Skip to content


  • Meeting abstracts
  • Open Access

Proceedings of Réanimation 2019, the French Intensive Care Society International Congress

Annals of Intensive Care20199:474

  • Published:

Oral communications Oral communications: Physiotherapists

COK-1 Bench assessment of the effect of a collapsible tube on the efficacy of a mechanical insufflation-exsufflation device

Romain Lachal (speaker)

Réanimation médicale, Hôpital de la Croix-Rousse, Hospices Civils de Lyon, Lyon, FRANCE
Correspondence: Romain Lachal -

Annals of Intensive Care 2019, 9(Suppl 1):COK-1

Introduction: Mechanical Insufflation-Exsufflation (MI-E) by using a specific device is commonly used to increase weak cough, as in patients with chronic neuromuscular weakness or in intensive care unit (ICU) patients with ICU-acquired neuro-myopathy. The assessment of the efficacy of MI-E device is commonly done by measuring peak cough flow (PCF). Upper airways collapse is frequently associated with neuromuscular disease and may compromise MI-E efficacy. Tracheomalacia is another disease that may impede PCF to increase with MI-E device. The goal of present study was to carry out a bench study to assess the effect of MI-E on PCF with and without the presence of a collapsible tube. Our hypothesis was that PCF was lower with than without collapsible tube.

Patients and methods: We used a lung simulator (TTL Michigan Instruments) with adjustable compliance (C) and resistance (R) to which a MI-E (CoughAssist E70, Philips-Respironics) was attached, with or without a latex collapsible tube. Flow and pressure were proximal to the lung simulator. Six C-R combinations were tested, each with and without the collapsible tube. For each C-R combination, we set ± 30, ± 40 and ± 50 cmH2O inspiratory expiratory pressure at the MI-E device. MI-E device was set in automatic mode with inspiratory time of 3 s, expiratory time of 3.2 s and pause of 2 s. Each set was recorded by using a data logger (Biopac 150, Biopac inc.) and the last 5 cycles were used for the analysis done by using Acqknowledge software (Biopac inc.). The peak expiratory flow during the first 100 ms after onset of expiration was taken as the surrogate of PCF. The corresponding pressure was also recorded.

Results: Contrary to our hypothesis, the peak expiratory flow during the first 100 ms of exsufflation phase is higher with than without the collapsible tube in every C-R condition, as shown in figure 1. For the C20R5 condition the effect of the collapsible tube on the intercept (− 0.35 cm H2O) was not significant but this was offset by a significant increase in slope (+ 0.12 L s cm H2O). For the other conditions, the collapsible tube significantly increased PCF at 30 cm H2O expiratory pressure and the gap further increased above this pressure because the slope increased with the collapsible tube.

Conclusion: We found that peak expiratory was higher with than without collapsible tube. In vivo measurements in patients should be done to confirm this finding.

COK-2 Early verticalization in neurologic intensive care units with a weight suspension system

Margrit Ascher (speaker), Francisco Miron Duran, Fanny Pradalier, Claire Jourdan, Kevin Chalard, Flora Djanikian, Isabelle Laffont , Pierre-François Perrigault

CHU Montpellier, Montpellier, FRANCE
Correspondence: Margrit Ascher -

Annals of Intensive Care 2019, 9(Suppl 1):COK-2

Introduction: Background. Current literature and French guidelines recommend early mobilization in Intensive Care Units (ICU), including verticalization and walking. Verticalization for neurologic patients in ICU is challenging because of neurological impairments, risks of falls and of clinical worsening. In the neuro-ICU of Montpellier university hospital, a weight suspension system (LiteGait®) is used. Objectives. To study the feasibility and safety of walking with the weight suspension system in a neuroICU. Feasibility involved proportion of patients who benefited from suspension walking, reasons for not using it, physiotherapists’ time required. Safety involved rate of adverse events, changes in vital parameters, pain.

Patients and methods: Design. Monocentric, prospective, descriptive study, including all neurogical patient hospitalized for > 48 h in ICU with initial mechanical ventilation from mid-February to mid-September 2018 (excluding deceased patients). Criteria for using suspension walking where respiratory stability without mechanical ventilation (tracheostomy and or oxygen therapy possible), hemodynamic and neurologic stability, sufficient respond to command (head control, testing of one quadriceps > 3 or two quadriceps > 2). Data included general description of patients + clinical status before suspension walking (pain, MRC testing, sitting balance, RASS, hemodynamic and respiratory parameters, medical equipment) + pain, hemodynamic and respiratory parameters during sessions + description of adverse events and consequences + duration of walking.

Results: Among 83 patients included (see table for characteristics), 25% benefited from suspension walking during their stay + for 25% of patients, suspension walking was needed but hindered by organization difficulties (such as timetable challenges, early discharge from ICU) + 22% patients could walk without suspension on first verticalization + 20% were too impaired to use suspension walking. A total of 41 suspension walking sessions were performed. Five sessions needed to be interrupted for the five following reasons- pain, hypotension, dizziness, diarrhea and dysfunction of the device. No adverse event had clinical consequences beyond the session. Pain score raised significantly for one patient only (5 points in BPS score). Mean delay between extubation (or tracheotomy) and first suspension walking was 7 days. Mean delay between first suspension walking and first walking without suspension was 8 days.

Conclusion: Verticalization with a suspension device in a neurologic ICU is feasible and safe, with a trained and supported team. Obstacles such as medical equipment, language impairments, absence of balance, heavy weight, poor tonicity and participation can be overcome using such a device. Suspension could enable several neurologic ICU patients to walk one week earlier.

COK-3 Effects of active exercise on red blood cell deformability in critically ill patients

Vinciane Scaillet (speaker)1, Mohera Potvin1, Damien Wathelet1, Adrien Fievet2, Ingrid Leclercq2, Karim Zouaoui Boudjeltia3, Patrick Biston 7, Michael Piagnerelli1

1Service de Kinésithérapie, Charleroi, BELGIUM; 2Ecole de Kinésithérapie-HEPH-Condorcet, Charleroi, BELGIUM; 3Laboratoire de Médecine Expérimentale. ULB 222 Unit., Charleroi, BELGIUM; 4Réanimation, Charleroi, Charleroi
Correspondence: Vinciane Scaillet -

Annals of Intensive Care 2019, 9(Suppl 1):COK-3

Introduction: Early mobilization is recommended in critically ill patients to limit the rapid decrease in muscle mass and function. Several methods, in addition of mobilization are used including passive active in-bed cycling and muscle electrostimulation (ESM). This latter could improve the muscle microcirculation assessed by tissue oxygen saturation and red blood cell (RBC) deformability in untrained athletes by a RBC-nitric oxide production mechanism. In a monocentric, randomized trial, we investigated in critically ill patients, the effects of a passive active exercise alone or associated with a daily ESM or with a daily in-bed cycling sessions on RBC deformability.

Patients and methods: Patients were randomized during the first 48 h of ICU admission in three groups- passive and or active mobilization (M) twice a day or M once a day associated with a daily session of 30-min legs passive active in-bed cycling (CYCLE) or M once a day associated with a daily session of 30-min legs ESM (ESM). Demographic data, PaO2 FiO2 and lactate levels were recorded before and after the session. RBC deformability was assessed ex vivo, before and after exercise, by the Laser-assisted Optical Rotational Cell Analyzer (LORCA, Mechatronics Instruments BV, AN Zwaag, Netherlands) with the elongation index (EI) in relation to the shear stress (0.3–50 Pa) applied on the RBC membrane. Data were expressed by the median value 25–75% of delta EI (EI after minus EI before exercise) and compared by Kruskal–Wallis one way test. A p value < 0.05 was considered as statistically significant.

Results: 10 patients in each group were included. PaO2 FiO2 (before- 318 (280–381) versus 310 (246–376) after + p = 0.46) and lactate levels (1.3(0.8–2.0) versus 1.4(1.0–2.0) mmol/L L + p = 0.65) were not modified whatever the type of exercise. Only, in the ESM group, delta EI was significantly decreased for several low shear stress (0.76, 1.21 and 1.93 Pa) compared to the two other groups.

Conclusion: Association of active passive mobilization and ESM alter at least for low shear stress, the RBC deformability in critically ill patients. Effects on long term treatment by this technique on RBC rheology and on muscle metabolism need to be investigated.

COK-4 Neurogenic para-osteoarthropathy in severe brain injury in the emergency department of CHU Oran Emergencies (Algeria)

Youcef Oubadi (speaker)1, Soumia Benbernou2, Khalida Bouyacoub1, Nabil Ghomari1, Abdelkader Azza1, Houria Pr Djebli France2

1Centre Hospitalo Universitaire d’Oran, Oran, ALGERIA; 2Faculté de médecine, Oran, ALGERIA
Correspondence: Youcef Oubadi -

Annals of Intensive Care 2019, 9(Suppl 1):COK-4

Introduction: Neurogenic para-osteoarthropathy (NOPO) is a common complication after severe head trauma, and can be complicated by joint damage ranging from amplitude limitation to ankylosis. This complication is frequently encountered in the unit. Objectives-Determine the frequency of NOPOs in the Emergency Resuscitation Department Establish a prevention protocol established by the physiotherapists of the unit.

Patients and methods: The exposed work was carried out at the URC resuscitation unit of CHUOran. This is a prospective study spread over two years (2015 and 2017) involving 76 patients. The study population consists of severe head injuries (GCT) with a Glasgow score of 8 or less. We studied the time of onset, the affected joint, the degree of limitation + risk factors.

Results: Of the 76 TCG patients, 29 patients presented with NOPO. The onset time was an average of 14 days with extremes ranging from 10 to 20 days. The elbow was the most affected joint with knee involvement in some cases. In this series the risk factors favoring the onset of NOPO were- severity of neurological lesions, periarticular lymphoedema, arterial compression of blood pressure cuffs, duration of sedation, type of drug used.

Conclusion: The NOPO is a complication occurring at the TCG, it causes functional sequelae that will jeopardize its socio-professional reintegration that only prevention will prevent.

COK-5 Standing the ARDS patient

Guillaume Fossat (speaker), Emmanuelle Desmalles

CHR, Orléans, FRANCE
Correspondence: Guillaume Fossat -

Annals of Intensive Care 2019, 9(Suppl 1):COK-5

Introduction: Standing the ICU patient has become a routine care in early mobility programs. In the awake patient, stand can improve minute ventilation, paO2 and awareness. Undesirable effect can be a decreased in Arterial Pressure. To perform stand position in the ICU patient the physiotherapist needs to move him out of the ICU bed. Recently, standing position can be performed directly in the bed. In ARDS patient, positioning is one of the major component of the care. Upright and prone positioning showed many benefits. Prone decreased the mortality rate up to 50% in ARDS patients. Upright position improves the oxygenation effects in ARDS patients. Instead, prone had several adverse effects such as pressure ulcer, endotracheal tube obstruction, and ventilator acquired pneumonia.

Patients and methods: In order to decrease the adverse effect of prone positioning we proposed to evaluate whether standing position is safe and can produced the same oxygenation effects as the prone positioning. The standing is performed directly in a dedicated ICU bed, in severe and moderate ADRS patients according to the BERLIN definition. To prevent hypotension, we use an anti shock pants that can be inflate up to 40 mmHg. Standing is performed on a multiple step angulation program, based on the clinical patient responses during 15 min each step-25° => 40° => 55° => 40° => 25° Arterial blood sample is performed before and at the completion of the technique. Static compliance is calculated at each angulation step.

Results: For now we only performed one Stand in ARDS patient. No adverse event occurred during the technique. The SpO2, the Blood Pressure remain stable and the anti-shock pants was no inflate. The heart rate rise up from 112 BPM to 132 BPM. The static compliance was stable between the begin and the end (25.00 and 26.92). The P F ratio increase from 171 to 211 at the end of the manoeuver.

Discussion: Upright position was already evaluated in some studies, it shows benefits only in responders patient during ARDS. Whereas upright is not standing, because the real incline angle during upright is 26°. With this technique, the patient can be stand for real on his feet, in a posture that is similar as the human stand position.

Conclusion: This technique appears to be safe and can produce positive effects on the P F ratio without change in static compliance during ARDS Volume Controlled Ventilation. Further patients need to test this technique to prove real effects.

COK-6 Aerodigestive tract ultrasound imaging- principles and interests for physsiotherapy

Carlos Diaz Lopez (speaker)1, Aymeric Le Neindre2

1Hôpital Forcilles, Férolles-Attilly, France; 2Hôpital Foch, Suresnes, FRANCE
Correspondence: Carlos Diaz Lopez -

Annals of Intensive Care 2019, 9(Suppl 1):COK-6

Introduction: The clinical assessment of swallowing disorders in ICU has many limitations in its reliability and accuracy. The reference standard imaging techniques, such as fluoroscopy, still rely their analysis on the operator’s subjective conclusion (see table 1) (1) (2). In this context, ultrasound imaging, that can be performed at the patient’s bedside, is a non-invasive tool allowing the evaluation of the main structures involved in all the swallowing phases and may be a promising tool to dysphagia assessment (3).

Patients and methods: This is a narrative review on the interests of ultrasound imaging for swallowing assessment as well as on its perspective of use in physiotherapy. The figures and practical descriptions are the result of the authors’ everyday use of ultrasound imaging.

Results: Ultrasound imaging allows a quantitative and qualitative examination of the majority of the swallowing structures, such as tongue, oropharyngeal muscles, larynx, and upper esophageal sphincter (UES). It assesses the tongue kinetics, looking for abnormal movements, and the contractility and thickness of the oropharyngeal muscles, which enables to assess a potential dysfunction (4) (5). Physiotherapists may also follow the larynx movements (in coronal and sagittal planes) evaluating the displacement of the hyoid bone (6). A decrease in the hyoid bone displacement and contraction of the tongue may explain a dysphagia (7). Finally, ultrasound imaging can measure the displacement and the opening-closing diameters of the upper esophagus sphincter (UES), allowing to detection of dysphagia related to UES disorders.

Conclusion: Ultrasound imaging of the aerodigestive tract may improve the quantitative and qualitative examination of the swallowing structures. The use of this method requires a deep understanding of the ultrasound imaging principles and of the swallowing physiopathology in order to be able to benefit from its potential. Ultrasound imaging seems to be a promising tool for physiotherapy, used as an outcome in clinical research or during clinical practice in order to improve the diagnosis of the dysfunctions related to dysphagia.

Oral communications: Doctors

CO-01 Impact of mechanical ventilation setting on the occurrence of acute kidney injury in ICU patients - insights from the MIMIC-III database

Guillaume Geri (speaker)1, Loic Ferrer2, Nam Tran3, Matthieu Jamme4, Leo Anthony Celi5, Joon Lee6, Antoine Vieillard-Baron 1

1Ambroise Paré Hospital, APHP, Boulogne-Billancourt, FRANCE; 2Curie Intitute, Paris, FRANCE; 3Waterloo University, Waterloo, CANADA; 4Versailles Saint Quentin University, Versailles, FRANCE; 5Massachusetts Institute of Technology, Cambridge, UNITED STATES; 6University of Waterloo, Waterloo, CANADA
Correspondence: Guillaume Geri -

Annals of Intensive Care 2019, 9(Suppl 1):CO-01

Introduction: Mechanical ventilation in ICU patients may induce acute kidney injury (AKI). We aimed to describe the effect of mechanical ventilation (MV) settings, on AKI worsening, as well as the potential role of mean perfusion pressure (MPP).

