (CSHR (IFI) = 3.55 [0.77; 16.46], CSHR (Death) = 1.53 [0.71; 3.31], CumRisk IFI (BDG > 80 pg/ml) = 13.4% [13.2%; 13.8%] versus CumRisk IFI (BDG < 80 pg/ml) = 5% [4.7; 5.2]). Neither a BDG > 250 pg/mL, nor an increase by 24% of BDG over time were associated with the occurrence of IFI. Similar results were found in the placebo subgroup.

Conclusion: Among high risk patients, a first measurement of BDG over 80 pg/mL was highly associated with the occurrence of IFI. Neither a cut-off of 250 pg/mL, nor repeated measurements of BDG over time seemed to be useful to predict the occurrence of IFI. The cumulative risk of IFI in the placebo group if BDG > 80 pg/ml is 25.39% questioning about the potential interest of empirical therapy in this subgroup.

Compliance with ethics regulations: Yes.

Fig. 1

Cumulative incidence of invasive fungal infection and death depending on the serum (1-3)-β-d-glucan (BDG) level at baseline

References

• 1. Clancy, J Clin Microbiol. 2018.

• 2. Timsit, JAMA. 2016.

F-104 Prehospital measurement of lactatemia and blood culture drawing for septic shock

Barbara Alves, Benoit Vivien, Romain Jouffroy

Anesthesia and Intensive Care Unit—SAMU 75—Assistance Publique des Hôpitaux de Paris –Necker Enfants Malades Hospital, Paris, France

Correspondence: Barbara Alves (brb.alv@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-104

Rationale: Since the 2016 Sepsis-3 conference, the distinction between sepsis and septic shock is based on blood lactate value. Septic shock may be encountered in the pre-hospital setting. In order to reduce the mortality, the precocity of treatments implementation has been emphasized, particularly early antibiotic administration. Prior antibiotic administration, and blood culture drawing must be performed. The aim of this survey was to clarify the capabilities of French prehospital emergency service (PEMS) to measure blood lactate and to draw blood culture prior to hospital admission for septic shock.

Patients and methods: We performed an electronic survey of 26 auto-questions addressed to the deputy chair of the French PEMS in 2018.

Results: Sixty PEMS (60%) participated in the survey. Twenty-five percent are able to measure blood lactate and 45% are able to draw blood culture in the prehospital setting. Ninety-five percent declared lactate measurement is helpful in assessing severity. Ninety percent claimed that the lactate value influences the hospital facility, emergency department vs. intensive care unit. Twenty-eight percent believe that the impossibility to draw blood culture precludes prehospital antibiotic administration. Sixty-three percent estimate that a protocol for septic shock management would be beneficial.

Conclusion: Few French PEMS are able to measure lactate and draw blood culture in the prehospital setting. The impact of blood lactate measurement and blood culture drawing by PEMS on septic shock outcome requires further studies.

Compliance with ethics regulations: Yes.

F-105 Head Injury Imaging-Timing is Critical

Mufaddal Jivanjee

¹London North West Healthcare NHS Trust, London, UK

Correspondence: Mufaddal Jivanjee (mufaddal.jivanjee@nhs.net)

Ann. Intensive Care 2020, 10 (Suppl 1):F-105

Rationale: Head injury is a common cause of morbidity and mortality in the first four decades of life, accounting for approximately 200,000 annual hospital admissions in the United Kingdom. The majority of patients recover without intervention, however some may develop a long-term disability or even die. The early detection of pathology is therefore absolutely critical in determining patients’ prognosis, helping to provide appropriate timely management. The National Institute for Health and Care Excellence (NICE) adult head injury guidelines, recommend that head injuries with specific risk factors should have a CT scan within 1 h of risk factors being identified. Furthermore the provisional report should be made available within 1 h of the scan. This audit assessed the compliance of staff to the NICE adult head injury guidelines.

Patients and methods: The previous 40 adult CT head scans, requested due to head injury, from the Emergency Department (ED) at London North West Healthcare NHS Trust were analysed for compliance to the NICE guidelines. The standards measured were: (1) Time from request of scan to completion of scan should be within 1 h; (2) Time from completion of scan to publication of provisional report should be within 1 h. The locally agreed target for both standards was 100%.

Results: On review of the 40 CT scans, 32 (80%) were completed within 1 h of request. From the 8 scans (20%) not completed within the hour, 4 were due to porter unavailability, 1 due to an uncooperative patient and the remaining 3 reasons were not clear from documentation. Following completion of the scan, 38 scans (95%) were provisionally reported within 1 h.

Conclusion: This study highlighted a good compliance by hospital staff in ensuring patients with head injuries are managed appropriately, following detection of risk factors indicating a CT head scan. Having said that, the locally agreed targets were just short of being met. One factor resulting in delayed scans was porter availability. An intervention recently introduced is the use of the “e-portering” application, which will endeavour to save time for referrers requesting porters and allow patient tracking. It is also worth educating porters, via email bulletins, on the importance of priority scans, such as CT head following injury. Furthermore, the findings of the audit were relayed to the radiology department to help improve reporting times and to the ED to re-emphasize prompt requesting of CT head scans when clinically indicated.

Compliance with ethics regulations: Yes.

F-106 Does continuous insufflation of oxygen without intubation improve mortality after cardiac arrest?

Arnaud Gaillard, Cécile Ricard, Guillaume Berthet, Olivier Baptiste, Vincent Peigne

Haute Savoie Fire Department, Meythet, France

Correspondence: Arnaud Gaillard (arnaud.gaillard@sdis74.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-106

Rationale: Continuous insufflation of oxygen (CIO) performed with specific endotracheal tube during cardiopulmonary resuscitation (CPR) is as effective as intermittent ventilation on endotracheal tube. Experimental data suggest that CIO improves the efficacy of external cardiac massage and reduces gastric dilatation. As endotracheal intubation is a cause of CPR interruption and requires skilled staff, a specific device has been developed to perform CIO without intubation. This device has been implemented progressively in our fire department since 2015. We evaluated this practice.

Patients and methods: Longitudinal study comparing the patients with out-of-hospital cardiac arrest managed by our fire department with CIO or bag-valve ventilation between January 2015 and April 2019. Patients who received mechanical chest compression were excluded. The main outcome was hospital survival. Secondary outcomes were the return of spontaneous circulation (ROSC) and CPR quality. Univariate and multivariate analysis was performed in the whole cohort and in the sub-groups of patient with shockable and non-shockable rhythms to take into account factors associated with survival (shockable rhythm, witness, age).

Results: Among the 793 patients included, 262 have been ventilated with CIO and 531 with valve-bag. The mortality was similar in the two groups (CIO: 85.1% valve-bag: 88.1% p 0.23). Mortality and ROSC were not associated with CIO in the multivariate analysis (odds ratio OR 1.19 95%-confidence interval CI95 [0.75–1.9] and 0.97 [0.59–1.6], respectively). CPR quality was better with CIO than with valve-bag regarding CPR fraction (ratio of duration of chest compressions on total duration of CPR, 84 versus 76% p < 0.0001) and adequacy to the guidelines of the rhythm and depth of chest compressions (70% vs 50% p < 0.0001 and 47% vs 38% p < 0.0001, respectively). In both subgroups of patients, CPR quality was still better with CIO than with valve-bag. In the subgroup of patients with shockable rhythm, univariate analysis showed a lower mortality among the 59 patients with CIO than among the 140 patients with valve-bag (57.6% vs 82.9% p < 0.001) but this difference was not confirmed by the multivariate analysis (OR 0.4 CI95 [0.10–1.03], p 0.06).

Conclusion: CIO without intubation is associated with an improvement of CPR quality but neither with mortality nor return of spontaneous circulation in case of out-of-hospital cardiac arrest.

Compliance with ethics regulations: Yes.

F-107 Can New Wearable Technology significantly increase the Efficacy Cardiopulmonary Resuscitation? A controlled, randomized trial testing

Michele Musiari¹, Samuele Ceruti², Roberta Sonzini³, Luciano Anselmi⁴, Andrea Saporito⁴, Xavier Capdevila⁵, Tiziano Cassina⁶

¹Hôpital Fribourgeois Regional (HFR), Fribourg, Switzerland; ²Hôpitaux Universitaires de Genève, Genève, Switzerland; ³Clinica Luganese, Lugano, Switzerland; ⁴Ente Ospedaliero Cantonale, Bellinzona, Switzerland; ⁵Centre Hôspitalier Universitaires, Montpellier, France; ⁶CardioCentro Ticino, Lugano, Switzerland

Correspondence: Michele Musiari (michelemusiari@bluewin.ch)

Ann. Intensive Care 2020, 10 (Suppl 1):F-107

Rationale: Cardiovascular accidents are a leading cause of death. A cardiopulmonary resuscitation (CPR) of quality has well shown that can reduce the mortality; despite this, survival rate has not changed significantly during last years. The aim of this study is to test a new wearable glove to provide lay people with instructions during out-of-hospital CPR.

Patients and methods: We performed a blinded, controlled trial on an electronic mannequin AmbuMan to test the performance of adult volunteers, non-healthcare professionals performing a simulated CPR both, without and with glove, following the glove instructions. The group without glove, also called “no-glove” is intended as control group. Each compression performed on the electronic mannequin AmbuMan was recorded by a connected laptop computer, drawing a depth frequency curve over the time. Primary outcome was to compare the accuracy of the two simulated CPR sessions in terms of depth and frequency of chest compressions performed by the same lay volunteers. Secondary outcome was to compare the decay of performance and percentage of time in which the candidate performed accurate CPR. Finally, the participants were asked if the glove was useful for CPR maneuvers. The difference between the two groups in regard to change in chest compression depth over time due to fatigue, defined as decay were also analyzed.

Results: 571 chest compressions were included: 293 in control group, 278 in glove group (Table 1). Mean depth of compression in the control group was 55.17 mm versus 52.11 mm in the glove-group (p = 0.000016). Compressions with an appropriate depth were not statistically different (81.9% vs 73.6%, p = 0.017). Mean frequency of compressions in the group with glove was 117.67 rpm vs 103.02 rpm in the control group (p < 0.00001). The percentage of compression cycles with an appropriate rate (> 100 rpm) was 92.4% in the group with the glove versus 71% in the control group, with an observed difference of 21.4% between the two groups, which was statistically significant (p < 0.0001, CI = 95%). A mean reduction over time of compressions depth of 5.3 mm (SD 10.28) was observed in the control group versus a mean reduction of 0.83 mm in the group wearing the glove (SD 9.91), but this mean difference in the decay of compressions delivery was not statistically significant (f-ratio = 5.680, SS = 579.61, df = 1, MS = 579.61, p = 0.018).

Conclusion: The visual and acoustic feedbacks provided by the device were useful in dictating the correct rhythm for non-healthcare professionals, translating in a significantly more accurate CPR.

Compliance with ethics regulations: Yes.

Fig. 1

Differences about compression frequency and cycles with rate greater than 100 rpm during CPR between the two groups. All values are intended as mean values

F-108 Early DiaGnosis of Anoxic brain injury for Resuscitated patients (EDGAR): value of continuous aEEG, quantitative automated pupillometry and Cardiac Arrest Hospital Prognosis (CAHP) score for early neurologic prognosis after adult cardiac arrest

Sébastien Gette, Rostane Gaci, Guillaume Louis, Cyril Cadoz, Damien Barraud, Boris Glavnik, Guillaume Jay, Julie Larde, Nicolas le Berre, Nouchan Mellati, Emmanuel Novy, Jessica Perny, Adeline Perrein, Yoann Picard, Kevin Podrez, Béatrice Schnitzler, Christophe Goetz, Serge Le Tacon

Hôpital Mercy CHR Metz-Thionville, Metz, France

Correspondence: Sébastien Gette (s.gette@chr-metz-thionville.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-108

Rationale: Neuroprognostication after cardiac arrest (CA) is a crucial issue and current guidelines recommend delayed multimodal approach. We aimed to describe reasons for death in a prospective cohort of CA patients and evaluate the diagnostic accuracy of early combined neurological prognostication tools such as automated pupillometry (AP), continuous amplitude electroencephalography (aEEG) and Cardiac Arrest Hospital Prognosis (CAHP) score performed 24 h after return of spontaneous circulation (ROSC).