Patients and methods: We included from the MIMIC-III database adult patients admitted for the first time in the ICU. We excluded patients with known chronic kidney disease, no KDIGO information and who suffered KDIGO3 AKI at day-1. The main outcome was one-KDIGO category AKI worsening (compared to the day before) and included as a categorical variable - discharged alive without AKI worsening and death before AKI worsening. We used a multinomial logistic regression at day 1 and day 2 according to a landmark-approach, with a two-days sliding perspective.

Results: 26,884 patients met the inclusion criteria (15,046 male, 56.0% + median age 65 [iqr 52, 78]). ICU and hospital mortality were 7.4 and 10.7%, respectively. Between day 1 and day 3, 529 patients died without AKI worsening, 11,230 were discharged alive and 1,656 suffered an AKI worsening. Between day 2 and day 4, 128 patients suffered an AKI worsening, 5,686 were discharged alive and 313 died. In multivariable analysis, using the “no MV” modality as a reference (n = 19,666), MV at day 1 was associated with AKI worsening at day 3 (relative risk ratio [RRR] 15.7 [7.7, 32], 23.2 [17.6, 30.6] and 55.6 [29, 106.4] for MV with PEEP < 5 (n = 283), PEEP between 5 and 8 (n = 4,153) and PEEP > 8cmH2O (n = 1,134)). **RRR of AKI worsening at day-4 were 2.3 [0.3, 19.4], 13.3 [5.8, 30.6] and 44.9 [17.2, 117.2] for MV with PEEP < 5 (n = 254), PEEP between 5 and 8 (n = 2,751) and PEEP > 8cmH2O (n = 1,088) at day 2 compared to no MV (n = 16,397). Hemodynamic parameters are shown on Figure. MPP significantly differed across MV groups (67, 70, 68 and 64 mmHg in patients receiving no MV, MV with PEEP < 5, PEEP 5–8 and PEEP > 8cmH2O, p < 0.001).

Conclusion: Mechanical ventilation was associated with AKI worsening in an increasing PEEP dependent manner at the early phase of ICU management in a large cohort of patients. Interaction between MPP and mechanical ventilation should be explored further.

CO-02 Transvenous Renal Biopsy of Critically Ill Patients- Safety and Diagnostic Yield

Marc Pineton de Chambrun (speaker)1, Philippe Cluzel2, Isabelle Brocheriou3, Nicolas Brechot1, Guillaume Hekimian1, Guillaume Franchineau1, Côme Bureau 1, Simon Bourcier1, Ania Nieszkowska1, Loic Le Guennec4, Zahir Amoura4, Alexis Mathian4, Matthieu Schmidt1, Alain Combes1, Charles-Edouard Luyt1

1Service de médecine intensive réanimation, ICAN, Hôpital La Pitié-Salpêtrière, Sorbonne Université, APHP, Paris, France., Paris, FRANCE; 2Hôpital La Pitié-Salpêtrière, Sorbonne Université, APHP, Paris, FRANCE; 3Service d’anatomopathologie, Hôpital La Pitié-Salpêtrière, Sorbonne Université, APHP, Paris, France, Paris, FRANCE; 4Service de médecine interne 2, Hôpital La Pitié-Salpêtrière, Sorbonne Université, APHP, Paris, France, Paris, FRANCE
Correspondence: Marc Pineton de Chambrun -

Annals of Intensive Care 2019, 9(Suppl 1):CO-02

Introduction: Introduction- Transvenous renal biopsy is an alternative way to obtain kidney samples from patients with bleeding-risk factors (e.g., antiplatelet therapy, anticoagulation or coagulation disorders…). This study was undertaken to determine the safety and diagnostic yield of transvenous renal biopsy of critically ill patients.

Patients and methods: Patients and Methods - Monocenter, retrospective, observational cohort study in a 26-bed French tertiary ICU. All patients undergoing in-ICU transvenous renal biopsy between January 2002 and February 2018 were included.

Results: Results- Eighty patients (male female sex ratio, 0.95 + mean ± SD age, 47.3 ± 18.3 years) were included. A histologic diagnosis was obtained for 77 (96.3%) patients, with acute tubular necrosis being the most frequent- 23 (29.9%). A potentially treatable cause was found for 47 (58.7%) patients. The numbers of patients with 0, 1, 2 or 3 factors (i.e., antiplatelet therapy, thrombopenia (< 150 G L) and preventive or curative anticoagulation) at the time of the biopsy were, respectively- seven (8.8%), 37 (46.2%), 31 (38.7%) and five (6.3%). Four (5%) and two (2.5%) patients, respectively, had renal hematoma and macroscopic hematuria + none required any specific treatment. Six (7.5%) patients died in-ICU and 90-day mortality was 8 80 (10%). No death was related to transvenous renal biopsy and median biopsy-to-death interval was 38 [IQR, 19.7 + 86] days.

Conclusion: Conclusions- Based on this cohort of ICU patients with acute renal failure, transvenous renal biopsy was safe and obtained a high diagnostic yield for these critically ill patients, even in the presence of multiple bleeding risk factors.

CO-03 The incidence of chronic kidney disease three years after non-severe acute kidney injury in critically ill patients - a cohort study

Arthur Orieux (speaker), Sébastien Rubin, Benjamin Clouzeau, Claire Rigothier, Christian Combe, Didier Gruson, Alexandre Boyer

CHU, Bordeaux, FRANCE
Correspondence: Arthur Orieux -

Annals of Intensive Care 2019, 9(Suppl 1):CO-03

Introduction: The risk of chronic kidney disease (CKD) following severe acute kidney injury (AKI) in critically ill patients is well documented. However, the long-term risk of CKD for less severe AKI in critically ill patients has never been assessed.

Patients and methods: This prospective single-center intensive care unit (ICU) observational 3-years follow-up study was carried out from 2013 to 2015 in Bordeaux (France). All patients with both severe (KDIGO 3) and non-severe AKI (KDIGO stage 1, 2) were enrolled. Patients with prior CKD were excluded. The primary outcome was the 3-years prevalence of CKD (defined by estimated glomerular filtration rate lower than 60 mL/min 1.73m2) in the non-severe AKI group. Secondary outcomes were risk factors for 3-years CKD, renal survival during follow-up and to identify the proportion among CKD patients followed by a nephrologist. Renal recovery was defined as return of creatinine at ICU discharge to < 26.5 µmol L above baseline.

Results: We screened 304 patients and after exclusion of 72 because of prior CKD 232 patients were enrolled. Severe AKI was observed in 120 (52%) and non-severe AKI in 112. At the end of the follow-up and after exclusion of patients who died, 41 and 30 patients remained in the non-severe and severe AKI group respectively. The global prevalence of CKD was 23 71 (32%). It was higher in the severe vs. non-severe AKI group (14 30 (47%) vs. 9 41 (22%) + p < 0.05) (table 1). Independent risk factors for 3-years CKD were diabetes (OR = 5.7 [1.5–21.5]) and absence of renal recovery at ICU discharge (OR = 14.2 [2.9–69.1]). Eleven out of 23 (48%) CKD patients were followed by a nephrologist.

Discussion: This study is the first to give the prevalence of CKD 3 years after non-severe AKI in critically-ill patients. No study has focused on stage 1–2 AKI in critically ill patients and only few studies have explored stage 1- 2 AKI in non-critically ill patients. The high prevalence of CKD at 3 years and the limited amount of CKD patients followed by a nephrologist confirm the need for accurate follow-up planning after ICU discharge in all patients diagnosed with AKI.

Conclusion: The risk of developing CKD at 3 years after non-severe AKI, despite lower than after severe AKI, remains high and is underestimated.

CO-04 Initiation strategies for renal replacement therapy for acute kidney injury and long-term survival- a follow-up from the AKIKI randomized controlled trial

Khalil Chaïbi (speaker)1, Frank Ehooman2, David Hajage3, Frederique Schortgen4, Laurent Martin-Lefevre5, Charles Verney6, Bertrand Pons 7, Eric Boulet8, Alexandre Boyer9, Guillaume Chevrel10, Nicolas Lerolle11, Dorothée Carpentier12, Nicolas De Prost13, Alexandre Lautrette14, Anne Bretagnol15, Julien Mayaux16, Saad Nseir17, Bruno Megarbane18, Marina Thirion19, Jean-Marie Forel20, Julien Maizel21, Yonis Hodane22, Philippe Markowicz23

1Avicenne Hospital, Bobigny, FRANCE; 2INSERM, Paris, FRANCE; 3Hôpital La Pitié Salpêtrière, Département Biostatistique, Santé Publique et Information Mé, Paris, FRANCE; 4CHU Henri Mondor, Creteil, FRANCE; 5Réanimation Médico-Chirurgicale, Centre Hospitalier Général, La Roche-Sur-Yon, FRANCE; 6Hopital Louis Mourier, Colombes, FRANCE; 7CHU Les Abymes Guadeloupe, Les Abymes, FRANCE; 8CH René Dubos, Pontoise, FRANCE; 9Hôpital Pellegrin-Tripode, Bordeaux, Bordeaux, FRANCE; 10CH Sud Francilien, Corbeil-Evry, Corbeil-Evry, FRANCE; 11CHU, Angers, FRANCE; 12CHU, Rouen, FRANCE; 13CHU Henri Mondor, Créteil, FRANCE; 14CHU Gabriel Montpied, Clermond-Ferrand, FRANCE; 15CHR, Orléans, FRANCE; 16Hôpital La Pitié Salpetrière, Paris, FRANCE; 17Hôpital R. Salengro, CHU, Lille, FRANCE; 18Hopital Lariboisière, Paris, FRANCE; 19CH Victor Dupouy, Argenteuil, FRANCE; 20Hopital Nord, Marseille, FRANCE; 21CHU, Amiens, FRANCE; 22CHU, Lyon, FRANCE; 23CH, Chollet, FRANCE
Correspondence: Khalil Chaïbi -

Annals of Intensive Care 2019, 9(Suppl 1):CO-04

Introduction: Epidemiological and experimental studies showed a link between acute kidney injury (AKI) and long-term outcome such as mortality and chronic kidney disease (CKD). Whether the timing of renal replacement therapy (RRT) during AKI eventually affects these outcomes is unknown. This study analyzes the long-term outcome of patients included in the Artificial Kidney Initiation in Kidney Injury (AKIKI) according to RRT initiation strategy.

Patients and methods: The AKIKI trial was a prospective, multicenter, open-label, two-arm randomized controlled trial that compared two RRT initiation strategies in patients receiving catecholamines and or mechanical ventilation and presenting with severe KDIGO3 AKI and no potentially life-threatening condition (NEJM, 2016 + 375-122-133). In the early strategy, RRT was started within 6 h after randomization criteria. In the delayed one, RRT was initiated only if one or more following criteria occurred- severe hyperkalemia, severe acidosis, severe pulmonary edema due to fluid overload resulting in severe hypoxemia, oligo-anuria > 72 h or serum urea concentration > 40 mmol/L. We assessed long-term (> 2 years) outcomes of the patients who survived 60 days after randomization. We collected vital status, last renal function value and health-related quality of life. We used a Kaplan–Meier estimator and a log rank test to analyze long-term survival according to randomization group.

Results: The median follow-up was 2.5 years (confidence interval [CI], 2.3 to 2.7 years). The early strategy group initially included 311 patients of whom 151 were alive 60 days after randomization and 145 were still alive at 2 years. The delayed strategy group initially included 308 patients of whom 153 were alive 60 days after randomization and 145 were still alive at 2 years (p = 0.94 for comparison between groups). Kaplan–Meier curves (see Figure 1) were superposed (p value- 0.79). The analyses of long-term renal function and health related quality of life are still in progress.

Conclusion: Although initial mortality of AKI in critically ill patients was high (49%), patients who survived the acute episode had a good vital prognosis at 2 years with only 6% additional deaths. Long-term death rate was not influenced by the initial RRT initiation strategy. Data on renal function and health quality will be available at the time of presentation.

CO-05 Epidemiology of Post-influenza Bacterial Pneumonia Due to a Panton-Valentine Leukocidin Positive Staphylococcus aureus in Intensive Care Unit- a retrospective nationwide study

Audrey Jacquot (speaker)1, Bruno Levy1, Charles Edouard Luyt2, Charles Vidal3, Sami Hraiech4, Jean-Pierre Quenot5, Lara Zafrani 6, Francis Schneider7, Pierre Kalfon8, Vincent Piriou19, Saadalla Nseir10, Paul Jaubert11, Jonathan Messika12, Muriel Fartoukh13, Xavier Valette14, Alexandre Lautrette15, Antoine Marchalot16, Nicolas Terzi17, Guillaume Schnell18, Charlène Le Moal19, Emmanuel Novy20, Jérémie Lemarié1, Arnaud Galbois21

1CHU, Nancy, FRANCE; 2APHP Pitié-Salpétrière, Paris, FRANCE; 3CHU Felix Guyon, La Réunion, FRANCE; 4APHM Hôpital Nord, Marseille, FRANCE; 5CHU, Dijon, FRANCE; 6APHP Saint Louis, Paris, Paris; 7CHU Hautepierre, Strasbourg, FRANCE; 8CH, Chartres, FRANCE; 9CHU Lyon Sud, Pierre Bénite, FRANCE; 10CHU, Lille, FRANCE; 11APHP Cochin, Paris, FRANCE; 12APHP Colombes, Colombes, FRANCE; 13APHP Tenon, Paris, FRANCE; 14CHU, Caen, FRANCE; 15CHU, Clermont-Ferrand, FRANCE; 16CH, Dieppe, FRANCE; 17CH Grenoble, La Tronche, FRANCE; 18CHU, Le Havre, FRANCE; 19CHU, Le Mans, FRANCE; 20CHU Nancy, Vandoeuvre-Les-Nancy, FRANCE; 21Hôpital privé Claude Galien, Quincy-sous-Sénart, FRANCE
Correspondence: Audrey Jacquot -

Annals of Intensive Care 2019, 9(Suppl 1):CO-05

Introduction: Superinfection of Influenza pneumonia with Staphylococcus aureus is a well-known adverse event. Conversely, published data on the epidemiology of superinfection with a Staphylococcus aureus producing the Panton-Valentine Leukocidin (SA PVL) remains scarse. This study evaluates the mortality and the epidemiology of this specific complication of influenza pneumonia in Intensive Care Unit (ICU).

Patients and methods: We conducted a multicenter retrospective epidemiological study over a period from January 2009 to December 2017. Were included patients admitted to ICU for SA PVL co infecting influenza pneumonia. The primary endpoint was mortality in ICU. Secondary endpoints included demographic, clinical characteristics of this population, description of complications during ICU stay, microbiological and therapeutic data.

Results: Of the 25 participating centres, 1970 patients admitted to ICU for influenza pneumonia were screened during the inclusion period. Prevalence of influenza pneumonia co infected with SA PVL was 1.1% (22 patients). Patients are young and had neither comorbidity, nor risk factors for severe influenza. The overall mortality in ICU of the cohort was 54.5% (12 patients). On admission to intensive care, patients were leucopenic (1.9 G L [0.7–4.8]) and 17 (81%) presented a severe ARDS (PaO2 FiO2 104 [66–320]) requiring vvECMO in 10 patients (45.5%). The ICU length of stay was 39.0 days [13.0–53.0] for survivors and 3.5 days [1.0–6.5] for non-survivors. No difference of anti-viral or antibiotic treatment between the two groups was recorded.

Conclusion: SA PVL superinfection of influenza pneumonia is rare but extremely severe complication, with a high and early mortality. Patients most often presented with severe ARDS frequently requiring vvECMO.

Declarations: Any conflict of interest for this work.

CO-06 Invasive pulmonary aspergillosis is a rare complication in critically ill patients with influenza

Anne Coste (speaker)1, Aurélien Frérou2, Jean Morin3, François-Xavier Blanc4, Jean Reignier5, Gilles Névez6, Patrice Le Pape 7, Jean-Marie Tonnelier8, Cédric Bretonnière9, Cécile Aubron10

1Service de maladies infectieuses, Brest, FRANCE; 2CHU Pontchaillou, Rennes, FRANCE; 3CHU - Service de pneumologie, Nantes, FRANCE; 4Service de pneumologie-CHU, Nantes, FRANCE; 5Réanimation médicale, Nantes, FRANCE; 6Département de mycologie parasitologie, Brest, FRANCE; 7Nantes University Hospital, Nantes, Nantes; 8Réanimation médicale, Brest, FRANCE; 9Réanimation médicale, Nantes, FRANCE; 10Réanimation médicale, Brest, FRANCE
Correspondence: Anne Coste -

Annals of Intensive Care 2019, 9(Suppl 1):CO-06

Introduction: Some studies show a high incidence of invasive pulmonary aspergillosis (IPA) in patients with severe influenza, and suggest to assess benefits of an antifungal prophylaxis. However, those studies come from the same area and lack of external validity. We aimed to measure the incidence of invasive pulmonary aspergillosis (IPA) and aspergillosis colonization in patients with severe influenza admitted to the intensive care unit (ICU).

Patients and methods: This retrospective multicenter cohort study recruited all patients with influenza admitted to four ICU in three French tertiary hospitals between September 1, 2009 and April 30, 2018. Patients were adults and had a confirmed influenza infection based on a positive airway PCR test. Patients with an Aspergillus-positive lower respiratory tract specimen culture (patients with Aspergillus) were diagnosed with Aspergillus colonization and invasive pulmonary aspergillosis (putative or proven) according to AspICU criteria. A multivariate logistic regression was performed to determine the factors associated with invasive pulmonary aspergillosis (IPA) and aspergillosis colonization.

Results: Four hundred and ninety-one patients were included. Three hundred and fifty-three patients (78.3%) were infected with Influenza A and 98 patients (21.7%) with Influenza B. Mean Simplified Acute Physiology Score II was 43.9 (SD = 20.1) and 97 patients (19.6%) died in ICU. Three hundred and fifty-six patients (72.2%) were under invasive mechanical ventilation and 32 patients (6.5%) underwent extra-corporeal membrane oxygenation. Twenty-six patients (5.2%) had an Aspergillus-positive lower respiratory tract specimen culture, including 12 patients (2.4%) diagnosed with IPA according to AspICU criteria. In multivariate analysis, factors associated with Aspergillosis-positive lower respiratory tract specimen culture were liver cirrhosis, haematological malignancy, use of vasopressors and H1N1 influenza (Table). Mortality rate did not significantly differ between patients with and without Aspergillus (26.9% and 19.2%, p = 0.32).

Conclusion: In this large French cohort study of patients with severe influenza, IPA was diagnosed in 2.4% of critically ill patients with Influenza infection. Prospective research is warranted to confirm our findings.

Table 1. Factors associated with Aspergillosis-positive lower respiratory tract specimen culture in multivariate analysis


Adjusted odds ratios

95% Confidence interval


Liver cirrhosis



9.4 × 10−6

Haematological malignancy



2.3 × 10−3





H1N1 Influenza infection




CO-07 Influenza virus infection is associated to global and persistent immunosuppression in ICU patients

Ana Catalina Hernandez Padilla (speaker)1, Robin Jeannet2, Thomas Lafon3, Olivier Barraud4, Sébastien Hantz4, Philippe Vignon5, Bruno François 5, Thomas Daix6

1Limoges, FRANCE; 2CNRS UMR 7276 Inserm U1262, CHU Dupuytren, Limoges, FRANCE; 3Inserm CIC 1435 Service d’Accueil des Urgences, CHU Dupuytren, Limoges, FRANCE; 4Inserm UMR 1092 Laboratoire de Bactériologie-Virologie-Hygiène, Université de Limoges CHU Dupuytren, Limoges, FRANCE; 5Inserm CIC 1435 Réanimation polyvalente Inserm UMR 1092, CHU Dupuytren, Limoges, FRANCE; 6Inserm CIC 1435 Réanimation polyvalente, CHU Dupuytren, Limoges, FRANCE
Correspondence: Ana Catalina Hernandez Padilla -

Annals of Intensive Care 2019, 9(Suppl 1):CO-07

Introduction: Seasonal Influenza virus infection (IVI) is associated to high morbidity and frequent complications. Immune dysregulation secondary to IVI may contribute in both ARDS and overmortality and increase the risk of secondary infections. This study aimed at characterizing the immune profile of ICU patients admitted for IVI.

Patients and methods: Prospective, single-center, observational study in immunocompetent adults admitted to the ICU for IVI confirmed by PCR from nasopharyngeal swabs between January and March 2018. Immune-profile was assessed by flow cytometry on peripheral blood at admission and during the first two weeks of ICU stay. Subsets analyzed included- immature (CD16-) and mature (CD16 +) granulocytes, activated monocytes (CD14 + CD16 +), HLA-DR- monocytes (CD14 + HLA-DR-) and T-CD3 + lymphocytes. Bacterial coinfection was defined as presence of positive cultures from samples acquired within standard of care associated to clinical evidence of infection.

Results: Thirteen patients (51 ± 14 years old + 9 men) were admitted to ICU (SOFA score 4 ± 3 + APACHE 16 ± 5) with confirmed IVI (Influenza A 69%). Twelve patients required invasive mechanical ventilation, 4 patients had coinfections at admission and 4 had secondary infections between days 8 and 16. All coinfections were of pulmonary origin. Most of the community acquired pneumonia (CAP) cases were related to pneumococcus coinfection (75%), while isolations related to ventilator-associated pneumonia (VAP) were Staphylococcus aureus (n = 2), Staphylococcus epidermidis (n = 1) and Aspergillus fumigatus (n = 1). Four patients (31%) died, all with documented coinfection (CAP n = 2, VAP n = 2). All patients showed T-lymphopenia at admission that persisted until day 7 (0.539 ± 0.501 G L to 0.561 ± 0.095 G L) (Figure 1A). Activated monocytes were also low at admission and trended towards decrease (0.051 ± 0.047 G L to 0.007 ± 0.006 G L) (Figure 1B). Concomitantly, the number of myeloid derived suppressive cells increased over time- HLA-DR- monocytes (0.099 ± 0.097 G L to 0.128 ± 0.118 G L) and immature granulocytes (0.592 ± 0.912 G L to 2.808 ± 2.014 G L) (Figure 1B and C).

Conclusion: Patients admitted to ICU for IVI had an immunosuppressive profile characterized by decreased mature active cell subsets with concurrent increase in immature suppressive ones. This immunosuppression lasted over the first week of ICU stay, and could explain the increased risk for secondary infections and its related worst prognosis.

CO-08 Impact of respiratory viruses in ICU patients with community acquired-pneumonia (CAP)- a one-year retrospective single center study

Marion Giry (speaker), Marie Gueudin, Déborah Boyer, Adeline Baron, Gaetan Beduneau, Soumaya Skallil, Marie-Anne Melone, Steven Grange, Dorothée Carpentier, Fabienne Tamion, Christophe Girault, Benoît Misset

CHU Rouen, Rouen, FRANCE
Correspondence: Marion Giry -

Annals of Intensive Care 2019, 9(Suppl 1):CO-08

Introduction: Pneumonia is the most frequent community-acquired infection responsible for ICU admission. Multiplex PCR enables early diagnosis of viral infection in daily practice. Few series have been described since this technique has been made available in routine. Our objective was to assess the prevalence and distribution of viruses among ICU patients with CAP and their relationship with severity and outcome.

Patients and methods: Retrospective analysis of the consecutive viral multiplex PCR (Eplex™, Genmark Dx) between November 2016 and October 2017 in a French 21 bed medical ICU admitting around 1,000 patients per year. Patients’ nasopharynx was sampled within 72 h following their ICU admission. We selected those patients with a diagnosis of CAP and split them into 4 groups according to causal agent- none, viruses, bacteria, combination of both. Comparisons were made with non-parametric Kruskall-Wallis and Fischer’s exact tests.

Results: 223 patients were sampled, of whom 109 had CAP, 38 aspiration or opportunistic pneumonia, 22 non-pulmonary infections, 11 pulmonary edema, 19 exacerbations of chronic lung disease, and 24 other diagnoses. Patients with CAP had the following characteristics- age 59 ± 16 y, male sex 60%, SAPS 2 score 40 ± 18, ICU length of stay (LOS) 8.7 ± 9.0 d, mortality 10%. No infectious agent was found in 32 (29%), a virus in 28 (26%), bacteria in 33 (30%) and both a virus and bacteria in 16 (15%). The main bacteria were S. pneumonia (23%) and H. influenza (6.5%). The main viruses were Rhinovirus Enterovirus (13.8%), Influenzae A (11%) and Parainfluenzae (5.5%). The most frequent virus-bacteria association was S. pneumoniae and Influenzae A (3%) and S. pneumoniae and Rhinovirus Enterovirus (3%). The SAPS 2 score was higher in the mixed group (p = 0.02). The ICU-LOS was 5.2 (no agent), 7.7 (virus), 10.1 (bacteria) and 14.8 (mixed) days respectively (p = 0.04). Mortality was similar among groups.

Conclusion: In our ICU population, respiratory viruses were present in 40% of CAP. Patients with mixed infection had higher severity at admission and longer ICU-LOS.

CO-09 Invasive Pulmonary Aspergillosis in Critically Ill Patients with Hematological Malignancies

Emmanuel Pardo (speaker)1, Virginie Lemiale2, Djamel Mokart3, Annabelle Stoclin4, Anne-SophieMoreau5, Lionel Kerhuel2, Etienne Ghrenassia 2, Laure Calvet2, Audrey De Jong2, Sandrine Valade2, Eric Mariotte2, Lara Zafrani2, Michael Darmon2, Elie Azoulay2

1CHU Saint-Antoine, Paris, FRANCE; 2Hôpital Saint-Louis, Paris, FRANCE; 3Institut Paoli-Calmettes, Marseille, FRANCE; 4Institut Gustave Roussy, Villejuif, FRANCE; 5Hôpital Albert Calmette, Lille, FRANCE
Correspondence: Emmanuel Pardo -

Annals of Intensive Care 2019, 9(Suppl 1):CO-09

Introduction: Invasive fungal infections remain associated with high mortality rates among patients with hematological malignancy. Recent medical advances in antifungal prophylaxis, diagnostic criteria and treatments of invasive pulmonary aspergillosis (IPA) have been reported. We sought to assess whether these advances translate into change in survival in patients with acute respiratory failure and IPA.

Patients and methods: Retrospective, multicenter study performed in four centers. Adult patients with hematological malignancy, proven or probable IPA, and acute respiratory failure requiring ICU between January 1998 and December 2017 were included. Results are reported as n (%) or median (IQR). A cox regression model was used to identify variables independently associated with 6 months survival.

Results: Overall, 219 patients were included. 138 (63%) were of male gender and median age was of 55 (IQR 44–64). Acute myeloid leukemia (30.1%) and non-Hodgkin lymphoma (22.8%) were the most frequent malignancies. 134 patients were neutropenic at study inclusion (62%), and 53 were allogeneic stem cell recipients (24.2% + including 64.2% who suffered from Graft-versus-Host Disease) and 22 receiving antifungal prophylaxis (10%). Median SOFA score at admission was 9 [7–12] and median time before introduction of invasive mechanical ventilation was 1 [0–3] day. 154 patients (70.3%) had positive serum galactomannan and 136 (62.1%) a positive culture. ICU and 6-months mortality remained unchanged during the 20-years study period, being respectively 58.4% and 80.1% (Figure 1A). Need for invasive mechanical ventilation, at admission or following failure of non-invasive techniques, was associated with a high mortality rate (Figure 1B). Use of Voriconazole (HR 0.67, IC95 0.48–0.94) and focal radiologic pulmonary infiltrate (HR 0.58, IC95 0.41–0.80) were independently associated with better survival by multivariable analysis adjusted on day-1 SOFA score (HR 1.1, IC95%1.06–1.15).

Conclusion: IPA still mostly affects patients with neutropenia and BMT recipients. Case fatality remains high and sustained over time. Voriconazole treatment and focal infiltrate (when compared to diffuse infiltrate) are associated with better survival. Strategies to improve survival in these high-risk patients are warranted.

CO-10 Changes in outcome of critically ill cancer patients over the last decades- Results of a systematic review on individual data

Michael Darmon (speaker)1, Aurélie Bourmaud2, Quentin Georges3, Marcio Soares4, Kyeongman Jeon5, Sandra Oeyen6, Chin Kook Rhee 7, Pascale Gruber8, Marlies Ostermann9, Quentin Hill10, Peter Depuydt11, Christelle Ferra12, Anne-Claire Toffart13, Peter Schellongowski14, Alice Muller15, Virginie Lemiale1, Fabien Tinquaut16, Djamel Mokart17, Elie Azoulay1

1Saint-Louis University Hospital, Paris, FRANCE; 2Robert Debre Hospital, Paris, FRANCE; 3Saint-Etienne University Hospital, Saint-Etienne, FRANCE; 4D’or Institute, Rio de Janeiro, BRAZIL; 5Samsung Medical Center, Seoul, REPUBLIC OF SOUTH KOREA; 6Ghent University Hospital, Ghent, BELGIUM; 7Catholic University of Korea, Seoul, Seoul; 8Royal Mardsen Hospital, Londres, UNITED-KINGDOM; 9Guy’s & St Thomas’ NHS Foundation Hospital, Londres, UNITED-KINGDOM; 10Leeds Teaching Hospitals, Leeds, UNITED-KINGDOM; 11Ghent University Hospital, Ghent, UNITED-KINGDOM; 12Universitat Autònoma de Barcelona, Barcelone, SPAIN; 13Grenoble University Hospital, Grenoble, FRANCE; 14Comprehensive Cancer Center, Vienne, AUSTRIA; 15Universidade Federal do Rio Grande do Sul, Porto Alegre, BRAZIL; 16Hygée Centre and Public Health Department, Saint-Etienne, FRANCE; 17Paoli Calmette Institute, Marseille, FRANCE
Correspondence: Michael Darmon -

Annals of Intensive Care 2019, 9(Suppl 1):CO-10

Introduction: Overall prognosis of critically ill patients is thought to have improved over the last decades (1). However, this assumption is based upon unadjusted findings and before after studies (1, 2). The aim of this study was to assess the influence ICU admission year on outcome of the critically-ill cancer patient population. Secondary objectives were to assess changes of prognosis in pre-specified subgroup of patients, and independent risk factors of mortality.