Patients and methods: We set up a monocentric prospective cohort of adult CA patients admitted in ICU after sustained ROSC and collected data according to Utstein style recommendations. Reasons for death were described under recently proposed classification: withdrawal of life-sustaining therapies (WLST) for neurological reasons, WLST due to comorbidities, refractory shock or recurrence of sudden CA or respiratory failure. For patients who kept abnormal neurologic state after ROSC with Glasgow Coma Scale < 15, we analysed accuracy of early neuroprognostication tools (AP, aEEG and CAHP score) to predict poor neurological outcome, i.e. Cerebral Performance Category (CPC) > 2 at hospital discharge.

Results: 144 patients were admitted after sustained ROSC from CA during the period (31.08.2018 to 10.07.2019). In-hospital mortality was 51%. Neurological WLST was the first reason for death (62%). Exhaustive early neuroprognostication with AP, aEEG and CAHP score was available for 75 patients. Among them, poor neurological outcome at hospital discharge (CPC > 2) was observed for 39 patients (52%). AP < 13% performed at H24 always predicted poor neurological outcome (PPV 100%; CI 95% [0.84–1]). Type I aEEG was associated with good neurological outcome (PPV 82%; CI 95% [0.62–0.92]). Prognostic values of combined tests (CAHP score + AP + aEEG) were respectively PPV 0.82 (CI 95% [0.67–0.91]) and NPV 0.81 (CI 95% [0.62–0.92]) if at least one positive, PPV 0.97 (CI 95% [0.8–1]) and NPV 0.70 (CI 95% [0.54–0.82]) if at least 2 positives, PPV 1 (CI 95% [0.87–1]) and NPV 0.55 (CI 95% [0.42–0.68]) if 3 positives.

Conclusion: Most deaths after CA resuscitation occur after WLST because of poor neurologic prognosis. A stepwise model with automated pupillometry and then aEEG if AP > 13% could discriminate about the neurologic prognosis at hospital discharge 72% of patients as soon as 24 h after ROSC (Fig. 1). This strategy would falsely misclassificate 4% of patients in a good neurologic outcome category. Other survivors (24%) should then be investigated with further classical delayed neuroprognostication tools.

Compliance with ethics regulations: Yes.

Fig. 1

Stepwise model for early neuroprognostication

F-109 Impact of delayed specialized advice on the quality of care at the emergency department

Hadil Mhadhbi, Meher Arafa, Mounir Hagui

ED HMPIT, Tunis, Tunisia

Correspondence: Hadil Mhadhbi (hadil.mhadhbi@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-109

Rationale: Management delay is one of the determining factors in the assessment of emergency department quality of care. Asking for a specialized advice seems to increase the time of delay. Our study aimed at measuring the delays in obtaining specialized advice and identify their major causes.

Patients and methods: We conducted a prospective study over the period of 1 month. We included all adult patients presenting to the emergency department who required specialized advice. Data of all patients was collected. Waiting times and influencing factors were studied.

Results: A total of 75 patients were included. The main reason for calling for a specialized advice was to ask for a department transfer in 68% of cases. The time of the day when specialized advice was solicited (n (%)): in the morning 51 (68); in the afternoon 17 (23); in the evening 9 (12). The main solicited specialties were (n (%)): visceral surgery 18 (24), trauma medicine 15 (20), cardiology 10 (13), urology 5 (7), and pulmonology 5 (7). The average waiting time between calling for and getting the specialized advice was 176 ± 115 min. Seventy-five percent of the specialized advice was obtained within 1 h. The causes of the delay were (n (%)): Physician busy in the operating room 22 (39), unreachable physician 9 (12), physician in the outpatient clinics 7 (10). The impact of the waiting time was (n (%)): conflict 20 (27), worsening patient state 4 (5). The average time between calling for the specialized advice and reaching a management decision was 210 ± 127 min.

Conclusion: The increasing length of stay of patients in the ED is strongly correlated to the delay in obtaining specialized advice. The implementation of a strategy to reduce the waiting time is necessary to avoid overcrowding the emergency departments and provide optimal care.

Compliance with ethics regulations: Yes.

F-110 Hypnosis to reduce anxiety and pain during prolonged ICU stay for Guillain-Barré syndrome (GBS)

Sylvie Calvat¹, Sylvie Colombani², Christophe Cracco¹, Charles Lafon¹, Stéphane Rouleau¹, David Schnell¹

¹Centre Hospitalier d’Angoulême, Angoulême, France; ²Institut Bergonié, Bordeaux, France

Correspondence: Sylvie Calvat (sylvie.calvat@ch-angouleme.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-110

Rationale: Hypnoanalgesia has been used since few years to reduce ICU-patients physical and psychological discomfort during invasive procedures. However, feasibility of overall well-being management of intubated patients with hypnosis has not been described.

Patients and methods: We report here the hypnotic accompaniment of a 16-year old patient without significant medical history hospitalized in our ICU for a severe GBS during 3 months. The GBS was diagnosed by electrophysiological study and immunologic markers. Patient had nearly complete paralysis of all extremities, but no facial or bulbar muscles. He received mechanical ventilation during 87 days, including weaning time. Tracheotomy was performed at day 15. Sedative drugs were stopped 2 days after intubation. Hypnosis sessions were started very early after intubation by one of our trained intensivist. Eight hypnotic sessions of hypnoanalgesia or hypnotherapy were performed after approval of the patient and his parents. Time distribution is reported in Fig. 1. First and second sessions were performed in order to induce relaxation and reduce anxiety. Following sessions were dedicated to: 1) decrease pain intensity (initially neuropathic, then induced by physiotherapy), 2) attenuate the negative perception of paralysis, 3) reduce the discomfort of tracheotomy 4) promote the belief in healing 5) facilitate swallowing exercises. Furthermore the patient was quickly trained to use self-hypnosis in order to dissociate him from pain, anxiety and ICU pollutions.

Results: Feasibility of hypnosis was judged satisfactory by the operating physician, despite mechanical ventilation. After extubation, final debriefing with the patient indicates that the most efficient sessions were those focused on anxiety disorders (using the suggestion of a safe place) and suggestions of mobility (using a mangas metaphor). The patient reported very positive perception of hypnosis use. He explained that self-hypnosis was effective to reduce many discomfort. He used it frequently (generally twice a day) for a puff of anxiety or before enoxaparin injection.

Conclusion: Our observation suggests that hypnosis seems feasible in ICU—awake patients and may be an interesting way to improve their ICU lived experience in combination with validated measures. Further investigations are needed to evaluate its effects on post-traumatic-stress disorder.

Compliance with ethics regulations: Yes.

Fig. 1

Time distribution of the eight hypnotic sessions

F-111 Resuscitation Hypnosis: Benchmarks for Effective Practice

Gwenael Prat¹, Anne Renault¹, Christelle Teiten¹, Montaine Lefevre², Erwan L’her¹

¹Medical Intensive care unit, CHRU Brest, Brest, France; ²Medical Intensive care unit, CH Morlaix, Morlaix, France

Correspondence: Gwenael Prat (gwenael.prat@chu-brest.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-111

Rationale: There is little medical reference for hypnosis in the intensive care field. Closed specialties such as anesthesia, emergency medicine can help and refer to hypnosis for certain technical procedures. Objective: To propose landmarks for a successful implementation of hypnosis by intensivists within the intensive care unit.

Patients and methods: This monocentric prospective observational study was performed from February 2018 to June 2019 in the 15-bed medical ICU of Brest University Hospital. Collected data were: characteristics of patients and hypnosis sessions performed, demographic data, physiological parameters (heart and respiratory rates) and objective and subjective evaluation of hypnosis sessions quality.

Results: 60 patients were included (mean age 55.4 ± 19 years, SAPS II 34.5 ± 14 points). 100 hypnosis sessions were performed, of which 1/3 under mechanical ventilation. Patterns of hypnosis sessions were: anxiety/comfort (53%), during a technical procedure (38%): TOE, CVC placement, thoracic drainage, upper digestive or bronchial endoscopy), initiation of noninvasive ventilation or before intubation. Most of time, the hypnotic trance was permitted by formal hypnosis techniques with travel and nature themes suggestion. Efficacy was qualitatively assessed and rated as “total effectiveness” for 60% of sessions. Qualitative evaluation by hypnotherapist, technical operator and observers was respectively 7.25 ± 1.5, 9.25 ± 0.5 and 9 ± 1/10. Heart rate decreased from 93 ± 17 to 88 ± 13 bpm and respiratory rate/min decreased from 22 ± 7 to 17.7 ± 5 rpm during sessions.

Discussion: After a meeting, the healthcare team carried out a brainstorming to propose hypnosis in our unit. Several difficulties were observed to explain implementation failures such as: finding competent patient, respiratory assistance, difficult communication, noisy environment, many nursing care, unexpected emergencies, etc.…). This experience allowed writing a vademecum to perform hypnosis in intensive care. Our aims are to get more trained caregivers and to integrate hypnosis during our post-resuscitation consultation, especially for post-traumatic stress.

Conclusion: Hypnotic tools can facilitate technical procedures and improve patients’ and caregivers’ quality of life within the ICU.

Compliance with ethics regulations: Yes.

F-112 Effect of a musical intervention during central venous catheterization in an intensive care unit: the MUSIC CAT prospective randomized pilot study

Sophie Jacquier, Brice Sauvage, Gregoire Muller, Thierry Boulain, Mai-Anh Nay

CHR, Orléans, France

Correspondence: Sophie Jacquier (sophie.jacquier@chr-orleans.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-112

Rationale: Evaluate the effect of a musical intervention on patient anxiety during a central venous access or a dialysis catheter implantation in an intensive care unit.

Patients and methods: The MUSIC CAT study was a prospective, single-centre, controlled, open-label, two-arm randomized trial, conducted from February 2018 to February 2019. Central venous catheterization with musical intervention was compared to standard care, i.e., the usual procedure of central venous catheterization without listening to music. Eligible patients had to be able to hear, understand explanations and consent. Randomisation was stratified according to ventilation type (mechanical ventilation or not) and catheter site (superior vena cava or femoral vein). The MUSIC CARE^® (Paris, France) application was used to make the patients listen to music through headphones. Each patient chose his/her musical topic on a digital tablet, just before the catheterization. The primary outcome was the change in anxiety Visual Analogic Scale (VAS) between the beginning and the end of the catheterization procedure (T0-TF anxiety VAS). Secondary outcomes included the patient’s pain VAS at the end of the procedure (TF pain VAS).

Results: 31 patients were included in the standard care group versus 36 in the musical intervention group. Main reasons for admission were the need of central catheter for chemotherapy (27, 40%), and sepsis and/or shock in both groups (26, 39%). Catheters were inserted in the internal jugular vein in most cases (56, 82%) and about one-third were tunnelled in both groups. There was no between-group difference regarding median T0-TF anxiety VAS: 0 [IQR:− 3 to 0] in the standard care group versus − 1 [− 3 to 0] in the music intervention group (P = 0.20) (Fig. 1), with no significant interaction between the variables of stratification or the operator experience and the intervention. The median TF pain VAS was not statistically different between groups: 0 [0 to 4.1] in standard care group and 0 [0 to 2] in music intervention group (P = 0.60), with no significant interaction between the variables of stratification or the operator experience and the intervention.

Conclusion: In this first randomized pilot study of musical intervention for central venous catheterization in awake patients in the intensive care unit, the musical intervention did not reduce patients’ anxiety as compared to usual care. As the study may have been underpowered, larger size trials are needed.

Compliance with ethics regulations: Yes.