Patients and methods: This study resulted from a systematic review and meta-analysis on individual data performed according to the PRISMA statements. Public-domain databases (PubMed and Cochrane) were searched by using predefined keywords. The research was restricted to articles published in English and studies focusing on critically ill adult patients from May 2005 to May 2015. The study protocol was registered in the PROSPERO database (CRD42015026347). Selected manuscripts’ authors were then contacted and individual data included in this analysis. Results are reported in n (%) or median (IQR). Adjusted mortality was assessed using a mixed logistic regression model taking year of ICU admission as both fixed and random effect, and study as random effect.

Results: Overall, 7,356 patients were included in this study, including 1,666 patients with neutropenia at ICU admission. Median age was 60 years (IQR 49–69). Median SAPSII score at ICU admission was 42 (IQR 28–57). Mechanical ventilation, vasopressors, and renal replacement therapy were required in 50.7% (n = 3.729), 41.1% (n = 3.024) and 16.1% (n = 1.174) of the included patients. Median ICU admission year was 2007 (IQR 2004–2010 + range 1994–2012). Hospital mortality was 47.4% in the overall population, ICU admission year being associated with a progressive decrease in hospital mortality (OR per year 0.94 + 95%CI 0.93–0.95). After adjustment for confounders, year of ICU admission was independently associated with hospital mortality (OR for hospital mortality per year- 0.96 + 95%CI 0.95-.97). This finding was confirmed in all of the subgroups except in patients with allogeneic stem cell transplantation (figure—mortality per year, trend + 95%CI).

Conclusion: Our results confirm that hospital mortality of critically ill cancer patients steadily decreased over time after adjustment for patients’ characteristics, patients’ severity and clustering effect. This finding was confirmed in pre-defined subgroups with the exception of allogeneic stem cell transplant recipients.

  1. 1.

    Shimabukuro-Vornhagen CA-Cancer 2016

  2. 2.

    Peigne ICM 2009


CO-11 Role of targeted therapy for cancer patients admitted to ICU

Virginie Lemiale (speaker)1, Anne-Pascale Meert2, Anne-Claire Toffart3, François Vincent1, Aude Gibelin4, Djamel Mokart5, Andry Van de Louw 6, Stefan Hatzl7, Karin Amrein8, Gaelle Rousseau-Bussac9, Philip Bauer10, Dorothée Carpentier11, Fabrice Bruneel12, Gabo Moreno13, Lucas-Maria Montini14, Anne-Sophie Moreau15, Pleun Hemelaar16, Elie Azoulay1

1APHP Saint Louis, Paris, FRANCE; 2Institut Jules Bordet, Bruxelles, BELGIUM; 3CHU, Grenoble, FRANCE; 4CHU Tenon, Paris, FRANCE; 5Institut Paoli Calmettes, Marseilles, FRANCE; 6Penn State College of Medicine, Hershey, UNITED-STATES; 7Medical University of Graz, Graz, AUSTRIA; 8CHI Creteil, Creteil, FRANCE; 9Mayo Clinic, Rochester, UNITED STATES; 10CHU, Rouen, FRANCE; 11Hôpital Mignot, Versailles, FRANCE; 12Hospital Universitari de Bellvitge, Barcelone, SPAIN; 13A. Gemelli University Hospital, Roma, ITALY; 14CHU Lille, Lille, FRANCE; 15VU medical center, Amsterdam, THE NETHERLANDS
Correspondence: Virginie Lemiale -

Annals of Intensive Care 2019, 9(Suppl 1):CO-11

Introduction: Mortality of cancer patients remains high. In the last years, new treatments such as targeted therapies, improved outcome of patients with solid tumors. However, those new therapies have been associated with complications leading to ICU admission. The incidence of those severe adverse events is currently unknown. The aim of this observational study was to describe severe adverse effects of targeted therapy.

Patients and methods: We conducted an observational multicentric study in 14 international centers belonging to our research group. Critically ill adult patients admitted for any reason between January 1st 2015 and December 31 2016 and treated for solid tumor with targeted therapy prior to ICU admission were included.

Results: 91 patients, aged of 61 (51–69) years were included, 56 (62%) were non- smokers patients. Underlying cancer were gasto-intestinal tract (n = 24.22%), breast (n = 18, 20%), kidney (n = 17, 19%), lung (n = 17, 19%) and melanoma (n = 4.4%). 81 (83% patients had metastasis, 44 (48%) patients were PS < 2.21 patients (23%) were treated with anti-EGFR, 25 (28%) patients with anti-VEGF and 5 (5.5%) anti-BRAF. Targeted therapy was started 72 days [17–151] before ICU admission. At ICU admission, the median SOFA score was 3 (1–6). Symptoms at ICU admission were respiratory failure (n = 31 + 35%), infectious diseases (n = 15 + 17%), cardiac failure (n = 14 + 19%) and metabolic disorder (n = 10 + 11%). For 16 (20%) patients, ICU admission was directly related to targeted therapy. Side-effects related to targeted therapy were anaphylaxia (n = 1), cardiac failure (n = 3), bleeding (n = 2), peritonitis (n = 1), metabolic disorder (n = 3), lung injury (n = 4), encephalopathy (n = 1) and SIRS without other etiology (n = 1). 12% (2 16) of them died during ICU stay (2 patients with digestive cancer treated with anti-VEGF who had peritonitis (1 patient) and cardiogenic failure (1 patient)). Other diagnosis were mostly acute respiratory failure related to tumor progression or pneumonia (n = 28), infectious diseases (n = 15) or cardiac failure (n = 13). 65 (71%) patients were discharged alive from ICU. Median ICU stay duration was 2 (1–6) days and1 month survival rate was 55% (n = 50).

Conclusion: In this preliminary study, 20% (IC 11.8–28.2%) of ICU admission were related to severe adverse event for targeted therapy. Intensivists should be aware of those complications particularly anti –VEGF when more patients would be treated with those therapies in the further years.

CO-12 Cytomegalovirus reactivation in intensive care unit patients with hematological malignancies- characteristics and clinical outcomes

Alistair Baber (speaker)1, Camille Vissac1, Maud Salmona2, Jérôme Le Goff3, Audrey De Jong1, Eric Mariotte1, Lara Zafrani1, Elie Azoulay1, Michael Darmon1, Laure Calvet1

1AP-HP, ICU, Saint Louis Teaching Hospital, Paris, FRANCE; 2AP-HP, Virology Department, Saint Louis Teaching Hospital, Paris, FRANCE
Correspondence: Alistair Baber -

Annals of Intensive Care 2019, 9(Suppl 1):CO-12

Introduction: Cytomegalovirus (CMV) reactivation occurs frequently in the critically ill and is associated with poor outcomes (1, 2). However, data on immunocompromised patients are scarce. The primary objective of this study was to describe characteristics and outcomes in critically-ill patients with hematological malignancies and CMV reactivation. Secondary objectives included description of CMV disease incidence and characteristics, associations between whole blood viral load and CMV disease and outcomes.

Patients and methods: Retrospective single center study (Jan 2010-Dec 2017). Adult patients, admitted to the ICU, having underlying hematological malignancy and CMV reactivation were included. CMV disease was defined according to recent guidelines (2). Results are reported as median (interquartile—IQR) or n (%). Factors associated with hospital mortality or CMV disease were analyzed using logistic regression.

Results: 178 patients were included (median age 55y [42–64], 123 patients [69.1%] male). Underlying malignancies were mainly non-Hodgkin’s lymphoma (n = 89 + 50%), acute leukemia (n = 34 + 19.1%) and myeloma (n = 16 + 9.0%) + 37 patients (20.8%) were allogeneic stem cell transplant (SCT) recipients. At admission, median SOFA score was 6 [4–9], and 98 patients required mechanical ventilation (55%), 85 vasopressors (47.8%) and 58 renal replacement therapy (32.6%). Overall, hospital mortality was 53% (n = 95). Median CMV load (whole blood) was 2.7 Log [2.3–3.5]. 44 (24.7%) patients developed CMV disease, including 7 probable or proven CMV diseases (3 pneumonia, 2 encephalitis, 1 retinitis, 1 colitis) and 37 possible CMV diseases (including 19 cytopenia, 11 hepatitis and 8 pneumonia). Mortality was associated with CMV viral load quartiles (Figure, P = 0.02). After adjustment for confounders, need for vasopressors (OR 2.53 + 95%CI 1.11–5.97) and CMV load (OR 1.88 per Log + 95%CI 1.29–2.85) were associated with hospital mortality. When forced one by one in the model, neither CMV disease nor CMV treatment were associated with outcomes. SCT (OR 2.55 + 95%CI 1.05–6.16), mechanical ventilation (OR 4.11 + OR 1.77–10.54) and viral load (OR 1.77 per Log + 95%CI 1.23–2.61) were independently associated with CMV disease.

Conclusion: In critically ill hematology patients, CMV viral load is independently associated with hospital mortality. Conversely, neither CMV disease nor treatment was associated with outcome suggesting CMV reactivations reflect an immunodeficiency rather than a cause of poor outcome.

CO-13 Diaphragm thickening fraction correlates with diaphragm electrical activity in patients under mechanical ventilation

Suela Demiri (speaker)1, Bruno-Pierre Dube2, Julien Mayaux1, Elise Morawiec1, Thomas Similowski1, Martin Dres1, Alexandre Demoule1

1Hôpital Universitaire La Pitié-Salpêtrière, Paris, FRANCE; 2Centre Hospitalier de l’Université de Montréal, Montreal, CANADA
Correspondence: Suela Demiri -

Annals of Intensive Care 2019, 9(Suppl 1):CO-13

Introduction: The evaluation of patient’s inspiratory effort is helpful to optimally tailor ventilator settings. While diaphragm electrical activity has been proposed as a valuable tool to assess inspiratory effort, it requires a dedicated tool, namely the NAVA (Neurally Adjusted Ventilatory Assist) catheter that is not widely available. By contrast, the use of ultrasound to evaluated diaphragm contractile activity has been growing in the ICU. However, the relationship between diaphragm contractility activity, i.e. diaphragm thickening fraction (TFdi) and diaphragm electrical activity (EAdi) has not been explored so far. Our hypothesis was that TFdi correlates with EAdi. In the present study, we sought to investigate the correlation between TFdi and EAdi through several conditions of ventilatory assistance.

Patients and methods: Patients intubated and ventilated for at least 24 h were eligible for inclusion in the study if they had been previously mechanically ventilated with NAVA mode and the NAVA catheter had been left in place. Ultrasound measurements of diaphragm contractile activity were made synchronously with the measurement of maximal EAdi (EAdimax) during 6 consecutive, randomly assigned ventilatory conditions (NAVA 1 μV/ml, pressure support levels of 5, 10, 15 and 20 cmH2O (with Positive end exipartory pressure PEEP of 5 cm H2O), and zero PEEP with 7 cm H2O pressure support. Overall correlation between EAdimax and TFdi was assessed using Spearman’s correlation coefficient. We also sought to determine the correlation between EAdimax and TFdi for each condition.

Results: Twenty patients 60 (53–67) years old were enrolled in the study after a median (25–75 IQR) period of mechanical ventilation of 6 days (3–67). Overall, the correlation between TFdi and EAdimax was r = 0.34, p < 0.001, as shown in Figure 1. The Spearman’s correlation coefficient decreased when the level of ventilatory assistance increased. It was 0.54 (p = 0.03) for zero PEEP with 7 cm H2O pressure support and 0.53 (p = 0.03), 0.53 (p = 0.3), 0.34 (p = 0.2) and 0.15 (p = 0.57) for a level of pressure support of 5, 10, 15 and 20 cmH2O with PEEP of 5 cmH2O and 0.38 (p = 0.14) in NAVA.

Conclusion: There was a moderate correlation between diaphragm thickening fraction and diaphragm electrical activity. Our findings suggest that diaphragm thickening fraction behaves as a relevant surrogate of patient’s inspiratory effort but the determinants of this correlation deserve further studies.

CO-14 Transpulmonary pressures and computed tomography in experimental ARDS

Jean-Christophe Richard (speaker)1, Maciej Orkisz2, Marcela Hernandez Hoyos3, Alfredo Morales Pinzon3, ClaudeGuérin4

1HCL, Tassin | Rhone, FRANCE; 2CREATIS UMR 5220, Villeurbanne, FRANCE; 3Systems and Computing Engineering Department, School of Engineering, Universidad de los Andes, Bogota, COLOMBIE; 4Hospices Civils de Lyon- Hôpital de la Croix-Rousse, Lyon, FRANCE
Correspondence: Jean-Christophe Richard -

Annals of Intensive Care 2019, 9(Suppl 1):CO-14

Introduction: Esophageal pressure to compute transpulmonary pressure (TPP) is an appealing technique to individualize PEEP in ARDS. TPP can be computed 1. as the difference between airway and esophageal pressures (absolute measurement PLABS) + 2. as the product of airway pressure by the ratio of lung to respiratory system elastance (elastance-derived measurement PLEL). The aim of the study was to evaluate the relationships between both TPP measurements and computed tomography (CT) in an experimental model of ARDS.

Patients and methods: ARDS was performed by saline lavage on 16 piglets. A recruitment maneuver was performed followed by mechanical ventilation with VT 6 ml kg body weight. PEEP was set to 20 cmH2O then decreased by 2 cmH2O-steps down to 2 cmH2O. Piglets were then randomized into 3 PEEP groups based on best compliance during PEEP trial, best end-expiratory lung volume during PEEP trial, or a PEEP-FiO2 table. Finally, 7 levels of VT ranging from 4 to 20 ml kg were applied at optimal PEEP. TPP measurements and CT were performed at end-expiration and end-inspiration after ARDS onset, during the PEEP trial, 1 h after setting optimal PEEP, and during the variable VT trial.

Results: PLABS ranged from -6 to 58 cmH2O, and PLEL ranged from 1 to 65 cmH2O. PLABS and PLEL were significantly correlated (R2 = 0.87, p < 0.001). Bias between PLEL and PLABS amounted to 4 cmH2O (limits of agreement ranging from -2 to 11 cmH2O). Non-inflated compartment significantly increased in deciles of end-expiratory PLABS and PLEL below 4.6 cmH2O and 11.8 cmH2O, respectively (cf. figure). The table provides the diagnostic performance of end-expiratory PLABS and PLEL to detect a non-inflated compartment below 5% and 10% of total lung volume. Overinflated compartment significantly increased in deciles of end-inspiratory PLABS and PLEL above 25.8 cmH2O and 30.7 cmH2O, respectively. ROC curve identified an end-inspiratory PLABS above 19.6 cmH2O and an end-inspiratory PLEL above 24.6 as best thresholds to identify an overinflated compartment greater than 2% of total lung volume.

Conclusion: Both TPP measurement techniques have similar diagnostic performance to identify near complete alveolar recruitment at end-expiration and hyperinflation at end-inspiration. Near complete alveolar recruitment may be achieved with PLABS or PLEL greater than 5 and 10 cmH2O, respectively.

CO-15 Prone positioning monitored with electrical impedance tomography on patients with severe ECMO-supported ARDS

Guillaume Franchineau (speaker)1, Nicolas Brechot1, Guillaume Hekimian1, Guillaume Lebreton2, SimonBourcier1, Come Bureau1, Loïc Le Guennec 1, Nieszkowska Ania1, Pascal Leprince2, Luyt Charles-Edouard1, Alain Combes1, Matthieu Schmidt1

1APHP, Pitié-Salpêtrière Hospital, Medical Intensive Care Unit, Paris, FRANCE; 2APHP, Pitié-Salpêtrière Hospital, Cardiac Surgery Department, Paris, FRANCE
Correspondence: Guillaume Franchineau -

Annals of Intensive Care 2019, 9(Suppl 1):CO-15

Introduction: Prone positioning (PP) during veno-venous extracorporeal membrane oxygenation (ECMO) is feasible but its relative effects have never been thoroughly described. Objectives are to describe through electrical impedance tomography (EIT) the impact of PP on regional ventilation and optimal PEEP level, and to derive from EIT baseline predictive factors to identify “PP responders” on ECMO.

Patients and methods: ECMO-supported severe ARDS patients, ventilated with a pressure-controlled mode, a 14cmH2O driving pressure, and an EIT-based “optimal PEEP” were included. Before, during, and after a 16 h-PP session, EIT based distribution and variation of tidal impedance, centre of gravity index, end-expiratory lung impedance (EELI), and static compliance were collected. “PP responders” were identified as patients who increased their static compliance by more than 3 ml cmH2O after 16 h of proning.

Results: Thirteen (62%) out of 21 studied-patients were considered as “PP responders” on ECMO. “PP responders” had a greater body mass index, more frequently a viral pneumonia and a shorter ECMO duration compared to “PP non-responders” (p < 0.01). For both groups, tidal volume and EELI were redistributed from ventral to dorsal regions during PP. However, “optimal PEEP” was significantly lower in PP than in supine position with 14 (12 -16) and 11 (8–14.5) vs 12 (10–14) and 8 (7.5 -10.5) cmH2O in “PP responders” and “PP non-responders”, respectively (figure 1). Lastly, baseline center of gravity index was lower in “PP responders” (p = 0.03).

Conclusion: EIT allows monitoring PP impact on ventilation and seems to be a relevant tool to identify patients who will more likely respond to PP on ECMO.

CO-16 Respiratory muscles contraction after successful ventilator liberation trial predicts extubation failure

Martin Dres (speaker)1, Domenico Luca Grieco2, Wissalle Ouechani1, Irene Telias2, Detajin Junhasavasdikul2, Luana Melo2, Thomas Piraino 2, Felipe Damiani2, Lilya Sergenyuk1, Lauriane Degravi1, Tai Pham2, Alexandre Demoule1, Thomas Similowski1, Laurent Brochard2

1APHP, Paris, FRANCE; 2St Michael Hospital, Toronto, CANADA
Correspondence: Martin Dres -

Annals of Intensive Care 2019, 9(Suppl 1):CO-16

Introduction: Even for patients who successfully pass a ventilator liberation trial, the period after extubation is at risk of complications. In the present study we investigated whether assessment of respiratory muscles contraction within 2 h after extubation could predict extubation failure (acute respiratory failure, re-intubation or death during a 7-days period following extubation).

Patients and methods: Patients from two intensive care units who were intubated since at least 48 h and who successfully passed a ventilator liberation trial were enrolled in the study. Within the two hours following extubation, right hemi-diaphragm (TFdi) and right upper intercostal muscle thickening fraction (TFci) were obtained with ultrasound. TFdi and TFic were compared between patients with and without extubation failure (unpaired test). Receiver operating characteristic (ROC) curves were constructed to predict the risk of extubation failure for each indice.

Results: Of the 122 patients who participated in the study, 22 (18%) had extubation failure. Respectively in patients with and without extubation failure, TFdi was 11.9% (9.8–19.4) and 21.5% (15.8–26.3) (p < 0.01) and TFic was 12.8% (8.2–22.2) and 6.5% (4.2–9.1) (p < 0.01). AUC of the ROC curves were 0.73 (0.59–0.86) and 0.76 (0.60–0.91) for TFdi and TFic respectively (p = 0.54). A value of TFdi lower than 16% predicted extubation failure with a sensitivity of 70% (46–88) and a specificity of 73% (63–82). A value of TFic greater than 9% predicted extubation failure with a sensitivity of 75% (51–91) and a specificity of 74% (63–82). Eventually, TFdi TFic ratio was 3.4 (2.0–5.4) and 0.8 (0.3–2.6) in patients with and without extubation failure respectively (p = 0.003). The AUC of the ROC curve for TFdi TFic was 0.82 (0.70–0.94) providing a sensitivity of 71% (60–81) and a specificity of 74% (49–91) to predict extubation failure with a cut-off of 2.2.

Conclusion: Assessment of respiratory muscles contraction within two hours after extubation could be used to identify patients at risk of extubation failure. The potential interest of our cut-offs could be used at the time of considering preventive post extubation ventilation strategy.

CO-17 Effect of empirical aminoglycoside combination therapy on mortality in patients with septic shock- a propensity-based analysis

Jean-François Llitjos (speaker)1, Simon Meslin1, Julienn Charpentier1, Alain Cariou1, Jean-Daniel Chiche1, Jean-Paul Mira1, Matthieu Jamme 2, Frédéric Pène1

1Médecine-Intensive Réanimation, Paris, FRANCE; 2Urgences néphrologues et transplantation rénale, Paris, Paris
Correspondence: Jean-François Llitjos -

Annals of Intensive Care 2019, 9(Suppl 1):CO-17

Introduction: Current guidelines recommend combination antibiotic therapy for patients with septic shock. However the supporting evidence is scarce. The aim of this study is to address the impact of aminoglycoside combination therapy on mortality in septic shock.

Patients and methods: This was a 9-year (2008–2016) monocenter retrospective study. All adult patients diagnosed with septic shock within the first 48 h were included. Septic shock was defined as a microbiologically proven or clinically suspected infection, associated with acute circulatory failure requiring vasopressors. Empirical combination therapy was defined by the addition of aminoglycosides within 36 h from the diagnosis of septic shock. A propensity score for aminoglycoside administration was built using day 1 demographic and clinical characteristics. The determinants of mortality were assessed using a propensity-adjusted Cox proportional hazards competing risk analysis.

Results: During the study period, 1040 patients were admitted for septic shock. Among them, 64% were males, the median age was 69 [57 + 78] years old. The crude in-ICU mortality rate was 35%. The antibiotic regimen was based on beta lactam therapy in almost all patients (99%), combined with aminoglycosides in 616 (59%) patients. First-line aminoglycosides were distributed into amikacin (61%), gentamicin (37%) and tobramycin (2%). Patients receiving aminoglycoside combination therapy were more frequently immunocompromised (40.4% vs. 31.84%, p = 0.005), had more positive blood cultures (39.1% vs. 25%, p < 0.0001) and had a higher creatininemia at admission (144 [95 + 228] vs. 131 [80 + 201] μmol L, p = 0.007). After propensity score-based matching, 348 patients receiving aminoglycoside combination therapy were compared to 348 patients treated without aminoglycosides. Demographics and clinical characteristics on admission were similar in the two matched groups. The mortality rates in patients treated or not with aminoglycoside combination therapy were 36.2% and 30% (p = 0.07 in univariate analysis), respectively. In propensity-adjusted Cox proportional hazards multivariate competing risk analysis, aminoglycoside combination therapy was not associated with mortality (CSH = 1.19, 95%CI = 0.90–1.55, p = 0.25). After matching, day 3 creatininemia was similar in both groups (91 [54 + 158] vs. 89.5 [54.75 + 158] μmol L, p = 0.86) whereas creatininemia at discharge was higher in patients treated with aminoglycoside combination therapy (92.5 [58–177] vs. 79.5 [54–137.5] μmol L, p = 0.02).

Conclusion: In septic shock patients, antibiotic combination with aminoglycosides was associated with neither benefit nor harm.

CO-18 Potential impact of real-time processing and rapid susceptibility testing of blood samples in severe gram-negative bloodstream infections

Sophie Alviset (speaker), Julien Charpentier, Hélène Poupet, Claire Poyart, Jean-Paul Mira, Solen Kerneis

APHP Hôpital Cochin, Paris, FRANCE
Correspondence: Sophie Alviset -

Annals of Intensive Care 2019, 9(Suppl 1):CO-18

Introduction: Timely appropriate therapy is critical in patients with gram-negative bloodstream infections. In standard care, blood cultures are processed during operating hours of the laboratory (8-30-18-30 in our setting). We evaluated the potential impact of real-time processing (24 h a day) and rapid antimicrobial susceptibility testing (AST) on antibiotic regimen in intensive care unit (ICU) patients.

Patients and methods: We retrospectively reviewed all episodes of gram-negative bloodstream infections occurring in the ICU of our 1500-bed hospital between 1 st January and 31s December 2017. We collected data on demographics, outcomes, timeframes of sample processing testing and antibiotic changes. Secondly, three scenarios were simulated on the dataset (1) Real-time processing + conventional techniques (gram stain and standard AST) + (2) Standard processing (8-30-18-30) + rapid AST + (3) Real-time processing + rapid AST**In scenarios 1 and 3, transportation time of blood samples to the laboratory was set at 2 h. In scenarios 2 and 3, time to AST result was set at 7 h.

Results: We included 86 episodes in 76 patients- 52 men (68%), median age 67 [interquartile range- 53–78]). Median Charlson’s score was 5 [0–15], and 28 patients (37%) died in the ICU. Infection episodes were mainly digestive (26 + 30%) or catheter related (16 + 19%). Most frequent pathogens were- E. coli (36, 33%), Enterobacter sp (20, 18%), P. aeruginosa (11, 10%). Reception of the AST led to a change of the antibiotic regimen in 53 86 (62%) episodes (de-escalation in 46). In standard care, median timeframe from blood collection to gram stain was 26 h [19–34] and from gram stain to AST 29 h [27–50]. Median timeframe for complete processing of blood samples (from collection to result of definitive AST) was 61 h [53–77] in standard care, 48 h [42–66] in scenario 1, 33 h [26–41] in scenario 2 and 22 h [20–27] in scenario 3. Over the whole sample, scenario 1, 2 and 3 had the potential to prompt antibiotic change to pathogen-directed therapy by an average of 5.6, 18.6 and 24.1 h respectively.

Conclusion: Real-time processing and rapid AST of blood cultures have a high potential to decrease time to antibiotic change in severe gram-negative bloodstream infections. This effect must be balanced with cost of diagnostic kits and around-the-clock staffing expenses.

CO-19 Prevalence of insufficient plasma concentration of beta-lactam antibiotics ICU patients- a one-year retrospective single centre study

Sacha Sarfati (speaker), Fabien Lamoureux, Elsa Desmarest-Durand, Soumaya Skallil, Thomas Clavier, Dorothée Carpentier, Steven Grange, Gaetan Beduneau, Fabienne Tamion, Christophe Girault, Benoît Misset

Correspondence: Sacha Sarfati -

Annals of Intensive Care 2019, 9(Suppl 1):CO-19

Introduction: Beta-lactam antibiotics effectiveness is time-dependent and requires that their plasmatic concentration is constantly over the Minimal Inhibiting Concentration (MIC) of the targeted bacterial strain. Dosing recommended by the pharmaceutical companies is based on data derived from patients with non severe diseases and may not be appropriate for ICU patients with sepsis, due to variable alteration of elimination or volume distribution. Our objective was to assess the prevalence of insufficient plasma concentrations of beta-lactam antibiotics in ICU septic patients treated with usual dosing and frequency.

Patients and methods: Retrospective analysis of therapeutic drug monitoring (mass spectrometry) between December 2016 and November 2017 for 6 molecules. We selected those assessments performed between 4 and 8 h after IV administration for amoxicillin (AMX) + 3 and 6 h for cloxacillin (CLO), 6 and 12 h for piperacillin (PIP), 6 and 12 h for meropinem (MEM), 8 and 24 h for cefepime (FEP), and 12 and 24 h for ceftriaxone (CRO). EUCAST inferior breakpoint was used for each molecule to approach usual MICs.

Results: 178 samples were assessed, of which 127 were residual concentrations (AMX = 24, CLO = 7, PIP = 62, MEM = 14, FEP = , 7, CRO = 13) in 88 patients. Age = 61 + − 4, SAPS 2 = 50 + − 19, weight = 75 + − 22 kg, body mass index = 28 + − 6, shock = 56%, renal replacement therapy = 17%, LOS in ICU = 15.2 + − 12.8 days, mortality = 25%. 30 127 plasma concentrations (23.6%) were below the EUCAST inferior breakpoint- AMX = 4 (17%), CLO = 1 (14%), PIP = 15 (24.1%), MEM 8 (57.1%), FEP = 1 (14.3%), and CRO = 1 (5.8%). Except for MEM and CRO, plasma concentrations were poorly correlated with the time from infusion to sample. Patients with insufficient concentrations were younger (52 + − 15 versus 64 + − 13, p 0.006) and had longer ICU stays before dosage (14 + − 9 vs 7 + − 6 days, p = 0.0001). They had similar ICU length of stay after dosage (11 + − 13 vs 7 + − 8 days, p = 0.16), sex ratio, SAPS 2 scores, weight, BMI, shock, RRT use, and mortality.

Conclusion: In our ICU population, usual dosing and frequency of beta-lactam antibiotics led to insufficient plasma level concentrations in 24% of cases, particularly for antibiotics with short half-lives and in younger patients. Insufficient concentrations were more frequently observed during late ICU antibiotic treatments.

CO-20 Impact of the Unyvero HPN test in ICU patients with ventilator-associated pneumonia or severe hospital-acquired pneumonia

Nathan Peiffer-Smadja (speaker), Lila Bouadma, Kahina Allouche, Martin Reboul, Philippe Montravers, Jean-François Timsit, Laurence Armand-Lefèvre

Hôpital Bichat - Claude Bernard, Paris, FRANCE
Correspondence: Nathan Peiffer-Smadja -

Annals of Intensive Care 2019, 9(Suppl 1):CO-20

Introduction: Early appropriate antibiotic therapy reduces morbidity and mortality of severe hospital-acquired pneumonia (HAP) and ventilator-associated pneumonia (VAP). However, the emergence of bacterial resistance requires the early use of antibiotics with the narrowest possible spectrum.The Unyvero Hospitalized Pneumonia (HPN, Curetis) test is a multiplex PCR detecting 21 bacteria and 19 resistance genes on respiratory samples within 5 h. We assessed the performance and impact of the Unyvero HPN test in ICU patients.