Fig. 1

Changes in anxiety VAS T0-TF in both groups

F-113 Sleep in non-intubated patients admitted to ICU for acute respiratory failure: A prospective observational monocenter study

Damien Marie^1,2, Stephanie Barrau^1,2, Clément Beuvon^1,2, Faustine Reynaud^1,2, Adrien Pépin-Lehalleur^1,2, Christophe Rault^1,3, Jean-Claude Meurice^1,4, Rémi Coudroy^1,2, Jean-Pierre Frat^1,2, René Robert^1,2, Xavier Drouot^1,3, Arnaud Thille^1,2

¹INSERM CIC 1402, Groupe ALIVE, Université de Poitiers, Poitiers, France; ²Réanimation Médicale, CHU de Poitiers, Poitiers; ³Neurophysiologie Clinique et Explorations Fonctionnelles, CHU de Poitiers, Poitiers; ⁴Pneumologie, CHU de Poitiers, Poitiers

Correspondence: Damien Marie (damien.marie01@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-113

Rationale: Sleep is markedly altered in ICU-patients under mechanical ventilation and may be due to noise, light, patient-care activities, patient-ventilator asynchronies, or the result of acute brain dysfunction induced by sedative drugs. To our knowledge, sleep has never been studied at ICU admission before any sedation. Our study aimed at assessing sleep quality of non-intubated sedation-free patients admitted to ICU for acute respiratory failure.

Patients and Methods: Observational study performed in a single centre of a teaching hospital. Patients admitted to ICU for acute respiratory failure (respiratory rate ≥ 25 breaths/min and PaO₂/FiO₂ < 300 mm Hg under high-flow nasal oxygen) could be enrolled. Patients with hypercapnia, central nervous disease, intubated early after admission and those with a do-not-intubate order were excluded. Sleep was evaluated by complete polysomnography (PSG) that started in the afternoon following admission and was continuously performed until the next morning.

Results: Over a 3-year period 128 patients were screened and 44 patients were included. Among them, 10 patients were excluded for the following reasons: 1 patient was intubated shortly after PSG initiation, 1 PSG was lost, and 8 EEG recordings (18%) were stopped before midnight (electrodes turned off or loss of signal). Therefore, 34 patients in whom PSG was complete during the nocturnal period were retained in the analysis (27 males [74%], age 62 years [60–69], SAPS II 32 [24–37], PaO₂/FiO₂ 133 mmHg [100–186], PaCO₂ 34 mmHg [30–39]). The recording time of polysomnography was 16 h in median [IQR 15–18] and total sleep time was 4.2 h [2.6–6.9] (28% of the total recording time). Sleep architecture was close to normal sleep with a median light sleep time (stage N1 and N2) of 60% [42–71] and a median deep sleep time of 29% [14–44]. By contrast, rapid eye movement (REM) sleep was completely absent in 13 patients (38%) and represented only 2.4% of total sleep time [0–7.8], that is 9 min [0–29]. Sleep was fragmented by numerous arousals and awakenings (11 events [9–16] per hour of sleep in median). Five of the 34 patients needed intubation after PSG (15%).

Conclusion: Our results suggest for the first time that patients admitted to the ICU for acute respiratory failure have a sleep time and architecture close to normal during the first night after admission, except for an almost compete disappearance of REM sleep, an essential stage of sleep.

Compliance with ethics regulations: Yes.

F-114 Evaluation of Association of Secondary Brain Insults and Outcome in Convulsive Status Epilepticus: a Post Hoc Analysis of the Hybernatus Trial

Candice Fontaine¹, Virginie Lemiale², Matthieu Resche-Rigon², Maleka Schenck³, Jonathan Chelly⁴, Thomas Geeraerts⁵, Aicha Hamdi⁶, Christophe Guitton⁷, Ferhat Meziani⁸, Jean-Yves Lefrant⁹, Bruno Mégarbane¹⁰, Hervé Mentec¹¹, Cendrine Chaffaut², Alain Cariou¹², Stéphane Legriel¹

¹Centre Hospitalier de Versailles, Le Chesnay, France; ²Centre Hospitalier Universitaire Saint Louis, Paris, France; ³Hôpitaux Universitaires de Strasbourg-Hôpital de Hautepierre, Strasbourg, France; ⁴Centre Hospitalier de Melun, Melun, France; ⁵Centre Hospitalier Universitaire de Toulouse, Toulouse, France; ⁶Centre Hospitalier De Montreuil, Montreuil, France; ⁷Centre Hospitalier du Mans, Le Mans, France; ⁸Hopitaux Universitaires de Strasbourg-Nouvel Hopital Civil, Strasbourg, France; ⁹Hopital Universitaire de Nimes, Nimes, France; ¹⁰Centre Hospitalier Universitaire Lariboisiere, Paris, France; ¹¹Centre Hospitalier d’Argenteuil, Argenteuil, France; ¹²Centre Hospitalier Universitaire Cochin, Paris, France

Correspondence: Candice Fontaine (candicefontaine@yahoo.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-114

Rationale: Convulsive status epilepticus (CSE) is a common neurological emergency associated with high mortality and morbidity rates. There are strong experimental data suggesting a potential impact of secondary brain insults (SBI) on outcome after CSE. However, there is no clinical proof to support this hypothesis. Our objective was to evaluate the association between SBI (mean arterial blood pressure, arterial partial pressure of carbon dioxide, arterial partial pressure of oxygen, temperature, natremia, and glycemia) at day 1 and neurological outcomes 90 days after CSE.

Patients and methods: This was a post hoc analysis of the HYBERNATUS multicenter open-label clinical trial randomized 270 critically ill patients with CSE requiring mechanical ventilation to either therapeutic hypothermia (32–34 °C for 24 h) plus standard care or standard care alone. 265 patients still alive at day 2 after inclusion were enrolled from March 2011 to January 2015 in 11 French medico-surgical ICUs. The primary outcome was favourable outcome 90 days after CSE defined as a Glasgow Outcome Scale score of 5.

Results: Median age was of 57 years [45–68]. A previous history of epilepsy was noted in 130 (49%) patients. Most episodes (173/265, 65%) occurred out-of-hospital, and 230 (87%) were witnessed from their onset. CSE was refractory in 86 (32%) patients and total seizure duration was 67 min (35–120). A favorable 90-day outcome occurred in 126 (48%) patients. Maximal glycemia value and hyperglycemia > 9.9 mmol/L at day 1 were the only SBI variables associated with outcome in univariate analysis. By multivariate analysis, age > 65 years (OR, 0.46; 95% IC, 0.26–0.83; P = 0.01), refractory CSE (OR, 0.50; 95% IC, 0.26–0.96; P = 0.04), and primary brain insult (OR, 0.50; 95% IC, 0.25–0.99; P = 0.047) were associated with an increased risk of poor outcome, and a bystander-witnessed onset of CSE (OR, 2.49; 95% IC, 1.05–5.59; P = 0.04) was associated with a decreased risk of poor outcome.

Conclusion: In our population, secondary brain insults were not associated with outcome in critically ill patients with convulsive status epilepticus; whereas age, bystander-witnessed onset of status epilepticus, refractory status epilepticus and primary brain insult were identified as strong predictors of 90-day functional impairment. Further studies are warranted to confirm our findings.

Compliance with ethics regulations: Yes.

F-115 Acute stroke admitted to the ICU: prognosis and functional outcome

Thibaut Carval¹, Charlotte Garret¹, Benoit Guillon², Amélie Seguin¹, Maelle Martin¹, Arnaud-Felix Miailhe¹, Hélène Migueres¹, Olivier Zambon¹, Laura Crosby¹, Jean-Baptiste Lascarrou¹, Jean Reignier¹, Emmanuel Canet¹

¹Medical ICU, Nantes, France; ²Neurology department, Nantes, France

Correspondence: Thibaut Carval (thibautcarval@hotmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-115

Rationale: Acute stroke (AS) is a leading cause of morbidity and mortality worldwide. However, data on the prognosis and functional outcome of patients with AS requiring ICU management is limited. Our purpose was to identify factors associated with good outcome (defined by a modified Rankin score (mRS) of 0–2) 6 months after ICU admission.

Patients and methods: Retrospective cohort of patients admitted to the medical ICU of a university-affiliated hospital between January 2014 and December 2018 and coded for acute stroke using the ICD-10 criteria. Patients with traumatic stroke and isolated subarachnoid hemorrhage were excluded.

Results: We identified 323 patients. Median age was 67 [54.5–77] years and 173 (53.6%) were males. Main reasons for ICU admission were coma (87%), hemodynamic instability (28.2%), acute respiratory failure (26%), and cardiac arrest (5.3%). Glasgow coma score at ICU admission was 6 [4–10] and SAPS 2 was 54 [35–64] points. Types of stroke were hemorrhagic in 248 (76.8%) patients and ischemic in 75 (23.2%). Mechanical ventilation was required in 257 patients (79.6%). Seizures occurred in 11.8% of the patients and convulsive status epilepticus in 3.1%. Pneumonia was diagnosed in 60 (18.6%) patients (aspiration pneumonia n = 20, ventilator associated pneumonia n = 40). Thrombolysis or thromboaspiration were performed in 18 (24%) patients with ischemic stroke. Surgical evacuation of expanding hematoma was performed in 21 (6.5%) patients, 19 (5.9%) had craniectomy, and 61 (18.9%) had external shunt for hydrocephalus. ICU and hospital mortality were 58.8% and 61%, respectively. Six months after ICU admission, 44 (13.6%) patients had a good outcome (mRS 0–2), 42 (13.1%) had significant disability (mRS 3–5), and 200 (61.9%) were deceased (lost follow-up n = 37, 11.5%). On multivariable analysis, age (OR 0.95 per year (0.91–0.99), p = 0.01), SAPS 2 (OR 0.95 per point (0.90–0.99), p = 0.03), and hemorrhagic stroke (OR 0.27 (0.08–0.9), p = 0.03) reduced the likelihood of good outcome (mRS 0–2) 6 months after ICU admission.

Conclusion: In our study, prognosis of acute stroke requiring ICU admission was poor and a good functional outcome occurred in less than 15% of the patients at 6 months. Age, severity at ICU admission, and type of stroke predicted outcome.

Compliance with ethics regulations: Yes.

F-116 Hemorrhagic brain injuries in critical care units: a systematic review of prognostic models

Jeanne Simon-Pimmel¹, Yohann Foucher^1, Maxime Leger², Fanny Feuillet³, Laetitia Bodet-Contentin⁴, Raphaël Cinotti⁵, Denis Frasca⁶, Etienne Dantan¹

¹INSERM UMR 1246 SPHERE, Nantes University, Tours University, Nantes, France; ²Medical Intensive Care, Angers University Hospital, Angers, France; ³INSERM UMR 1246 SPHERE, Nantes University, Tours University-Biostatistics and Methodology Unit, Nantes University Hospital, Nantes, France; ⁴Intensive Care Unit, Tours University Hospital, Tours, France; ⁵Anaesthesia and critical care department, Nantes University Hospital, Nantes, France; ⁶Anaesthesia and critical care department, Poitiers University Hospital, Poitiers, France

Correspondence: Jeanne Simon-Pimmel (jeanne.pimmel@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-116

Rationale: In intensive care units, severe spontaneous hemorrhagic brain injuries have a poor prognosis for mortality and functional outcomes. Affected patients face particular ethical issues regarding the difficulty of anticipating their eventual recovery. In this context, prognostic scores can help clinicians in patients/relatives counseling and therapeutic decisions. The previous reviews pointed out many prognostic tools for intracranial hemorrhage and subarachnoid hemorrhage but did not focus on injuries explicitly severe nor assessed the methodological limitations of the models. Our systematic review aimed to assess methodologically prognostic tools for functional outcomes in severe spontaneous haemorrhagic brain, with particular attention to their clinical utilities.