Patients and methods: In this prospective single-center study, conducted in 3 ICUs, we performed a Unyvero HPN test on bronchoalveolar lavage (BAL) or plugged telescoping catheters (PTC) samples of patients suspected of VAP or HAP with Gram-negative bacilli or clustered Gram-positive cocci on Gram staining. We evaluated the performance (sensitivity and specificity) of the test compared to conventional culture methods and its potential impact on the choice of antibiotic therapy.

Results: We analyzed 71 samples (54 BAL and 10 PTC) from 63 patients (52 males, median age 64 years). A total of 69 84 bacteria were detected by the Unyvero test with a sensitivity of 82% (95% CI, 72–90%) and a specificity of 99% (95% CI 98–100). The performance of the test is shown in the table below. For Pseudomonas aeruginosa, the most frequently isolated pathogen (n = 22), sensitivity and specificity were 100% and 98%, respectively. Two cases of relapses of pneumonia were observed with bacteria detected by the Unyvero test but not initially found with the conventional culture. The 4 extended-spectrum beta-lactamases (ESBL) and 2 carbapenemases (NDM) were all detected by the Unyvero test. The clinical evaluation of 33 episodes of pneumonia showed that the Unyvero test would have led to 22 (67%) therapeutic changes. Of these, 3 corresponded to antibiotic therapy initiations and 19 to modifications in antibiotic treatment including 11 (33%) de-escalations, 5 (15%) adequations or improvement of antibiotic therapy and 3 (10%) less adapted therapies.

Conclusion: The Unyvero HPN test has good performance. Its use could lead to changes in antibiotic therapy that are mainly favorable for patients and limit the selection pressure by reducing the antibiotic spectrum early.



Correct identification

(Unyvero + / Culture +)

False positive

(Unyvero + / Culture −)

Sensitivity (%)

Specificity (%)

Positive predictive value (%)

Negative predictive value (%)

Gram-positive bacteria

Staphylococcus aureus







Streptococcus pneumoniae








Escherichia coli







Enterobacter cloacae complex







Proteus spp.







Klebsiella pneumonia







Klebsiella oxytoca







Serratia marcescens







Morganella morganii







Non-fermenting bacteria

Pseudomonas aeruginosa







Actinetobacter baumanii complex







Strenotrophomonas maltophilia








Legionella pneumophila







Haemophilus influenzae















*Total including all organisms in the panel

CO-21 Medical rhabdomyolysis in critically ill patients- initial presentation and short- and long-term outcomes

Ines Gragueb-Chatti (speaker), Romain Hernu, Marie Simon, Thomas Baudry, Thomas Madelaine, Adeline Grateau, Martin Cour, Laurent Argaud

Hospices Civils de Lyon, Lyon, FRANCE
Correspondence: Ines Gragueb-Chatti -

Annals of Intensive Care 2019, 9(Suppl 1):CO-21

Introduction: Rhabdomyolysis is a common pathology in intensive care unit (ICU), often associated with acute kidney injury (AKI) and high mortality rate. To date, clinical course and prognostic factors of medical rhabdomyolysis in ICU are poorly documented. The aim of the study was to describe characteristics and evolution of critically-ill patients with medical rhabdomyolysis and to identify prognostic factors associated with short and long-term outcomes.

Patients and methods: A retrospective study was conducted between January 2006 and December 2016, in the three medical ICUs of Lyon (France). All adult patients hospitalized with severe medical rhabdomyolysis (creatine kinase [CK] > 5,000 IU L) were included. Demographic features, comorbidities (Charlson score), clinical and laboratory data including AKI (according to KDIGO classification), organ failure support, day-28 outcome and one-year mortality were collected.

Results: A total of 427 patients were included. The mean age was 55 ± 17 years, and sex ratio was 1.8. The most common causes of rhabdomyolysis were immobilization (n = 159, 37%) and sepsis (n = 116, 27%). At admission, SOFA score was 8.8 ± 5.1, and mean CK level reached 16.814 ± 28.616 IU L. During ICU stay, 324 patients (76%) developed AKI and 132 (31%) required renal replacement therapy. ICU, day-28, and one-year mortality were 27, 26 and 32%, respectively. Initial CK level did not significantly differ between patients alive (18.186 ± 30.238 IU L) and dead (12,644 ± 22,617 IU L) at day 28 (p = 0.05). After multivariate analysis, Charlson score and organ failures (cardiovascular, respiratory or neurological) were found to be independently associated with day-28 mortality (p < 0.05). Age, Charlson score and highest SOFA score during ICU stay were independent factors associated with one-year mortality (p < 0.05).

Conclusion: These results emphasize the poor outcome of critically-ill patients with severe medical rhabdomyolysis. In this population, comorbidities and organ failures, but not initial CK level, appear to be major determinants of short and long-term prognosis.

CO-22 Are ambulance resources associated with outcome in out-of-hospital cardiac arrests? Insights from the Paris- Sudden Death Expertise Centre registry

Florence Dumas (speaker)1, Richard Chocron1, Thomas Loeb1, Lionel Lamhaut1, Daniel Jost1, Frédéric Adnet1, Eric Lecarpentier1, Wulfran Bougouin2, Franckie Beganton3, Philippe Juvin4, Eloi Marijon4, Xavier Jouven4, Alain Cariou5

1Emergency medicine, Paris, FRANCE; 2ICU, Paris, FRANCE; 3CEMS, Paris, FRANCE; 4HEGP, Paris, FRANCE; 5Cochin, Paris, FRANCE
Correspondence: Florence Dumas -

Annals of Intensive Care 2019, 9(Suppl 1):CO-22

Introduction: In out-of-hospital cardiac arrest (OHCA), geographic disparities in outcomes may reflect baseline variations in patients’ characteristics but may also result from differences in the number of ambulances providing basic life support (BLS) and advanced life support (ALS). We aimed at assessing the influence of allocated ambulance resources on outcome in OHCA patients in a large urban community.

Patients and methods: From May 2011 to January 2016, we analyzed a prospectively collected Utstein database for all OHCA adults. Cases were geocoded according to 19 neighborhoods and the number of BLS (firefighters performing cardiopulmonary resuscitation and applying automated external defibrillator) and ALS ambulances (medicalized team providing advanced care such as drugs and endotracheal intubation) was collected. We assessed the respective influence of Utstein parameters, socio-economic characteristics and ambulance resources of these neighborhoods using a mixed-effect model with successful return of spontaneous circulation (ROSC) as the primary endpoint and survival at hospital discharge as a secondary endpoint.

Results: During the study period, 8754 non-traumatic OHCA occurred in the Greater Paris area. Overall ROSC rate was 3675 8754 (41.9%) and survival rate at hospital discharge was 788 8754 (9%), ranging from 33% to 51.1% and from 4.4% to 14.5% respectively, according to neighborhoods (p < 0.001). Patients’ and socio-demographics’ characteristics significantly differed between neighborhoods (p for trend < 0.001). After adjustment, a higher density of ambulances was associated with successful ROSC (respectively aOR = 1.31 (1.14–1.51) + p < 0.001 for ALS ambulances > 1.5 per neighborhood and aOR = 1.21 (1.04–1.41) + p = 0.01 for BLS ambulances > 4 per neighborhood). Regarding survival at discharge, only the number of ALS ambulances > 1.5 per neighborhood was significant (aOR = 1.30 (1.06–1.59) p = 0.01).

Conclusion: In this large urban population-based study of out-of-hospital cardiac arrests patients, we observed that allocated resources of Emergency Medical Service (EMS) are associated with outcome, suggesting that improving healthcare organization may attenuate disparities in prognosis.

CO-23 Assessment of severe eosinopenia for the diagnosis of infection in the Emergency Department and comparison with other biomarkers

Jérémy Rosman (speaker)1, Simon Hainguerlot2, Aurélien Cordonnier3, Karelle Staffe3, Olivier Gallon4, Thomas Beuvelet5, Xavier Fontaine 2, Philippe Mateu1

1Service de Médecine Intensive Réanimation, Centre Hospitalier de Charleville-Mézières, Charleville-Mézières, FRANCE; 2Service d’Accueil des Urgences, Centre Hospitalier de Charleville-Mézières, Charleville-Mézières, FRANCE; 3Département de l’Information Médicale, Centre Hospitalier de Charleville-Mézières, Charleville-Mézières, FRANCE; 4Service des Maladies Infectieuses, Centre Hospitalier de Charleville-Mézières, Charleville-Mézières, FRANCE; 5Laboratoire de Biologie, Centre Hospitalier de Charleville-Mézières, Charleville-Mézières, FRANCE
Correspondence: Jérémy Rosman -

Annals of Intensive Care 2019, 9(Suppl 1):CO-23

Introduction: Infectious diseases represent a frequent cause of consultation in the Emergency Department (ED). Biomarkers may help clinicians to diagnose infection from other causes. Severe eosinopenia may reflect level of systemic inflammation. The aim of our study was to evaluate severe eosinopenia, as defined a value < 0.01 G L (< 10 mm3), as a marker of infection.

Patients and methods: Patients admitted in the ED with at least one complete blood count were included. Biomarkers of infection were evaluated by multivariable logistic regression.

Results: Among 15.638 patients screened, 11.520 patients were finally included, 14% were infected. Severe eosinopenia was present in 49% of infected patients and 23% in non-infected patients (p < 0.001). Severe eosinopenia had a modest accuracy for diagnosis of infection with 49% sensitivity, 77% specificity, 26% positive predictive value, 90% negative predictive value, 2.1 positive likelihood ratio and 0.47 négative likelihood ratio. Accuracy was lower than other biomarkers as C-reactive protein, fibrinogen, procalcitonin and (neutrophilic) leukocytosis. In mutivariable analysis, severe eosinopenia was an independant factor of infection [adjusted odds ratio 1.3, (95% confidence interval 1.1–1.5), but lower than other biomarkers).

Conclusion: Severe eosinopenia, a reflect of systemic inflammation, was an independent factor associated with infection, but is not better than other usual biomarkers.

CO-24 QSOFA as predictor of mortality and prolonged ICU admission in Emergency Department patients with suspected infection

Emmanuel Canet (speaker)1, David Mcd Taylor2, Richard Khor2, Vivek Krishnan2, RinaldoBellomo2

1Intensive Care Unit, Nantes, FRANCE; 2Austin Health Hospital, Melbourne, AUSTRALIA
Correspondence: Emmanuel Canet -

Annals of Intensive Care 2019, 9(Suppl 1):CO-24

Introduction: We assessed the quick Sequential Organ Failure Assessment (qSOFA) score as a predictor of in-hospital mortality or prolonged ICU stay in Emergency Department (ED) patients with suspected infection.

Patients and methods: We measured qSOFA in a cohort of 11,205 ED patients with suspected infection. The primary outcome was in-hospital mortality and or ICU stay ≥ 3 days.

Results: The qSOFA score was positive in 2,429 (21.7%) patients. In-hospital mortality, and in-hospital mortality or ICU stay ≥ 3 days were 12.8% and 17.2% respectively for qSOFA positive patients vs 2.2% and 4.2% for qSOFA negative patients (p < 0.0001). For the prediction of in-hospital mortality, a positive qSOFA had a positive predictive value (PPV) of 13% (95% CI, 11–14) and a negative predictive value (NPV) of 98% (95% CI, 97–98). For the prediction of in-hospital mortality or ICU stay ≥ 3 days, the PPV and NPV of a positive qSOFA were 17% (95% CI, 16–19) and 96 (95% CI, 95–96), respectively.

Conclusion: Among ED patients with suspected infection, a positive qSOFA identified those at much greater risk of mortality and longer ICU stay.

CO-25 Prevalence, management and prognosis of hypoxemia among obese patients in the ICU- insights from the SPECTRUM study

Florence Boissier (speaker)1, David Grimaldi2, Sami Hraiech3, Philippe Michel4, Jean-BaptisteLascarrou5, Tai Pham6, Jean-Christophe Richard7, Arnaud W Thille8, Stephan Ehrmann9, Nadia Aissaoui10, Jean-Claude Lacherade11, Gregoire Muller12

1Service de réanimation médicale CHU, Poitiers, FRANCE; 2Intensive care unit, Erasme Hospital, Bruxelles, BELGIUM; 3réanimation des détresses respiratoires et infections sévères, Hopital Nord APHM, Marseille, FRANCE; 4Service de réanimation, CH René-Dubos, Pontoise, FRANCE; 5Médecine Intensive Réanimation CHU, Nantes, FRANCE; 6Critical Care Medicine, St Michael’s Hospital, Toronto, CANADA; 7Réanimation médicale, hôpital de la Croix Rousse, Lyon, FRANCE; 8Réanimation médicale, CHU, Poitiers, FRANCE; 9Médecine Intensive Réanimation, CHRU, Tours, FRANCE; 10Réanimation médicale, HEGP, APHP, Paris, FRANCE; 11Médecine Intensive - Réanimation, CHD Vendée, site de la Roche sur Yon, La Roche-Sur-Yon, FRANCE; 12Service de Médecine Intensive Réanimation, Orléans, FRANCE
Correspondence: Florence Boissier -

Annals of Intensive Care 2019, 9(Suppl 1):CO-25

Introduction: Using the SPECTRUM study, we aimed to evaluate whether hypoxemia in ICU among obese patients had different causes and management compared to non-obese patients.

Patients and methods: Subgroup analysis of a prevalence-point-day conducted in 117 French-speaking ICU aiming to report the patterns and outcomes of hypoxemic patients (defined by P F < 300 mmHg). Obesity was defined as a body mass index (BMI) > 30 kg m2.

Results: Among 1571 patients hospitalized in ICUs the day of the study with BMI data, 428 were obese (27%). 247 of them were hypoxemic (57.7%) as compared with 597 1143 (52.2%) non-obese patients (p 0.05). They exhibited more frequent obesity-hypoventilation syndrome (23% versus 1.5% p < 0.001) and sleep apnea (19.4% versus 3.5% p < 0.001), less frequent non-obstructive chronic respiratory disease (2% versus 8.6% p < 0.001) and interstitial pulmonary disease (1.2% versus 4.4% p 0.02) and had more frequently home non-invasive ventilation (NIV) (5.7 versus 2.9% p 0.05). Hypoxemia was mild in 52%, moderate in 39% and severe in 9%, similar to non-obese patients. ARDS criteria were fulfilled in 21.5% (versus 20.6% in non-obese patients). They required high flow oxygen, NIV and invasive ventilation respectively in 5.7, 6.9 and 58.7% of cases, which was not different from non-obese patients. PEEP was higher (7 (IQR 5–10) versus 6 (IQR 5–8) cmH2O, p 0.03). Tidal volume in intubated patients was 7 ml kg (IQR 6–8.4) in obese patients versus 6.8 (6.1–7.7) (p 0.06). Plateau pressure was not different between the 2 groups (24 versus 22 cm H2O), as well as PaO2, PaCO2 and pH. Diagnosis of atelectasis was frequent in the 2 groups (23% versus 25.7%). There was no difference in the use of prone position (5.2 versus 4.3%, p 0.66). ICU mortality of obese patients was 20.6% versus 27.4% (p 0.04). Multivariate Cox model confirmed a negative independent association between obesity and ICU mortality.

Conclusion: Obese patients represent more than 25% of the patients hospitalized in ICU the day of the study. Hypoxemia seems more frequent in obese patients. Respiratory support and hypoxemia severity were similar between obese and non-obese patients. PEEP was higher in obese patients. Obesity was associated with a lower mortality.

CO-26 Do all immunocompromised patients with ARF respond equally to oxygenation strategy?

Virginie Lemiale (speaker)1, Audrey De Jong2, Guillaume Dumas3, Jordi Rello4, Michael Darmon5, Philip Bauer6, Andry Van de Louw 7, Julien Mayaux8, Ignacio Martin-Loesche9, Djamel Mokart10, Peter Schellongowski11, Sangeeta Mehta12, Achille Kouachet13, Frédéric Pene14, Peter Pikkers15, Gaston Burghi16, Massimo Antonelli17, Fabrice Bruneel18, Andreas Barratt-Due19, Miia Valkonen20, Victoria Metaxa21, Anders Perner22, Julien Dessajan5

1APHP Saint Louis, Paris, FRANCE; 2Anesthesiology and Intensive Care + Anesthesia and Critical Care Department B, Montpeller, FRANCE; 3Medical ICU APHP Saint Antoine, Paris, FRANCE; 4Vall d’Hebron Hospital, Barcelone, SPAIN; 5Medical ICU APHP Saint Louis, Paris, FRANCE; 6Mayo Clinic, Rochester, UNITED STATES; 7Penn State College of Medicine, Hershey, UNITED STATES; 8Medical ICU APHP Pitié Salpétriere, Paris, FRANCE; 9St James’s University Hospital, Dublin, IRELAND; 10Institut Paoli Calmettes, Marseilles, FRANCE; 11Medical University of Vienna, Vienne, AUSTRIA; 12Mount Sinai Hospital, Toronto, CANADA; 13Medical ICU CHU, Angers, FRANCE; 14Medical ICU APHP Cochin, Paris, FRANCE; 15Radboud University Medical Center, Nijmegen, THE NETHERLANDS; 16Montevideo University hospital, Montevideo, URUGUAY; 17Catholic University of Rome- A. Gemelli University Hospital, Roma, ITALY; 18Medical ICU A Mignot, Versailles, FRANCE; 19Oslo University Hospital, Oslo, NORVÈGE; 20Helsinki University Hospital, Helsinki, FINLANDE; 21King’s College Hospital, Londres, UNITED-KINGDOM; 22Rigshospitalet, University of Copenhagen, Copenaghe, DENMARK
Correspondence: Virginie Lemiale -

Annals of Intensive Care 2019, 9(Suppl 1):CO-26

Introduction: In immunocompromised patient with acute respiratory failure (ARF), mortality remains high. First oxygenation strategy with non-invasive ventilation or high flow nasal oxygen has not been clearly demonstrated according to the patient status. We assessed assess outcomes in patients with hematological malignancies and acute respiratory failure (ARF) according to the initial ventilation strategy, radiological lesion and ARF etiology.

Patients and methods: All patients with ARF included in MINIMAX, TRIALOH and IVNICTUS studies and who were not intubated at admission, were included in this post hoc analysis of three multicenter studies. An external validation was then performed in the EFRAIM cohort.

Results: 847 patients admitted with ARF were included. At ICU admission, radiological pattern was subnormal (n = 75, 9%), focal lesion (n = 159 + 20%), diffuse alveolar lesion (n = 444 + 55%) or interstitial lesion (n = 127 + 16%). Diagnosis of ARF was mainly related to bacterial or viral pneumonia (335, 40%). Diagnosis of ARF was not found for 147 (17%) patients. First oxygenation strategy was standard oxygen (n = 310), NIV (n = 400), HFNO (n = 65) and NIV + HFNC (n = 72). Bilateral alveolar pattern (OR = 2.05 (1.00–4.22), p = 0.05) was independently associated with day-28 mortality after adjusting on NIV use within the 2 first days (OR = 1.76 (1.14–2.72), p = 0.01), SOFA score without respiratory item at ICU admission (OR = 1.16 (1.09–1.24), p < 0.001) and PaO2 FiO2 < 100 at ICU admission (OR = 1.69 (1.16–2.50), p = 0.007). Opportunistic infection (OR = 2.16 (1.14–4.09), p = 0.01) and no identified cause (OR = 1.97 (1.08–3.58)) were independently associated with day-28 mortality after taking into account NIV use within the 2 first days (OR = 1.86 (1.22–2.83), p = 0.0004), SOFA score without respiratory item at ICU admission (OR = 1.17 (1.10–1.25), p < 0.001) and PaO2 FiO2 < 100 at ICU admission (OR = 1.69 (1.16–2.50), p = 0.01). The analyses performed in the validation cohort confirmed the results found in the initial cohort.

Conclusion: NIV use, opportunistic or no identified diagnosis and bilateral alveolar radiological pattern were associated with mortality, after taking into account the severity of ARF disease using SOFA score and PaO2 FiO2 ratio.

CO-27 High-flow oxygen therapy vs non invasive ventilation- a prospective cross-over physiological study of alveolar recruitment in acute respiratory failure

Elise Artaud-Macari (speaker), Michael Bubenheim, Gurvan Le Bouar, Dorothée Carpentier, Steven Grangé, Déborah Boyer, Gaëtan Béduneau , Benoît Misset, Antoine Cuvelier, Fabienne Tamion, Christophe Girault

Pulmonary, Respiratory Intensive Care and Thoracic oncology Department, Rouen University Hospital, UPRES EA-3830 I, Rouen, FRANCE
Correspondence: Elise Artaud-Macari -

Annals of Intensive Care 2019, 9(Suppl 1):CO-27

Introduction: High-flow oxygen therapy (HFNC) has shown a benefit for the prognosis of patients with hypoxemic acute respiratory failure (ARF), while noninvasive ventilation (NIV) remains debated in this indication. We evaluated the effect of HFNC on alveolar recruitment and lung volumes in hypoxemic ARF compared to NIV and facial mask (FM).

Patients and methods: A prospective cross-over physiological study was conducted. Eligible patients had to present a hypoxemic ARF due to pneumonia requiring HFNC and or NIV according to ICU physician. Cardiogenic pulmonary oedema and underlying respiratory disease were excluded. Each patient was investigated with the Pulmovista® (Dräger, Lübeck, Germany) device and underwent 15 min periods of HFNC or NIV in a randomized order, interspersed with 15 min periods of FM used as reference. The primary endpoint was the comparison of global and regional end-expiratory electrical lung impedance (EELI) between NIV and HFNC. Secondary endpoints were the comparison, between the 3 techniques, of lung volumes (global and regional tidal variations (TV), respiratory parameters, hemodynamic tolerance, dyspnea and comfort.

Results: NIV and HFNC significantly increased the global EELI compared with FM (2056 [1070 + 2825] vs. 4083 [2928 + 5134], p = 0.001 and 1448 [1028 + 3542] vs 2921 [1706 + 4850], p = 0.0001, respectively). No global EELI difference was found between NIV and HFNC (4083 [2928 + 5134] vs 2921 [1706 + 4850], p = 0.4) (fig 1.1). Global and regional TV increased under NIV compared to HFNC (p < 0.05) or FM (p < 0.05), while HFNC did not modify TV over FM. NIV significantly improved the SpO2 FiO2 ratio compared to HFNC (p = 0.001) (fig 1.2). HFNC significantly reduced respiratory rate vs FM (p = 0.04) but not NIV. No difference was found for dyspnea score between the 3 techniques. Patient comfort was similar between HFNC and FM but decreased with NIV.

Conclusion: This study demonstrates a similar benefit of HFNC and NIV on alveolar recruitment with the settings used, as compared to FM. By contrast to HFNC and despite a better oxygenation, NIV also increases lung volumes which may contribute to its potentially deleterious effect during hypoxemic ARF leading to the recent concept of Patient Self Inflicted Lung Injury or P-SILI (Brochard L et al. AJRCCM 2017 + 195-438-42).

CO-28 Lung and chest wall mechanics of patients admitted to Intensive Care Unit

Elise Yvin (speaker), Pierre-Yves Olivier, Lise Piquilloud, Satar Mortaza, Alain Mercat, François Beloncle

Département de Médecine Intensive-Réanimation et Médecine Hyperbare, CHU d’Angers, Angers, FRANCE
Correspondence: Elise Yvin -

Annals of Intensive Care 2019, 9(Suppl 1):CO-28

Introduction: Data concerning respiratory mechanics of intubated patients separately analyzing the lung and the chest wall are scare. This study aims at describing respiratory mechanics (respiratory system compliance (CRS), lung compliance (CL), chest wall compliance (CCW) and end-expiratory lung volume (EELV)) of all intubated patients admitted to Intensive Care Unit (ICU).

Patients and methods: This is an interim analysis of a prospective single-center study. All patients admitted to ICU, intubated and paralyzed as part of the routine care, without contraindication to esophageal pressure measurement were included. Respiratory mechanics measurements were performed at their admission to ICU with standardized ventilator settings. CL and CCW were measured using esophageal pressure measurement (Nutrivent catheter, Sidam ®, San Giacomo Roncole, Italy). EELV was measured, at a positive end-expiratory pressure of 5 cmH2O, using the nitrogen washout-washin technique (CRF inview, GE-Healthcare ®, Madison, WI, USA). Results are presented as median [IQR]. Correlations were analyzed using Spearman test.

Results: 30 patients were included in the study. ICU admission diagnoses were sepsis other than pulmonary (9 patients, 30%), pneumonia (5 patients, 17%), cardiac arrest (6 patients, 20%), acute cardiac failure (5 patients, 16%) or neurologic disorder (5 patients, 17%). Ten patients (33%) met Acute Respiratory Distress Syndrome criteria at the day of respiratory mechanics measurement. Distributions of CRS, CL and CCW are presented in figure 1. There was a wide distribution of CCW values. CCW impairment (defined as CCW < 100 mL cmH2O) was observed in 7 patients (23%). No clinical history of abdominal hypertension was found in these patients. CCW was not correlated with the body mass index (p = 0.21 + r = 0.24). CL was well correlated with EELV (p < 0.001 + r = 0.80).

Conclusion: This study shows that, in a non-selected population of intubated patients admitted to ICU for various reasons, distributions of CL and CCW are wide. CCW impairment is not rare and cannot be predicted by the patient’s medical history or morphology. The good correlation between CL and EELV is consistent with data establishing that CL represents the aerated lung volume.

CO-29 Incidence of ventilatory acquired pneumonia in PICU- a one-year prospective multicenter data-base (the INCIPAVE study)

Stephane Dauger (speaker)1, Yves Gallien2, Josephus Van Gestel3, Michael Levy1, Astrid Botte4, Fleur Cour-Andlauer5, Camille Guillot 6, Fabrice Lesage7, Jérôme Rambaud8, Catherine Vanbaelen9, Maryline Chomton1, Matthieu Resche-Rigon10

1Réanimation pédiatrique, hôpital Robert Debré, Paris, FRANCE; 2Hôpital Saint-Louis-Sce de Biostatistique et Information Médicale, Paris, FRANCE; 3PICU, Utrecht, THE NETHERLANDS; 4Réanimation pédiatrique, Hôpital Pellegrin, Bordeaux, FRANCE; 5Réanimation pédiatrique, HFME, Lyon, FRANCE; 6Réanimation pédiatrique, Hôpital Jeanne de Flandres, Lille, FRANCE; 7Réanimation pédiatrique, Hôpital Necker Enfants malades, Paris, FRANCE; 8Réanimation pédiatrique, Hôpital Armand Trousseau, Paris, FRANCE; 9Réanimation pédiatrique, Hôpital des enfants, Toulouse, FRANCE; 10Service de Biostatistique et Information Médicale, Hôpital Saint-Louis, Paris, FRANCE
Correspondence: Stephane Dauger -

Annals of Intensive Care 2019, 9(Suppl 1):CO-29

Introduction: Ventilatory Acquired Pneumonia (VAP) is one of the main nosocomial infection in adult ICU. Only one prospective multicenter study performed during six months in 16 PICUs of the US has prospectively described pediatric VAP. We design the INCIPAVE study to report the occurrence of VAP in european PICUs, focusing on patients, risk factors (RF), micro-organisms involved, diagnostic tools and antibiotics used. The first aim of this study was to calculate the incidence of VAP.

Patients and methods: Multicenter prospective cohort study from 03/04/2017 to 03/04/2018 including all patients mechanically ventilated (MV) at least once in eight PICUs, one in the Netherlands and seven in France. VAP was defined using the 2015 CDC criteria, applied during PICU stays, excluding the 48 h preceeding and following PICU. Patients were described on admission and main RF ever tested in the medical litterature were daily included by a pediatric intensivist of each PICU in an electronic database on a securized dedicated website. The Ethics Committee of the French Society of Intensive Care approved the study, which has been declared to the CNIL and recorded on All parents were individually informed by a dedicated sheet. Descriptive data are reported as number (%) or medians [first-third quartiles]. Incidence was calculated as the number of VAP for 1000 days of MV.

Results: These results are based on declarative information reported in the INCIPAVE database during its first opening on September 2018, before cleaning. During one year, 2047 episodes of MV were included in 1856 patients (26.5 months [6 + 92], 12kgs [6.4 + 23.22], 56% of males). Main reasons for admission to PICU were post-operative care (35.1%) and respiratory (23.7%), neurological (17.1%) or circulatory (12.2%) failures. PIM-2 score was 2.8% [1.3 + 7.5] with an observed mortality of 10.9%. PELOD-2 score on day-1 was 5 [4.7]. Patients were ventilated via uncuffed tubes in 19.6% of cases (7.8% of tracheostomy). A total of 158 VAP was declared during 11685 days of MV. The incidence of VAP was 13.5 1000 days of MV (IC95% - [11.4 + 15.6]).

Conclusion: This incidence of VAP in INCIPAVE study is higher than the incidence reported recently by Gupta (7.0 1000 days of MV) (1). New analyses are planned after cleaning of the database, including a special reading of each case of VAP.

CO-30 Severe sepsis and septic shock in hematological pediatric patients admitted to picu - a bicentric retrospective study from 2011 to 2017

Aurélia Alimi (speaker)1, Jérôme Rambaud2, Julie Sommet1, Sandrine Jean2, Michael Levy1, Maryline Chomton1, Fleur Le Bourgeois1, Stephane Dauger1

1Hôpital Robert Debré APHP, Paris, FRANCE; 2Hôpital Armand Trousseau APHP, Paris, FRANCE
Correspondence: Aurélia Alimi -

Annals of Intensive Care 2019, 9(Suppl 1):CO-30

Introduction: Pediatric sepsis remains a burdensome public health problem, especially in patients with immuno-deficiency. Adherence to Survival Sepsis Campaign (SSC) would be associated with lower mortality. However, this correlation had never been evaluated in children with hematologic malignancies.

Patients and methods: Retrospective bicentric cohort study of children with identified hematologic malignancy or hematopoietic stem cell transplantation and requiring intensive care for severe sepsis between January 2011 and August 2017. Detailed description of sepsis resuscitation just before and within six hours after the diagnosis of sepsis. Evaluation of adherence to SSC recommandations, before and after admission to Pediatric Intensive Care Unit (PICU). Accurate analysis of microbiologic etiologies. Assessment of mortality 6 months after the sepsis.