Patients and methods: Following PRISMA recommendations, we queried Medline, Embase, Web of Science, and the Cochrane by February 19, 2019. We included multivariate prognostic models explicitly developed or validated on adults with severe intracranial or subarachnoid haemorrhage. We evaluated the articles following the CHARMS recommendations (CHecklist for critical Appraisal and data extraction for systematic Reviews of prediction Modelling Studies) and the TRIPOD statements (Transparent Reporting of a multivariable prediction model for Individual Prognosis.

Results: Our review confirmed the multiple publications of prognostic scores, as we found 71 articles aiming to develop or validate prognostic tools. Relying on guidelines, we discarded 52 articles due to the lack of prognostic capacities, validation, or predictor selection. 8 articles developed and validated a prognostic tool and 19 externally validated existing models (Fig. 1). No score was of good methodological quality in intracranial hemorrhage. We highlighted two prognostic scores in subarachnoid hemorrhages: the SAHIT predicting unfavorable outcome or mortality at 6 months and the FRESH predicting unfavorable outcome at 12 months.

Conclusion: Although prognostic studies on haemorrhagic brain injuries abound in the literature, they generally lack of methodological robustness or show incomplete reporting. With the numerous published scores, we believe that it is time to stop developing new scores. Ongoing validation, recalibration, and impact studies would keep improving existing good tools. The use of “patient-centered” approaches could also enhance them, and be more appropriate to inform patients and families about their long-term potential recovery. These considerations should drive future research in the modern era of neurocritical care prognosis.

Compliance with ethics regulations: NA.

Fig. 1

Flow chart of the systematic review

F-117 RETRACT System: a computerized 3D video analyzing system for the monitoring of respiratory retraction signs in children

Haythem Rehouma¹, Gabriel Masson², Sandrine Essouri², Rita Noumeir³, Philippe Jouvet²

¹École de Technologies Supérieure (ETS) of Montreal (QC) Canada, Montreal, Canada; ²CHU Sainte-Justine (CHUSJ), Mother and Child University Hospital Center, Montréal, Canada, Montreal, Canada; ³École de Technologies Supérieure (ETS) of Montreal (QC) Canada, Montreal, Canada

Correspondence: Haythem Rehouma (rhoumahaythem@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-117

Rationale: Respiratory pattern analysis by a visual examination is an important part of clinical assessment but is dependent on caregiver expertise and is subjective. Furthermore, there is no easy medical device used in PICU to measure Tidal Volume (Vt) and Minute Ventilation (MV) in spontaneous breathing patients. The clinical research unit in critical care of CHUSJ and ETS have developed a non-invasive computerized 3D video analyzing system (RETRACT System) to detect and perform a video analysis of respiratory status in children. The aim of this study is to test the reliability of the RETRACT System to monitor respiratory distress in critically ill children.

Patients and methods: The RETRACT System is detailed in reference 1. In summary, cameras reproduce in 3D the thorax and abdomen of a subject. The respiratory status (respiratory rate (RR), tidal volume (Vt), minute ventilation (MV)) assessed by the RETRACT System was compared on a bench test (high-fidelity mannequin) and in critically ill children, to the ventilator measurements and clinician expert evaluation (gold standard). Bland–Altman plots were used for comparison.

Results: We observed a significant agreement, on mannequin, between RETRACT System and gold standard method in estimating Vt, RR and MV, i.e. 95% of the paired differences were within the limits of agreement in Bland–Altman plots, as illustrated in Fig. 1. In critically ill children (n = 2), the correlation between the pairs of measures was also high (r > 0.95, p < 0.001) and the coefficient of determination with a high fit (0.90 < R2 < 0.99, p < 0.0001). For good correlation, the RETRACT System needed to have a visual access to thorax and abdomen in a quiet subject.

Conclusion: The RETRACT System measurements of Vt, RR and MV for respiratory distress monitoring in patients seems reliable. More testing are required to validate this method in usual practice and to develop the retractions signs video analysis.

Compliance with ethics regulations: Yes.

Fig. 1

Bland-Altman plots of phantom’s quantitative measures. The plots are showing the agreement between the gold-standard method (ventilator) and our method (3D system) in estimating Vt (a), RR (b), MV (c) quantitative measures

Reference

• 1. Rehouma H, Noumeir R, Bouachir W, Jouvet P, Essouri S. Comput Med Imaging Graph 2018;70:17–28.

F-118 Design and evaluation of a training support for pediatric tracheostomy care

Thibault Lassalle, Thomas Sagardoy, Astrid Botte

CHU Bordeaux, Bordeaux, France

Correspondence: Thibault Lassalle (thibaultlassalle40@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-118

Rationale: Pediatric tracheostomy care of the child is unknown and feared by health professionals. The e-learning courses include the advantage of training a maximum of people over a minimum of time, and harmonize everyone’s knowledge. The objective of this study is to create and validate a training video for pediatric tracheostomy care, accessible by e-learning for health professionals.

Patients and methods: To design this training video, we followed the ADDIE 5-step educational engineering model: context and needs analysis, video design, video development, video implementation, evaluation of the effectiveness of the video. The evaluation of the video was carried out according to the first 3 phases of the Kirkpatrick model: first, learner satisfaction, second, learning evaluation (knowledge assessment and self-assessment of comfort before video) before video, immediately after video and at 3 months, third behavior modification. We have included health professionals performing tracheostomy care (pediatric nurses, childcare auxiliaries, physiotherapists, medical residents, senior doctors) in the main tracheostomy treatment centers in the New Aquitaine region who have agreed to participate to our study.

Results: 262 health professionals were included. 163 responded 3 months after the video. Learners were satisfied with the video. Their knowledge and skills were significantly improved (p < 0.001) (Fig. 1). Their improvement persisted at 3 months (p < 0.001). 93% of health professionals had modified their practice at 3 months.

Conclusion: This study made it possible to create and validate a training video for pediatric tracheostomy care which is accessible by e-learning for health professionals.

Compliance with ethics regulations: Yes.

Fig. 1

Knowledge evaluation

F-119 Nurse-driven non-invasive ventilation weaning protocol in severe bronchiolitis in PICU is feasible and safe

Camille Guillot, Mathilde Periot-Jarry

Hopital Jeanne de Flandre-CHU Lille, Lille, France

Correspondence: Camille Guillot (cami.guillot@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):F-119

Rationale: Severe bronchiolitis requires hospitalization in paediatric intensive care unit (PICU). Non-invasive ventilation (NIV) has been demonstrated to treat them since twenty years, its use is well defined but there is no consensus for the weaning. This study evaluated the application of a nurse-driven NIV weaning protocol in hospitalized infants with severe bronchiolitis and verified its safety.

Patients and methods: This was a retrospective monocentric study in a PICU of Robert Debré hospital-Paris, France. In the epidemic period of bronchiolitis between 2015 and 2017, all patients under one year old with severe bronchiolitis and requiring NIV were included. Two groups were compared: one group using the nurse-driven NIV weaning protocol and one group without using this protocol. Occurrences of complications, duration of ventilatory support and length of stay (LOS) in PICU and total LOS were compared.

Results: 191 patients were included in the study, 115 in the no-protocol group, and 72 in the protocol group. The nurse-driven protocol was using at the rate of 90% (n = 72/80). The use of protocol did not increase the occurrence of complications (n = 45/115, 37.8% in no-protocol group versus n = 22/72, 30.6% in protocol group p = 0.308). The ventilatory support duration and LOS did not increased by the protocol utilization. The ventilation support total duration was 41 h [19–74] in the no-protocol group versus 49 h [29–89] in the protocol group (p = 0.082), and the CPAP (Continuous Positive Airways Pressure) duration was 27 h [17–49] in the no-protocol group and 30 h [13–51] in the protocol group (p = 0.522). PICU LOS were 3.5 days [3–5] in the no-protocol group versus 4 days [3–5.5] in the protocol group (p = 0.383), hospital LOS was 10 days [7–14] in the no-protocol group versus 10 days [7–12] in the protocol group (p = 0.447) (Fig. 1).

Conclusion: The use of this first nurse-driven NIV weaning protocol was feasible and simple with a very good application rate. Its utilization was safe. The occurrence of complications did not increase by the use of this protocol. It would allow an optimal NIV weaning without prolonging the ventilatory support duration nor PICU LOS or hospital LOS. The professional practices appeared to be coordinated and the nurses appeared to be more autonomous.

Compliance with ethics regulations: Yes.

Fig. 1

No-protocol and protocol groups comparison: CPAP duration (1), ventilatory support duration (2), PICU LOS (3), hospital LOS (4)

F-120 Weaning from Non-Invasive Ventilation and High-Flow Nasal Cannula therapy in acute bronchiolitis in infants—A survey of practice in French-speaking countries

Marie Suzanne¹, Isabelle Pin1, Guillaume Mortamet²

¹Service de pédiatrie, CHU de Grenoble, Grenoble, France; ²Service de réanimation pédiatrique, CHU de Grenoble, Grenoble, France

Correspondence: Marie Suzanne (msuzanne@chu-grenoble.Fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-120

Rationale: First-line management of severe acute bronchiolitis in infants is mainly based on Non-Invasive Ventilation (NIV) and High-Flow Nasal Cannula (HFNC) therapy. However, pediatric data regarding weaning from NIV/HFNC are lacking. This study aims to identify the weaning practices from NIV/HFNC in children with severe bronchiolitis.

Patients and methods: The WeaNIV-survey is a cross-sectional survey. A questionnaire was sent to French-speaking physicians with key roles in pediatric intensive care units.

Results: A total of 73% (24/33) of French University Hospital were represented in the study. Only 9% of pediatric centers used a protocol for weaning from NIV/HFNC and nurses were considered as key-actors of the weaning process for half of participants. Continuous Positive Airway Pressure (CPAP) was the mode of ventilation mainly used as the first-line therapy in clinical practice. The main criteria considered to initiate weaning process were: no or slight respiratory distress, a FiO₂ < 40%, a respiratory rate < 50/min and no significant apnea. Three strategies to discontinue NIV/HFNC were identified: 1/gradual decrease of ventilatory parameters (pressure or flow), 2/abrupt discontinuation and 3/gradual increase in off-ventilation time. Abrupt weaning strategy was the most commonly used, no matter the mode of ventilation. A significant level of respiratory distress, the presence of apneas, an increase in oxygen requirement, and a respiratory rate > 60/min were identified as weaning failure criteria by most pediatric intensive care physicians.

Conclusion: In most centers, the weaning process does not follow any protocol. Abrupt weaning seems to be commonly used as weaning strategy in children with severe bronchiolitis supported by NIV/HFNC. Based on the study findings, we suggest that criteria for weaning initiation and for weaning failure must be defined and weaning protocols generated.

Compliance with ethics regulations: Yes.

F-121 Prone positioning for severe pediatric acute respiratory distress syndrome: an observational study in four French intensive care units

Laura Labro, Jerome Rambaud

Pediatric and neonatal intensive care unit, Armand-Trousseau hospital, Sorbonne university, Paris, France

Correspondence: Laura Labro (laura.glt@wanadoo.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-121

Rationale: Prone positioning is frequently used for management of adult Acute Respiratory Distress Syndrome (ARDS). However, no positive effects were proved in pediatric population. The first objective of our study was to evaluate the use and the efficacy of prone positioning in severe Pediatric-ARDS.

Patients and methods: We performed a retrospective descriptive multicenter study including all children aged from 28 days to 18 years old and suffering from severe ARDS in four referral paediatric intensive care units from January 2007 to December 2018. We evaluated the use of prone positioning, oxygenation parameters and ventilator settings at the beginning, 6 h and the end of positioning, survival and complications secondary to prone positioning.

Results: 93 patients were included indeed 59 (63%) benefited from prone positioning for a mean of 2.5 runs for 18.2 h. We found a decreasing of OSI (17.8 vs 10.8, p = 0.001), OI (26 vs 13, p = 0.0006) and an improvement of the PaO₂/FiO₂ ratio (75 vs 101, p = 0.002) throughout the first 6 h of prone positioning and between the beginning and the end of prone positioning (17.8 vs 6.7, p < 0.001), (26 vs 8, p = 0.007) and (75 vs 144, p = 0.007). Unfortunately, the survival rate was not significantly higher in the prone positioning group (80% vs 62%, p 0.07). Complications secondary to prone positioning occured for 6 patients (12.2%).

Conclusion: This first study, which evaluate prone positioning efficacy in severe P-ARDS shows evidence that prone positioning improves oxygenation parameters and survival rate. These results highlight the necessity to develop a multicentric prospective randomized study to confirm these conclusions.

Compliance with ethics regulations: Yes.

F-122 Extracorporeal membrane oxygenation (ECMO) for immunocompromised children with acute respiratory distress syndrome (ARDS): a French ECMO center cohort

Jerome Rambaud, Blandine Robert

Pediatric and neonatal intensive care unit, Armand-Trousseau hospital, Sorbonne university, Paris, France

Correspondence: Jerome Rambaud (jerome.rambaud@aphp.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-122

Rationale: Immunocompromised children are likely to develop refractory acute respiratory distress syndrome (ARDS) and the usefulness of providing extracorporeal life support (ECLS) to these patients is a subject of continuing debate. The aim was to report outcomes and to compare factors associated with mortality between immunocompromised and non-immunocompromised children respiratory supported with ECMO.

Patients and methods: We performed a retrospective monocentric study in the pediatric French ECMO center of Armand Trousseau, including all pediatric patients aged from 1 month to 18 years requiring ECLS for ARDS.

Results: From 2007 to 2018, one hundred and eleven (111) patients were respiratory supported with ECMO; among them twenty-five (25) were immunocompromised. Survival rate 6 months after hospital intensive care discharge was significantly lower for immunocompromised patients compared to immunocompetent ones (41.7% vs. 62.8%; p = 0.04). ARDS severity was similar between groups. Fungal pneumonias were reported only for immunocompromised patients (12.5% versus 0% in the control group; p = 0.001). None of the clinical or biological data gathered prior to ECMO were different between the two groups. Bleeding complications were significantly more frequent in the immunocompromised group and blood product transfusions were also more frequently required in this group. There was no significant difference for the occurrence of nosocomial infections between groups.

Conclusion: Survival rate of immunocompromised children supported with ECMO for pediatric ARDS is lower than that of non-immunocompromised ones. But, the expectation for a favorable outcome is real and it is worth if the prognosis of their underlying disease is likely to be compatible with a good long-term quality of life.

Compliance with ethics regulations: Yes.

F-123 Impact of establishment of a pediatric mobile ECMO team

Géraldine Poncelet¹, Mathieu Genuini¹, Jérôme Naudin¹, Pierre-Louis Leger², Guillaume Geslain¹, Arielle Maroni¹, Maryline Chomton¹

¹Réanimation et surveillance continue pédiatriques, CHU Robert Debré, Paris, France; ²Réanimation et surveillance continue pédiatriques, CHU Armand Trousseau, Paris, France

Correspondence: Géraldine Poncelet (geraldine.poncelet@aphp.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):F-123

Rationale: Extracorporeal Membrane Oxygenation (ECMO) is used as a rescue therapy for critically ill children when usual respiratory and hemodynamic therapies fail to maintain vital functions. Since our hospital is not a pediatric ECMO center, children requiring this therapy need to be transferred to a referral ECMO center. Since November 2014, a pediatric mobile ECMO team composed of a medical transport team and a pediatric ECMO team is able to initiate ECMO in our Pediatric Intensive Care Unit (PICU). The aim of this study was to describe our population who required to be transported to a referral ECMO center.

Patients and methods: A retrospective observational study in a PICU, including all patients aged < 18 years old from 2014 to 2019 who required transfer to an ECMO center. Patient conditions, respiratory and hemodynamic parameters, time taken in charge by mobile ECMO team, immediately before cannulation were collected.

Results: Forty-three patients were included, median age was 21 months and 9 were newborns. Main indications for ECMO were Acute Respiratory Distress Syndrome (40%), refractory septic shock (23%), cardiac diseases (16%), Pertussis (9%), diaphragmatic hernia (5%) and cardiac arrest (2%). Survival at discharge was 53%. Number of organ failures (4 vs 2) and Vasoactive-Inotropic Score (VIS) (205 vs 50) were significantly higher in the non-survivor group. Cannulation was veno-venous (14%) or veno-arterial (67%) and 8 patients (19%) were finally not initiated on ECMO. We observed an increase of patients cannulated in our PICU over time (Fig. 1). There was no significant difference in mortality between patients transported on ECMO after cannulation in our PICU and those who were transported to be cannulated in a referral ECMO center. The median time between the decision and the cannulation was 3.25 h and the median time taken in charge by PICU transport team was approximately 1 h. These periods were not significantly different between cannulation on site or in an ECMO center and between survivors and not-survivors.

Conclusion: In our study, multiple organ dysfunction, particularly hematologic and acute renal failures, seems to be a risk factor of mortality. The delay between decision and management is similar whatever the cannulation site. Specific ECMO mobile team and PICU transport team seem to be essential, fast and trained to transfer these patients. It would be interesting to compare our cohort with children requiring ECMO already hospitalized in a referral ECMO center.

Compliance with ethics regulations: Yes.

Fig. 1

Number of patients transferred to an ECMO center over time

E-posters

P-001 The Use of ICU Resources in Metastatic Breast Cancer

Guillaume Beinse¹, Jean-Jacques Tudesq¹, Luis Teixeira², Edith Borcoman¹, Adrien Mirouse¹, Yannick Hourmant¹, Asma Mabrouki¹, Michaël Darmon¹, Virginie Lemiale¹, Elie Azoulay¹

¹Medical ICU, Saint-Louis Hospital, AP-HP, Paris, France; ²Breast diseases unit, Saint-Louis Hospital, AP-HP, Paris, France

Correspondence: Guillaume Beinse (guillaume.beinse@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):P-001

Rationale: Life expectancy in patients with metastatic breast cancer (MBC) has substantially improved over the last decade. Life threatening complications result from advanced diseases, infection and treatment-related toxicity. Only few studies have assessed outcomes in this setting. We performed a hospital-wide study to investigate how ICU resources are needed in patients with MBC.

Patients and methods: All patients with MBC managed at our hospital between 2010 and 2019 were retrospectively included. The primary outcome was overall survival (OS). Factors associated with ICU mortality were identified using a multivariable Cox proportional hazard model with sensitivity analysis. Results are expressed as median [interquartile ranges] unless stated otherwise.

Results: Among the 1128 patients managed at our hospital, 68 (6%, including 1 male) were admitted to the ICU (8 [2–15] patients per year). Age was 55 [49–67] years. Patients were receiving their 2nd [1st–3rd] line of treatment and had 3 [2–3] metastatic sites. SOFA score at admission was 3 [1–8]. Main reason for ICU admission was sepsis (n = 23, 34%), acute respiratory failure (n = 22, 32%), coma (n = 9, 13%) and metabolic disorder (n = 7, 10%). Invasive mechanical ventilation was required for 18 patients (26%) and renal replacement therapy for 10 (15%). Sixteen (24%) patients died in ICU. Following ICU discharge, median OS was 6.4 months (95% CI [1.7–17.9]) and 22/52 (42.3%) patients died within 3 months. An antineoplastic treatment was resumed for 33/52 (62%) patients alive after ICU discharge. Factors independently associated with mortality were performance status ≥ 2 (HR 1.85, 95% IC [1.01–3.40]) and SOFA score at day 1 (HR 1.19 per point, 95% IC [1.11–1.27]). After sensitivity analysis, the number of treatment lines at ICU admission was not associated with mortality.

Conclusion: ICU admission is required in the course of the MBC disease for 6% of the patients. Determinants of short term outcomes rely on performance status and disease severity but not on the characteristics of the underlying disease. Ongoing analyses will assess whether ICU survivors reach life expectancy of patients never admitted to the ICU.

Compliance with ethics regulations: Yes.

P-002 Factors associated with survival of patients with solid cancer alive after intensive care unit (ICU) discharge between 2005 and 2013

Hubert Gheerbrant¹, Jean-François Timsit², Nicolas Terzi¹, Stephane Ruckly³, Mathieu Laramas¹, Matteo Giaj Levra¹, Emmanuelle Jacquet¹, Loic Falque¹, Denis Moro-Sibilot¹, Anne-Claire Toffart¹

¹CHU Grenoble Alpes, Grenoble, France; ²APHP, Paris, France; ³OutcomeRea, Bobigny, France

Correspondence: Hubert Gheerbrant (hgheerbrant@chu-grenoble.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):P-002

Rationale: The prognosis of critically ill cancer patients admitted in intensive care unit (ICU), remains an issue. Our objective was to assess the factors associated with 3- and 6-month survival of ICU cancer survivors.

Patients and methods: Based on the French OutcomeRea™ database, we included solid cancer patients discharged alive, between December 2005 and November 2013, from the medical ICU of the university hospital in Grenoble, France. Patient characteristics and outcome at 3 and 6 months following ICU discharge were extracted from available database.

Results: Of the 361 cancer patients with unscheduled admissions, 253 (70%) were discharged alive from ICU. The main primary cancer sites were digestive (31%) and thoracic (26%). The 3- and 6-month mortality rates were 33% and 41%, respectively. Factors independently associated with 6-month mortality included ECOG performance status (ECOG-PS) of 3–4 (OR, 3.71; 95% IC: 1.67–8.23), metastatic disease (OR, 2.24; 95% IC: 1.22–4.09), admission for cancer progression (OR, 2.64; 95% IC: 1.32–5.30), SAPS II of 45 to 58 (OR, 4.41; 95% IC: 1.85–10.53), and treatment limitation decision at ICU admission (OR, 3.89; 95% IC: 1.60–9.43). Interestingly, cancer chemotherapy prior to ICU admission was independently associated with lower 3-month mortality (OR, 0.38; 95% IC: 0.2–0.75). Among patients with an ECOG-PS 0–1 at admission, 70% (n = 66) and 61% (n = 57) displayed an ECOG-PS 0–2 at 3 and 6 months, respectively. At 3 months, 74 (55%) patients received anticancer treatment, 13 (8%) were given exclusive palliative care.

Discussion: Factors associated with 6-month mortality are almost the same as those known to be associated with ICU mortality. We highlighted that most patients recovered an ECOG-PS of 0–2 at 3 and 6 months, in particular those with a good ECOG-PS at ICU admission, and could benefit from an anticancer treatment following ICU discharge.

Conclusion: These results should be taken into account when deciding upon ICU admission. It is of paramount importance to have an evaluation of both patient’s general condition and anticancer treatment opportunities following ICU discharge.

Compliance with ethics regulations: Yes.

Fig. 1

ECOG-PS at 3 and 6 months according to ECOG-PS at ICU admission

P-003 Initiation of chemotherapy within ICU: a retrospective study

Maxence Chicoisneau, Lieveke Ameye, Marianne Paesmans, Jean-Paul Sculier, Anne-Pascale Meert

Institut Jules Bordet, Brussels, Belgium

Correspondence: Maxence Chicoisneau (Maxence.Chicoisneau@ulb.ac.be)

Ann. Intensive Care 2020, 10 (Suppl 1):P-003

Rationale: The decision to urgently initiate medical anti-cancer treatment in cancer patients admitted to intensive care unit for cancer-related organ failure is an issue. We currently lack criteria to select patients who may benefit from the treatment initiation. The purpose of our exploratory study was therefore to evaluate the characteristics of patients whose medical anti-cancer treatment is initiated in ICU and to identify prognostic factors for in-hospital mortality. in these patients.

Patients and methods: We analyzed retrospectively, over a period of 11 years (1/1/2007 to 31/12/2017), cancer patients over 18-year old admitted to our ICU Bordet and in whose anti-cancer medical treatment was initiated during in-ICU stay. To identify prognostic factors for in-hospital mortality, we carried out a multivariate analysis of the factors influencing this mortality, considered as a binary. We also analyzed the long term survival of patients alive after their hospital stay (from the day of going out of hospital).

Results: Overall, 147 patients were included, 78 men (53%) and 69 women (47%), with a median age of 58 years (18–86). Of these, 79 patients (54%) had a solid tumor and 68 (46%) had a hematological tumor. In-ICU mortality is 23% (95% CI 17–31%) and in-hospital mortality 32% (95% CI 25–40%). The prognostic factors for in-hospital mortality were age (mean 61 vs 54 in those who survived), the SOFA score (median 6 vs 2), the SAPS II score (mean 41 vs 31), the Charlson score (mean 8 vs. 6.5), the number of organ failure (mean 2.5 vs 1.4) and the presence of a therapeutic limitation (NTBR stated within 24 h: 68% vs 16%). Survival at 1 year of patients who survived the hospital stay was 37% and median survival time was estimated to be 0.5 year (95% CI 0.4–0.9). In patients with a solid tumor, 1-year survival was 17% and 61% in those with a hematological tumor (p < 0.001).

Conclusion: We observed, in selected cancer patients admitted to the ICU for a cancer-related complication, that the initiation of an anti-cancer medical treatment is feasible and can lead to interesting results, particularly in patients with a hematological tumor.

Compliance with ethics regulations: Yes.

P-004 Adequacy and prognosis of ICU admission or denial in critically ill onco-hematology patients

Wafa Zarrougui¹, Emna Ennouri¹, Tarek Ben Ahmed², Khaoula Meddeb¹, Radhouane Toumi¹, Amal Baccari¹, Imen Ben Saida¹, Mohamed Boussarsar¹

¹Medical Intensive Care Unit, Farhat Hached University Hospital. Université de Sousse, Faculté de médecine de Sousse, LR No LR12SP09. Heart Failure, Sousse, Tunisia; ²Department of medical oncology, Farhat Hached University Hospital, Sousse, Tunisia

Correspondence: Wafa Zarrougui (wafa.zarrougui91@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):P-004

Rationale: Considerable progress in the management of onco-hematology (OH) malignancies led to an increase in the number of patients proposed for Intensive Care Unit (ICU) admission. Several guidelines offer decision models for ICU transfer of these patients. We aimed to describe prognosis, adequacy of ICU admission and denial in onco-hematology patients.

Patients and methods: We included all OH patients proposed for ICU admission in a Tunisian medical ICU, between January 2016 and July 2019. From an admission proposal registry, were collected patient underlying condition, functional status, malignancy and predicted prognosis, acute critical illness and its reversibility, adequacy of ICU admission approval or denial according to the 2016 AFSOS guidelines, and patients outcomes.

Results: During the study period, 173 patients were proposed for ICU admission, only 8 (4.6%) were admitted. They were 96 (55%) male; 44.6 ± 15.6 mean aged; 3.6 ± 2.3, mean Charlson Index; 49 (28.3%) had World Health Organisation (WHO) score ≥ 3. Underlying Malignancy were; 23 (13.3%) non metastatic cancer; 43 (24.9%), metastatic cancer; 103 (59.5%), haematological malignancy. Predicted prognosis was estimated at under 3 months for 68 (39.3%) of patients. Acute critical illness were mainly acute circulatory failure, 66 (38.2%) and acute respiratory failure, 52 (30.1%) with 99 (57.2%) considered reversible. Mortality was 7 (87.5%) for admitted and 83 (50.3%) for denied patients. Decision of ICU admission was adequate with guidelines in 5 (62.5%) and 139 (84.2%) in denials.

Conclusion: Physicians’ clinical judgement regarding ICU admission of OH patients seems satisfying. However, a subset of patients could benefit from ICU but are denied admission due to ICU beds unavailability and misuse.

Compliance with ethics regulations: Yes.

P-005 Characteristics and Outcomes of Critically Ill Cancer Patients admitted to a Tunisian Intensive Care Unit

Imen Ben Saida¹, Hela Kallal¹, Yosri Ben Ali¹, Nesrine Fraj¹, Wafa Zarrougui¹, Imtinene Belaid², Mohamed Boussarsar¹

¹Medical intensive care unit, Farhat Hached hospital, Université de Sousse, Faculté de Médecine de Sousse, UR No LR12SP09. Heart Failure, Sousse, Tunisia; ²Oncology department, Farhat Hached hospital, Université de Sousse, Faculté de Médecine de Sousse, Sousse, Tunisia

Correspondence: Imen Ben Saida (imen.bensaida@yahoo.com)

Ann. Intensive Care 2020, 10 (Suppl 1):P-005

Rationale: Cancer patients frequently need intensive care support for a life-threatening condition due to the underlying neoplasm or an adverse therapy-related event. However, there are poor data on their characteristics and outcomes in the intensive care setting. The aim of the present study was to describe clinical characteristics and to identify factors associated with in-ICU mortality in critically ill cancer patients.

Patients and methods: It is a retrospective study conducted in the medical ICU of Farhat Hached teaching hospital between January 2007 and December 2018. All cancer patients with complete records were included. Baseline characteristics, clinical parameters, severity of illness, primary tumor location and outcomes were collected. Univariate and multivariate regression analyses were carried out to identify factors independently associated to poor prognosis.

Results: During the study period, 3569 patients were admitted, 59 (1.65%) had malignancy. Among these, 27 (45.8%) had hematological malignancies and 32 (54.2%) had solid tumors, of whom 15 (25.4%) had evidence of metastases. Clinical characteristics were: mean age, 55.3 ± 14.7 years; male, 38 (64.4%); WHO Performance status 0 to 1, 55 (93.2%); median SAPSII, 53 [41–67]; invasive mechanical ventilation (IMV), 45 (76.6%); median duration of IMV, 2 [1–5]days; vasopressors use, 43 (72.9%). The main reasons for admission were: septic shock, 21 (35.6%); coma, 11 (16.9%); pneumonia 9 (15.3%); pulmonary edema, 8 (13.6%) and miscellaneous 11 (18.6%). Median length of ICU stay was 3 [2–8] days and mortality rate was 67.8%. On univariate analysis, the factors associated with mortality were, IMV on admission (26.3% vs 80%; p < 0.001); vasopressors use (26.3% vs 95%;p < 0.001) and septic shock (45% vs 15.8%; p = 0.029). Multivariate regression model identified two factors as independently associated to mortality: IMV on admission (OR, 9.4; 95% IC, [1.6–55.1]; p = 0.013) and vasopressors use (OR, 46.5; 95% IC, [6.5–330.3]; p < 0.001).

Conclusion: In the present study, invasive mechanical ventilation and vasopressors use on ICU admission were the independent predictive factors of mortality in critically ill cancer patients.

Compliance with ethics regulations: Yes.

P-006 Prognosis and mortality risk factors of denied onco-hematology patients for ICU admission

Emna Ennouri¹, Wafa Zarrougui¹, Tarek Ben Ahmed², Khaoula Meddeb¹, Radhouane Toumi¹, Amal Triki¹, Imen Ben Saida¹, Mohamed Boussarsar¹

¹Medical Intensive Care Unit, Farhat Hached University Hospital. Université de Sousse, Faculté de médecine de Sousse, LR No LR12SP09. Heart Failure, Sousse, Tunisia; ²Department of medical oncology, Farhat Hached University Hospital., Sousse, Tunisia

Correspondence: Emna Ennouri (m.na.ennouri@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):P-006

Rationale: Thanks to therapeutic advances, prognosis of patients with malignancies significantly improved during the last decades. However, occurrence of life-threatening condition could be a turning point in those patients evolution, especially when ICU facilities are not available.

We aimed to determine prognosis and risk factors of mortality of denied onco-hematology patients requiring ICU admission.

Patients and methods: A retrospective study was conducted in a Tunisian medical ICU between January 2016 and July 2019. Were included all consecutive denied onco-hematology patients proposed for ICU admission. Patients baseline characteristics, underlying malignancy and 3 months prognosis, severity of illness and its reversibility and outcome were collected. Univariate and multivariate regression analysis were used to identify factors independently associated with mortality.

Results: During the study period, 165 onco-hematology patients were denied to ICU admission out of 173 proposed patients. Mean age, 45.1 ± 15.6 years; sex ratio 1; mean Charlson Index, 3.6 ± 2.4; median [IQR] World Health Organisation (WHO) score, 2 [0–2]. Underlying malignancies were: non metastatic cancer, 21 (12.5%); metastatic cancer, 42 (25.5%) and haematological malignancies, 98 (59.4%). Mortality rate was 83 (50.3%). Univariate analysis yielded the following as factors associated with mortality: 3-month malignancy prognosis (p = 0.000), predicted reversibility of critical illness (p = 0.000), WHO score ≥ 2 (p = 0.005), acute neurological impairment (p = 0.006) and pulmonary embolism (p = 0.008). Multivariate regression model identified 3 factors to be independently associated with mortality: reversibility of critical illness (OR, 7.4; 95% IC, [2.5–22.9]; p = 0.000); WHO score ≥ 2 (OR, 0.44; 95% IC, [0.19–0.99]; p = 0.048) and acute neurological impairment (OR, 0.3; 95% IC, [0.1–0.9]; p = 0.035).

Conclusion: The present study among onco-hematological patients denied for ICU admission, identified, underlying condition, neurological impairment and predicted reversibility of critical illness, as factors associated with short term in-hospital mortality.

Compliance with ethics regulations: Yes.

P-007 Influence of underlying hematological malignancy grade on outcome of critically ill hematological patients

Pablo Pagliarani, Vincent Rebiere, Lara Zafrani, Virginie Lemiale, Djamel Mokart, Alexandre Demoule, Martine Nyunga, Achille Kouatchet, Frédéric Pene, Michaël Darmon, Elie Azoulay

APHP, Paris, France

Correspondence: Pablo Pagliarani (pablopagliarani@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):P-007

Rationale: Prognostic impact of underlying malignancy seems limited in most studies assessing outcome of critically ill cancer patients [1]. However, only limited number of characteristics, namely disease progression status and preexisting stem cell transplantation, were usually assessed [1]. Primary objective of this study was to assess influence of hematological malignancy aggressiveness on hospital outcome. Secondary objective was to assess influence hematological malignancy aggressiveness on type of infection.

Patients and methods: Post-hoc analysis of prospective multicenter cohort performed in 17 hospitals in France and Belgium and including critically ill adults with underlying hematological malignancy admitted in ICU from Jan 2010 to May 2011. A Cox model was used to adjust for confounding variables then a propensity score matching on characteristics associated with underlying malignancy aggressiveness was performed.

Results: Of the 1011 included patients, 300 (29.7%) had low grade malignancy (LG), the most frequent being myeloma (n = 126), chronic lymphocytic leukemia (n = 76), and myelodysplasia (n = 46). Patients with LG malignancy were older, underwent more frequently autologous stem cell transplantation (SCT) and had less frequently altered performans status. They had more severe organ failure at ICU admission (SOFA score 6 [4–9] vs. 5 [3–8], P = 0.03). Before adjustment, mortality was 36% (n = 108) and 39.9% (n = 284) respectively in patients with and without LG malignancy (P = 0.27). After adjustment for confounder using a Cox model, a higher mortality was associated with non-low grade malignancy (OR 1.49; 95% IC 1.17–1.9). A propensity score then allowed a 1:1 matching upon variable associated with malignancy aggressiveness. After matching unadjusted mortality was 36% (n = 108) in patients with LG malignancy and 48.8% (n = 146) in patients with high grade malignancy (P = 0.002) (Figure). In the matched cohort and after adjustment for confounder, high grade malignancies were associated with lower mortality (OR 1.42; 95% IC 1.10–1.84). Risk of fungal infection was unchanged by underlying malignancy before adjustment (6% vs. 8.6% of patients with and without LG malignancy; P = 0.26) or after adjustment (HR 1.29; 95% IC 0.71–2.41).

Conclusion: Despite anti-cancer advances, aggressiveness of hematological malignancies is associated with overall ICU outcome. Low-grade malignancies displaying a better prognosis than non-low grade. Aggressiveness of the underlying malignancy is not associated with risk of fungal infection.

Compliance with ethics regulations: Yes.

Fig. 1

Survival probability according to underlying malignancy after propensity score matching

P-008 Pulmonary embolism in ICU of the CHU of ORAN: retrospective study about 22 cases

Soulef Bousbia, Soumia Benbernou, Nabil Ghomari, Houria Djebli

Faculté de medecine d’Oran, Oran, ALGERIA

Correspondence: Soulef Bousbia (bsoulef90@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):P-008

Rationale: The pulmonary embolism (PE) remains a serious condition that requires intensive care unit (ICU) admission.

Patients and methods: We aimed to determine the clinical, therapeutic and features scalable of patients with PE within the Department of resuscitation of the CHU of ORAN. It is a retrospective study including patients admitted to the CHU of ORAN from January 2014 to December 2017 with PE. Several parameters have been collected including age, gender, reason for admission, symptomatology, PE risk factors, the Wells score, therapy and evolution.

Results: We collected 22 patients suffering of EP among 1038 patients admitted to the ICU. The median age was 64 years, 75% of the patients were female, the main symptoms were dyspnea in more than 50% of the patients, onset of symptomatology was brutal in 87.5%. Physical signs were dominated by tachycardia in all patients, blood gases highlighted hypoxia in 60% of the patients. D. dimeres plasma levels were increased in all patients. Angioscanner found proximal thrombus in 75% of patients. All the patients received curative dose HBPM and finally mortality was around 50%.

Conclusion: Tachycardia and polypnea are the main clinical characteristics of patients with PE in the emergency ward. Beside tools for PE diagnosis and risk stratification may be helpful in patient management.

Compliance with ethics regulations: Yes.

P-009 Pulmonary embolism feature in patients with Guilliain Barre syndrom

Olfa Turki, Rezk Gorbel, Mariem Dlela, Najeh Baccouche, Abir Bouattour, Mabrouk Bahloul, Mounir Bouaziz

HABIB BOURGUIBA University Hospital, Sfax, Tunisia

Correspondence: Olfa Turki (olfa.turki.rea@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):P-009

Rationale: Guillain-Barré syndrome is the most common cause of acute flaccid paralysis and is associated with pulmonary embolism due to the mobility limitation. The aim of this study is to describe the incidence, the severity of pulmonory embolism in patients admitted to an intensive care unit (ICU) for Guillain–Barre syndrome (GBS).

Patients and methods: Twenty-eight adults patients with confirmed diagnosis of GBS were admitted to the ICU in our university hospital center over a 10-year period and they were all included. Prevalence, risk factors and course of VTE were analyzed in ICU patients with various forms and severity of GBS.

Results: During the study period, 23 adult GBS patients were included. Five (17.9%) developped pulmonary embolism. The mean age was 51.2 ± 16.7 years and the sex ratio was 0.86. The comparaison betewen the 2 groups with and without PE showed that factors associated with the development of this complication were: respiratory failure requiring mecanical ventilation (p = 0.03), infectious complications (p < 0.001), blood pressure lability (p = 0.029), the delay of ICU admission (p = 0.02), the delay to treatment initiation (p = 0.036), the SOFA score (p = 0.03) and the presence of quadriplegia (p = 0.031).

Conclusion: Pulmonary embolism is a frequent complication in patients with GBS. Factors associated with this complication were: respiratory failure requiring mecanical ventilation, infectious complications, the delay of ICU admission, the delay to treatment initiation, a high SOFA score and the presence of quadriplegia. Preventive measures in this category of patients have to be improved.

Compliance with ethics regulations: Yes.

P-010 Acute respiratory distress syndrome among burns in Tunisia: state of play and prognosis

Wael Chemli, Lilya Debbiche, Hana Fredj, Sarra Ben Zarrouk, Hana Benali, Manel Ben Saad, Amel Mokline, Amenallah Messaadi

Burns intensive care unit, CTGB, Ben Arous, Tunisia

Correspondence: Wael Chemli (wchemli90@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):P-010

Rationale: Acute respiratory distress syndrome (ARDS) is a life-threatening pathology associated with very high morbidity and mortality (35–45%) in intensive care units (ICU) and with even higher mortality among the severly burned patients worldwide (36 à 80%). The aim of our study was to describe in Tunisia burn patients with ARDS and to identify prognosis factors.

Patients and methods: We conducted a descriptive retrospective study between 01-01-2017 to 31-12-2018, in burns ICU, in Ben Arous, in Tunisia. All burns who presented an ARDS, according to the Berlin 2012 definition, during their stay in the ICU, were included. When clinical or gasometric data was uncomplete, these patients were excluded.

Results: During the study period, 691 patients were admitted to our burn unit including 246 ventilated patients. Fifty patients presented an ARDS: fifteen patients were excluded for lack of information, and 35 patients were retained. The sex ratio was 2.5. Patients had a mean age of 36 ± 12 years, an average burned area of 44% ± 22%, an average unit of burn skin score (UBS score) of 94 ± 77 and an average sequential organ failure assessment score (SOFA score) of 4. None of the patients had a history of cardiovascular or pulmonary diseases. The average time of onset of ARDS was 5 ± 4 days. ARDS was mild in 1 case, moderate in 11 and severe in 23. The etiology of ARDS was pulmonary in 25 cases (71%) and extra-pulmonary in 10 (29%). The pulmonary ARDS had as cause pneumonia isolated in 15 patients, an isolated pulmonary burn in 6 patients and a combination of pneumonia and lung burns in 4 patients. Extra-pulmonary ARDS were all due to sepsis and mainly to bacteremia. Septic shock was associated with ARDS in 20 patients (57%). The treatment was a conventional treatment based on protective ventilation, curarization and prone positioning in addition to the etiological treatment. The average length of stay in ICU was 9 days and mortality was 85% in these patients.

Conclusion: Mortality from ARDS in burns in Tunisia, is important especially in those with pulmonary burns as well as those with sepsis. The introduction of new treatments, such as extracorporeal membrane oxygenation, remains essential to improve the prognosis of burn patients.

Compliance with ethics regulations: Yes.

P-011 Risk factors and outcomes of aspiration pneumonia

Oussama Jaoued, Makni Saba, Abid Emna, Sik Ali Habiba, Fekih Hassen Mohamed, Elatrous Souheil

Hôpital Taher sfar, Mahdia, Tunisia

Correspondence: Oussama Jaoued (oussamajaoued@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):P-011

Rationale: Aspiration pneumonia (AP) is common in intensive care unit (ICU). The incidence of AP among adults hospitalized with pneumonia ranges between 5 and 53.2%. Usually one or more risk factors are identified to be involved in AP. The aim of this study was to determine the risk factors and predictors of mortality on patients with AP.

Patients and methods: We retrospectively included patients aged more than 18 years and who were hospitalized in our ICU for AP. Patients were excluded if they had history of tuberculosis, if they have bronchiectasis or metastatic brain tumor.

Results: A total of 102 patients were included. History of diabetes, hypertension, epilepsy and ischemic stroke were found respectively in 22.2%, 21.5%, 16.7%, and 7.8% of cases. The reason of ICU admission were coma (35%), acute respiratory failure (33%), poisoning (27%) and cardiac arrest (5%). The incidence of acute respiratory distress syndrome (ARDS) was 17%. The most common organism isolated was staphylococcus aureus (4 cases). Risk factors for AP were epilepsy (20%), swallowing disorders (18%), ischemic stroke (12%), COPD (9%) and degenerative neurological disease (5%). The mortality rate was 17.6%. The median duration of mechanical ventilation was 12 days [IQR 10–23]. In multivariate logistic regression analysis; SAPS II score (OR = 1.05, 95% IC [1.001–1.1], p = 0.046) and ARDS (OR = 44.04, 95% IC [3.91–495.57], p = 0.002) were independently associated with mortality.

Conclusion: Risk factors for aspiration pneumonia were epilepsy, swallowing disorders and ischemic stroke. ARDS and SAPS II score were independent predictive factors of mortality.

Compliance with ethics regulations: Yes.

P-012 Hyperoxia in intensive care: impact on morbidity and mortality

Rim Jemmeli, Samia Ayed, Amira Jamoussi, Dhouha Lakhdhar, Jalila Ben Khelil, Mohamed Besbes

Abderrahmen Mami Hospital, Ariana, Tunisia

Correspondence: Rim Jemmeli (rimjemmeli@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):P-012

Rationale: Oxygen therapy is a common treatment in intensive care unit (ICU). While its efficiency in hypoxia is well-known, several studies have shown a harmful effect of oxygen when prescribed at high doses. The threshold value defining pathological hyperoxia remains undetermined. The aim of this study was to evaluate the impact of hyperoxia on morbidity and mortality.

Patients and methods: This was a prospective study performed in the ICU of Abderrahmen Mami Hospital during a 4-month period. All patients admitted in ICU during the study-period were included. Those who didn’t need oxygen therapy or in end of life stage were excluded. Arterial blood gases were analyzed daily and each day with at least one value of oxygen arterial saturation (SaO2) > 92% was considered as a day with hyperoxia. For each patient included, the number of times and days spent in hyperoxia was recorded as well as complications during the ICU stay and the outcome.

Results: During the study-period, 140 patients were included but only 112 were eligible. Mean age was 58 ± 18 years. Acute on chronic respiratory failure was the most frequent reason of admission (67%). Non-invasive ventilation was required for 18% of patients and invasive mechanical ventilation was necessary in 65% of cases. Overall mortality was 32%. Hyperoxia was observed in 96% of cases, with an average of 10 ± 10 times during the ICU stay and 6 ± 6 days. A statistically significant association was observed between a long duration of hyperoxia and the occurrence of ventilator acquired pneumonia (p < 10–3), ventilator acquired bronchitis (p = 0.001), acute respiratory distress syndrome (p < 10–3), atelectasis (p < 10–3), septic shock (p < 10–3), rythm disorders (p = 0.003), reintubation (p < 10–3) and tracheostomy (p = 0.038). On multivariate analysis, independent factors of mortality were: simplified acute physiology score II, cardiac failure, need for invasive mechanical ventilation and septic shock. Hyperoxia was not independently associated with mortality.

Conclusion: Hyperoxia is frequent in ICU. It is significantly associated with ICU complications but not independently associated with mortality.

Compliance with ethics regulations: Yes.

P-013 Experience of the practice of prone position in patients with acute respiratory distress syndrome in intensive care (CHU Oran)

Nabil Ghomari, Soumia Benbernou, Djebli Houria

Faculté de medecine d’Oran, Oran, Algeria

Correspondence: Nabil Ghomari (nabilghomari@hotmail.fr)

Ann. Intensive Care 2020, 10 (Suppl 1):P-013

Rationale: Mechanical ventilation (MV) in the prone position (PP) and low tidal volume have become recommendations with a high level of scientific evidence in recent years. The PP has been practiced for 7 years in the CHU Oran emergency resuscitation service. We wanted to report the service experience in the practice of PP in patients with ARDS.

Patients and methods: Retrospective study performed in patients with severe hypoxia ARDS with SPO2 < 88% under FIO2 > 80% or PAO2/FIO2 < 150 during the period March 2011 to December 2018.

Results: 38 patients received ventilation in PP. ARDS was secondary to thoracic trauma in 42% of patients, septic shock in 32% and aspiration pneumonitis in 26%. Analysis of the success factors and improvement of oxygenation found that lobar ARDS, the delay < 72 h and a duration of PP ≥ 18 h were statistically significant.

Conclusion: The PP must be integrated into the arsenal of care of the patients in ARDS especially in our country where we do not have all the therapeutic options.

Compliance with ethics regulations: Yes.

P-014 Impact of Cannula Size on Clinical Outcomes in Venovenous Extracorporeal Membrane Oxygenation

Julien Goutay, Nicolas Cousin, Thibault Duburcq, Erika Parmentier-Decrucq

CHU de Lille, Pôle de Réanimation, Hôpital Salengro, Lille, France

Correspondence: Julien Goutay (julien.goutay@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):P-014

Rationale: In Veno-Venous Extracorporeal Membrane Oxygenation (VV-ECMO) therapy, blood flow is the main determinant of arterial oxygenation and should be 60–80 mL/kg/min in adults. This flow rate is determined by several factors including the size of the inflow cannula. The impact on clinical outcomes of arterial cannula’s size in Veno-Arterial ECMO (VA-ECMO) has already been studied, and showed no difference for survival to discharge, weaning success rate and initial flow rate between a small cannula group and a larger one. Our first objective was to describe the impact of inlet cannula size on the assistance flow rate in patients treated with VV-ECMO. Secondary objectives were to analyze its impact on ECMO weaning, mechanical ventilation characteristics and mortality.

Patients and methods: We retrospectively reviewed all cases of respiratory failure treated with VV-ECMO admitted in the medical intensive care unit (ICU) of Lille’s teaching hospital from January 1st, 2013 through March 31st, 2019. Inlet cannula size was collected and divided into two groups: the “small cannula” group had inlet cannula less than or equal to 23Fr, while “large cannula” were larger than 23Fr. Primary endpoint was the initial flow rate according to the inlet cannula size, and its changes during the first 48 h of assistance. Secondary endpoints were the analysis of predictive factors associated with the choice of a larger inlet cannula, and the impact of its size on clinical outcomes such as successful ECMO weaning.

Results: 74 patients treated with VV-ECMO were admitted in our hospital. Eleven (15%) were cannulated with a large inlet device. Mean initial ECMO flow rate was statistically higher in the “large cannula” group than in the “small cannula” one: 5.8 L/min (± 0.7) versus 4.7 (± 0.8) respectively, p < 0.0001. The difference was also significant during the first 48 h of assistance. We found no difference between the two groups on clinical outcomes such as ECMO weaning time. In univariate analysis, weight was heavier in the “large cannula” group [94 (± 26) kg] than “small cannula” [81 (± 20)], p < 0.05.

Conclusion: ECMO initial flow rate was higher in a “large inlet cannula” group (internal diameter more than 23 Fr) compared with a “small cannula” group. We found no correlation with cannula-related haemorrhagic or thrombotic complications. Inlet cannula size did not influence ECMO weaning, and duration time, but this may be a lack of statistical power. Further prospective studies should confirm this results.

Compliance with ethics regulations: Yes.

Fig. 1

Difference and evolution of maximum ECMO flow rate in patients with small (n = 60) and large inlet cannula (n = 10)

P-015 Colonization with carbapenemase-producing Gram-negative bacilli in burn patients in Tunisia

Sonia Ben Behi¹, Sarra Dhraief¹, Hana Fredj², Lilya Debbiche², Amenallah Messaadi², Lamia Thabet¹

¹Centre de Traumatologie et des Grands Brûlés de Ben Arous-Laboratoire de biologie médicale et banque du sang, Ben Arous, Tunisia; ²Centre de Traumatologie et des Grands Brûlés de Ben Arous-Service de réanimation des brûlés, Ben Arous, Tunisia

Correspondence: Sonia Ben Behi (benbehi.sonya@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):P-015

Rationale: Burn patients are at risk of multidrug-resistant (MDR) bacterial infections with high mortality rate. Therefore, monitoring the emergence of MDR pathogens in these vulnerable patients is important. This study aimed to assess digestive colonization with carbapenemase-producing Gram-negative bacilli (CP-GNB) in patients admitted to the burn intensive care unit.

Patients and methods: Our study was prospective and conducted over a one-year period (January 2018 to December 2018). Every admitted patient was subjected to the screening. A double swab set was used to collect rectal swab specimens. One swab was used for MDR screening by disk diffusion method on selective media; the other for multiplex real-time PCR (Cepheid’s GeneXpert^®) allowing detection of the most common carbapenemase-encoding genes (CEG) (blaOXA-48, blaKPC, blaNDM, blaVIM and blaIMP).

Results: Among the 125 studied patients, 71 (56.8%) were detected positive at admission for CP-GNB by the GeneXpert^® Carba-R assay. Eleven patients, initially not colonized, acquired positive faecal carriage subsequently during their hospital stay. Forty-two colonized patients (59.1%) developed CP-GNB infection during their hospitalization. The CEG blaNDM quantitatively dominated by far with 54 detections; either alone (15 cases) or associated with other CEG (39 cases). The second most frequent gene was blaOXA-48. It was detected alone eight times and in association with other CEG 38 times. Forty-three patients carried blaVIM gene, usually in association with other CEG (93%). However, only one patient carried blaKPC gene. The parallel screening by classical microbiology methods (disk diffusion on selective media) detected the presence of CP-GNB in all molecular positive samples.

Conclusion: Our study describes the characterization of carbapenemase in burn patients and highlights their alarming spread. This emphasizes the importance of an active surveillance program by early detection of CP-GNB carriers and an isolation policy to limit the MDR infections expansion.

Compliance with ethics regulations: Yes.

P-016 Evaluation of Antifungal Therapy in Burn ICU

Selma Abid, Hana Fredj, Amel Mokline, Bahija Gasri, Manel Ben Saad, Amenallah Messaadi

Intensive Burn Care Departement, Burn and Trauma Center-Tunis, Tunisia, Ben Arous, Tunisia

Correspondence: Selma Abid (darly2910@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):P-016

Rationale: Invasive fungal infections are increasingly observed in the ICUs especially in burn units. In the absence of simple and accessible techniques for early microbiological diagnosis, the use of antifungal treatment is increasing. Little is known about the extent of the problem of antifungal prescription in burn ICUs. We aimed to evaluate the antifungal prescription in major burn patients.

Patients and methods: During the study period (2018–2019), all prescriptions of antifungals were analysed. Analysis concerned demographics, clinical circumstances, as well as the basis of antifungal prescribing (targeted vs. empiric).

Results: Among the 691 patients admitted in this period, 58 patients were treated with antifungals (sex ratio: 1.4; mean age: 38 ± 19 years, with low associated comorbidity). The TBSA was 34.25% [19.5–53.25], UBS was 57 [32.75–91]. Most of the patients (87.9%) were transferred from another hospital structure within 62 ± 96 h. Antifungal treatment was started at the average of the seventh day post wound injury, more often on an empiric basis. SOFA score at the beginning of the treatment was 11 ± 3.9. Lymphopenia was present in 32% and thrombopenia was present in 41%. Index of colonisation was positif only in 4 cases. The average candida score was 2.72 ± 1.15. Microbiological results were tardily collected, within 3 weeks, in 27%. Fungal urine infections were found in 9 cases. Candidemia and catheter-related infections were considered only in 2 cases. The risk factors of fungal infection as described in literature were found in most of the patients including mechanical ventilation (84.4%), length period of stay (16 days [8.75–29.5]), central venous line (100%), severe sepsis or septic shock (100%), large-spectrum antibiotherapy for more than 3 days (100%).

Conclusion: The management of antifungal infections in major burn patients is still challenging. Antifungal prescription is based on clinical presumption. The empirical prescription reflects the lack of efficient laboratory support and late microbiological results prompting physicians to rely on clinical informations. The management of fungal infections is based on the improvement of mycological investigations.

Compliance with ethics regulations: NA.

P-017 Epidemiology of candidemia in a medical ICU: a 12-years retrospective study

Salma Klaii¹, Ahlem Trifi², Aicha Kallel¹, Foued Daly², Yosr Touil², Sami Abdellatif², Kalthoum Kallel¹, Salah Ben Lakhal²

¹Parasitology/Mycology laboratory, la Rabta hospital, Faculty of Medicine of Tunis, Tunis, Tunisia; ²Medical ICU, la Rabta hospital, Faculty of Medicine of Tunis, Tunis, Tunisia

Correspondence: Salma Klaii (kalai.salma89@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):P-017

Rationale: Invasive candidiasis is a widespread and alarming infection in intensive care units (ICU) patients. Its diagnosis is often difficult because of the lack of specificity of clinical signs and the low sensitivity of blood cultures. While the Candida albicans species remain the most common cause of bloodstream infections, non-albicans are emerging. These infections are serious, associated with high mortality rate and requiring early diagnosis and appropriate treatment. In Tunisia, few data are available. We aimed to determine the epidemiological profile of a series of candidemia in ICU, the risk factors associated with the occurrence of candidemia and to describe the modalities of the mycological diagnosis of candidemia and their etiological profile.

Patients and methods: A retrospective longitudinal descriptive study conducted in the Parasitology—Mycology laboratory with the collaboration of the medical ICU of la Rabta hospital—Tunis over a 12-year period from January 1, 2006 to December 31, 2017. All hospitalized ICU patients with at least one Candida-positive blood culture were included.

Results: Forty-three patients among 2585 hospitalized patients during the study period had at least one candidemia infection. The main risk factors for development of candidemia infection include invasive procedures, a prior use of antibiotics and parenteral nutrition. C. Albicans was the most common species, detected in 41.9% of patients. Non-albicans Candida species were prominent (58.1%), represented by C Parapsilosis, followed by C. Tropicalis and C. Krusei then C. Glabrata and finally C. Lusitaniae. All the isolates tested were sensitive to the common antifungal agents. The mortality rate of our patients was high (76.7%), and the detection of the Albicans species in blood cultures was the only prognostic factor identified (OR = 9.38 [0.83–105.23], p = 0.06).

Conclusion: Candidemia in the medical ICU patients is common and is associated with high mortality rate. Despite the progress of biological tools, the diagnosis is difficult and needs to take into account the risk factors of the patients as well as scores based on clinical and microbiological parameters. A better identification of risk patients may help to early initiate empirical antifungal treatment.

Compliance with ethics regulations: Yes.

P-018 Necrotizing soft-tissue infections in the intensive care unit: a retrospective hospital-based study

Kais Regaieg, Sabrine Nakaa, Arnaud Mailloux, Madjid Boukari, Johana Cohen, Dany Goldgran-Toledano

Groupe hospitalier intercommunal Le Raincy-Montfermeil, Montfermeil, France

Correspondence: Kais Regaieg (kais.regaieg@gmail.com)

Ann. Intensive Care 2020, 10 (Suppl 1):P-018

Rationale: The objective of our study is to describe the epidemiological and clinical characteristics of Necrotizing soft-tissue infections (NSTI) and to improve therapeutic management.

Patients and methods: We conducted a retrospective observational study that included patients admitted in the intensive care unit (ICU) of general hospital between September 2014 and Aout 2019 with a primary or secondary diagnosis of NSTI. We collected demographic and clinical data, cultured pathogens, lengths of stay, and in-ICU mortality.

Results: During the study period, a total of 20 patients admitted to the ICU were diagnosed with NSTI (0.3% of the total number of patients). The mean of age was 69 years. The sex ratio (M/W) was 0.6. Ten patients (50%) were directly admitted to the ICU, others were transferred from medical or surgical wards. The mean of SAPS II was 61.6 (26.6). The main indication to admission in ICU was shock (50%). The most common comorbidity was diabetes (50%). The other co-morbidities associated with NSTI were cardiovascular diseases (33%), obesity (25%) and carelessness (25%). The sites most commonly affected were extremities in 15 patients (75%) and abdomen/ano-genital in 4 patients (20%). In ICU, a total of 17 patients (85%) were mechanically ventilated [ (median duration: 4.70 days (8.63)], 15 patients (75%) were given vasopressors, and 4 patients (20%) underw