Results: 65 patients were included from the 78 who have been screened. Before admission in ICU - 38% of them did not receive oxygen, 47% of patients showed low blood pressure for age without continuous infusion of inotropes, and lactic acid was measured in few patients (14%). Within 6 h of PICU admission, 28% did not receive oxygen and only 24% of patients with continuous infusion of inotropes were ventilated. Lactic acid and central venous oxygen saturation measurements were obtained in only one third of patients. An infectious organism was isolated in 51% of patients. The most common primary site of infection was central venous catheter (70%) and 5 patients (13%) showed a fungic infection. The mortality within 6 months was evaluated at 18.5%.

Discussion: Adherence to SCC recommandations for hematologic children with sepsis appeared inadequate. However, mortality was lower in our study than in other previous data.

Conclusion: Higher adherence to SSC recommandations for children with hematologic malignancies and better sensitization of medical and nurse staff would be potentially associated with much better prognosis.

CO-31 High-flow nasal canula (HFNC) in infant hospitalized with moderate bronchiolitis- results of a multicenter open-label RCT in pediatric health care (Bronchopti study)

Philippe Durand (speaker)

Assistance Publique Hôpitaux de Paris, Le Kremlin-Bicêtre, FRANCE
Correspondence: Philippe Durand -

Annals of Intensive Care 2019, 9(Suppl 1):CO-31

Introduction: HFNC has emerged as a promising method to provide respiratory support in bronchiolitis. Only two RCT had evaluated HFNC in less severe bronchiolitis admitted in general wards but failed to demonstrate a reduction in the length of oxygen therapy or in the proportion of patient transferred in PICU (Kepreotes 2017, Franklin 2018). Therefore, we performed a superiority trial to test the hypothesis that HFNC could reduce the proportion of treatment failure requiring non-invasive ventilation among infants in these setting.

Patients and methods: Inclusion- first episode of bronchiolitis in infant (less 6 months-old), SpO2 in room air < 95%, m-WCAS score > or = 2 and < or = 5. Exclusion - urgent need for NIV, mWCAS > 5 or 6, lack of consent. Randomization - standard oxygen therapy (up to 2 L min)(control) or HFNC (3 l kg min)(experimental). Cross-over wasn’t allowed. Failure criteria- FiO2 > or = 40% or nasal flow oxygen > 2L min to maintain SpO2 target, increasing m-WCAS score or > 5–6, refractory apnea and or increasing PaCO2 at H6. Primary endpoint- proportion of patient in treatment failure requiring non-invasive within 7-days following randomization. Secondary endpoint- percentage admitted in ICU, assessment of short term respiratory status, general ward unit LOS, oxygen and nutritional - support free days.

Results: During the 6 months study period (2016–2017), 2630 patients admitted for bronchiolitis into the 17 PED’s network were screened of whom 271 underwent randomization (268 in the intention-to-treat analysis) (table 1). HFNC didn’t improved the primary outcome. Failure occurred in 21 of 133 patients (15%) in the HFNC group and 27 of 135 (20%) in the control group (OR IC 95% + 0.75 [0.40 + 1.40] + p = .36). HFNC didn’t reduced the risk of admission in ICU (21 (15%) in HFNC group versus 26 (19%) in control (OR IC 95% + 0.78 [0.41 + 1.41] + p = .45). Any patient was intubated. The main reason for treatment failure was the worsening of mWCAS score. Mean LOS on general ward unit was lower in the control group (3.8 + - 2.7 days versus 4.4 + -2.4 days + p = 0.04). Short-term assessment of respiratory status didn’t shown difference except for mWCAS score and RR in favor HFNC. Three pneumothorax was reported in HFNC group.

Conclusion: There was neither evidence of lower rate of non-invasive ventilation support among patients receiving HFNC therapy nor difference in the rate of ICU admission.


HFNC group (N = 133)

Standard group (N = 135)


Age, Mean (SD)-d

68 ± 48

65 ± 46


Weight, Mean (SD)-kg

5,1 ± 1,5

4,9 ± 1


Female sex-no.(%)



Gestational age, Mean (SD)-weeks

38 ± 2

38 ± 2


-proportion of premature, no.(%)

16 (12%)

16 (11%)


Duration of symptoms before randomization, Mean (SD)-days

3,3 ± 2,1

3,1 ± 2,2


Temperature, Mean (SD) -°C

37,2 ± 0,6

37,2 ± 0,5


RR, Mean (SD) -bpm

53 ± 13

55 ± 14


Heart rate, Mean (SD)-bpm

156 ± 18

154 ± 18


SpO2, Mean (SD)-% in room air

90 ± 3

90 ± 3


m-WSCA score, Mean (SD)

3,3 ± 0,8

3,1 ± 0,7


PtCO2, Mean (SD)-mmHg

50 ± 11

50 ± 10


pH, Mean (SD)

7,34 ± 0,07

7,33 ± 0,05


Viral cause-no./total no.(%)

Number tested, no




RVS status, no. (%)

85 (82%)

87 (82%)


CO-32 Prospective bicentric observationnal study of non-invasive ventilation in pediatric acute chest syndrome - the nivipacs cohort

Charlotte Idier (speaker)1, Pierre-Louis Leger2, Julie Sommet3, Julia Guilbert1, Fleur Le Bourgeois3, Jérôme Naudin3, Michael Levy 3, Stephane Dauger3

1CHU Gatien de Clocheville, Tours, FRANCE; 2Hopital Armand Trousseau, AP-HP, Paris, FRANCE; 3Hôpital Robert Debré APHP, Paris, FRANCE
Correspondence: Charlotte Idier -

Annals of Intensive Care 2019, 9(Suppl 1):CO-32

Introduction: Acute chest syndrome (ACS) is one of the most frequent and severe manifestations of sickle cell disease in childhood. The physiopathological mechanisms of ACS are complex and resulted in severe abnormalities of ventilation perfusion ratio. Our hypothesis is that non-invasive ventilation (NIV) could be an important component of ACS treatment. The main objective of this study is to describe the tolerance of NIV in the pediatric ACS.

Patients and methods: Bicentric prospective study from 01 09 2016 to 01 09 2017, including all sickle cell patients admitted to pediatric intensive care unit (PICU) for ACS. The main composite objective was to describe the tolerance of NIV by identifying the following four components- i) frequency of use, ii) effects on work of breathing evaluated by the clinical respiratory score (CRS), iii) circumstances of NIV weaning, iv) child’s comfort scored by the COMFORT scale, and the occurrence of complications.

Results: From the thirty-seven children included, 30 (81%) were ventilated with NIV for 34 [22 + 53] hours for a duration of stay in PICU of 3.0 [2.0 + 4.1] days. All children were discontinuously ventilated in Pressure Support (PS) mode. The CRS dropped rapidly between H0 and H4- from 6 [4 + 7] at H0 to 4 [3 + 6] at H4 (p = 0.0006), without change in the SpO2 FIO2 ratio during the same time (385 [320 + 463] versus 398 [331 + 438], p = 0.25). The COMFORT score was the same with or without NIV. We did not report any NIV complication.

Conclusion: The majority of children admitted to PICU with an ACS are broken down into NIV in PS mode. NIV was very well tolerated and was able to decrease respiratory distress in the first hours without modifying oxygen requirement.

CO-33 Association between early endotracheal intubation and ICU mortality in septic shock - a prospective multicentric observational study

Sophie Jacquier (speaker)1, Agathe Delbove2, Cédric Darreau3, Marjorie Saint-Martin3, Frédéric Martino5, Jean-François Hamel6, Mai-Anh Nay7, Nicolas Terzi8, Geoffrey Ledoux9, Ferran Roche-Campo10, Laurent Camous11, Frédéric Pene12, Thibault Balzer13, François Bagate14, Julien Lorber15, Pierre Bouju16, Clémence Marois17, René Robert18, Stéphane Gaudry19, Morgane Commereuc20, Matthieu Debarre21, Nicolas Chudeau2, Pierre Labroca22

1CHRU, Tours, FRANCE; 2CHU, Nantes, FRANCE; 3CH, Le Mans, FRANCE; 5CHU Guadeloupe, Abymes, FRANCE; 6CHU, Angers, FRANCE; 7CH, Orléans, FRANCE; 8CHU, Grenoble, FRANCE; 9CHRU, Lille, FRANCE; 10Hospital of Santa Creu i Sant Pau, Barcelone, SPAIN; 11CHU Bicetre, Paris, FRANCE; 12CHU Cochin, Paris, FRANCE; 13CHU, Brest, FRANCE; 14CHU Mondor, Paris, FRANCE; 15CH, La Roche-Sur-Yon, FRANCE; 16CH, Lorient, FRANCE; 17CHU La Pitié Salpétrière, Paris, FRANCE; 18CHU, Poitiers, FRANCE; 19CHU Colombes, Paris, FRANCE; 20CHU HEGP, Paris, FRANCE; 21CH, Saint-Brieuc, FRANCE; 22CHU, Nancy, FRANCE
Correspondence: Sophie Jacquier -

Annals of Intensive Care 2019, 9(Suppl 1):CO-33

Introduction: By contrast to neurological and respiratory failure, the place for endotracheal intubation and mechanical ventilation is not detailed in guidelines on septic shock, evidencing a lack of dedicated studies. The objective of this study was to assess the association between early endotracheal intubation (before H8 following vasopressor initiation - H0) and outcome (ICU survival) in septic shock patients, taking into account presence or absence of classical intubation criteria (i.e. neurologic and or respiratory failure).

Patients and methods: This prospective multicenter observational study was conducted from May 2016 to October 2017 in 32 ICUs (France and Spain). All successive adult patients suffering from septic shock and admitted in participating ICUs were considered. Criteria defining three groups of patients were defined a priori according to the potential neurologic respiratory motivation of intubation between H0 and H8—1) “classical indication group” (CIG) in presence of straightforward indication for endotracheal intubation, 2) “intermediate group”(IG) in presence of moderate impairment, 3) and non-classical motivation group (NCG) in absence of significant impairment. Then, a propensity score was constructed to establish the probability of being intubated.

Results: Eight hundred and fifty-nine patients were recruited. Two hundred and twenty-six patients were sorted into the CIG, 329 into the IG and 190 into the NCG. Early intubation was performed in 51% of the CIG, 22% of IG and 9% in NCG. Multivariate analysis showed that groups (CIG > IG > NCG), higher SAPS II and early intubation (OR = 1.65 [1.07–1.56]) vs. no early intubation, were independently associated with higher ICU mortality. The propensity score was constructed entering 586 patients. An increased risk of death was observed throughout the range of the propensity score in early intubated patients vs. not early intubated patients, all the more that the propensity score was increased (p < 0.001), see figure *In a multivariate analysis, delayed intubation showed no impact on mortality vs. early intubation with OR = 0.99 [0.52–1.87] + p = 0.97 for patients intubated between H8-H24 vs. patients intubated before H8 and OR = 1.66 [0.76–3.65] + p = 0.21 for patients intubated between H24 and H72.

Conclusion: An association between early intubation and higher mortality was observed both in a multivariate model and using a propensity score in septic shock patients. This study opens the way for a prospective interventional trial.

CO-34 VA-ECMO to rescue refractory septic shock with severe myocardial dysfunction

Nicolas Bréchot (speaker)1, David Hajage1, Julien Demiselle2, Antoine Kimmoun3, Santi Montero1, Matthieu Schmidt1, Guillaume Hékimian1, Guillaume Lebreton1, Charles-Edouard Luyt1, Elie Zogheib4, Erwan Flecher5, Daniel Brodie6, Bruno Levy3, Pierre Asfar2, Alain Combes1

1La Pitié-Salpêtrière University Hospital, Paris, FRANCE; 2Angers University Hospital, Angers, FRANCE; 3Nancy-Barbois University Hospital, Vandoeuvre-Lès-Nancy, FRANCE; 4Amiens University Hospital, Amiens, FRANCE; 5Rennes University Hospital, Rennes, FRANCE; 6Columbia University Medical Center, New-York, UNITED STATES
Correspondence: Nicolas Bréchot -

Annals of Intensive Care 2019, 9(Suppl 1):CO-34

Introduction: Some patients with septic shock may develop very severe myocardial dysfunction refractory to medical treatments. Several cohorts and case-report studies reported salvage therapy with veno-arterial extracorporeal membrane oxygenation (VA-ECMO) in those patients. The aim of this study was to assess the usefulness of VA-ECMO as rescue therapy during refractory myocardial dysfunction associated with septic shock.

Patients and methods: In this multicenter international retrospective study, patients from 4 ECMO centers implanted with VA-ECMO as a rescue therapy from 2008 to 2018 for refractory septic shock with severe myocardial dysfunction (cardiac index below 3 L min m2 and or left ventricular ejection fraction below 35%) were included. As controls, 228 patients exhibiting severe myocardial dysfunction during the same period of time but who did not receive ECMO were isolated from 3 large septic shock databases, and were included the first day they met criteria for severe myocardial dysfunction. ECMO and non-ECMO patients outcomes were compared after matching for disease severity. The primary end point was mortality at 90 days.

Results: (mean ± SD and n (%) are shown) Eighty-two patients rescued with VA-ECMO were included. Patients exhibited extremely severe myocardial dysfunction (mean cardiac index at 1.54 ± 0.54 L min m2 and left ventricular ejection fraction 28.1 ± 5.7%), despite very high doses of catecholamines (mean inotropic score 279 ± 247 µg kg min). They also had profound lactic acidemia (pH 7.13 ± 0.15 and lactatemia 8.9 ± 4.4 mmol/L) and severe multiple organ failure (SOFA score 16.6 ± 2.9 and SAPS-II score 78.3 ± 16.1). Seventy-eight of them could be matched with non-ECMO patients based on a propensity score for disease severity (lactatemia, cardiac index and SOFA score). Mortality at 90 days was dramatically reduced in matched ECMO patients compared to controls, (41% vs. 87%, p < 0.0001). In a sensitivity analysis, all patients with cardiac index monitoring were weighted using matching weight ponderation on inotropic score, lactatemia, SOFA score, age and cardiac index at inclusion. Mortality at 90 days remained significantly reduced in weighted ECMO patients. Finally, survivors in the ECMO group reported acceptable SF-36 evaluated health-related quality of life in the months following ICU discharge.

Conclusion: In this retrospective propensity-matched study, rescue therapy with VA-ECMO was associated with a dramatic reduction in 90 days mortality for patients with refractory septic shock and severe myocardial dysfunction.

CO-35 Persistent fluid responsiveness is infrequent after fluid expansion- a multicenter, prospective physiological study

Hélène Beringuer (speaker)1, Antoine Kimmoun2, Pierre-Eric Danin3, Julien Pottecher4, Florence Fagot-Gandet4, Eliane Albuisson2, Xavier Monnet5, Jean-Louis Teboul5, Martin Dres1

1APHP, Paris, FRANCE; 2CHRU de Nancy, Vandoeuvre-Les-Nancy, FRANCE; 3CHU, Nice, FRANCE; 4Hôpitaux Universitaires de Strasbourg, Strasbourg, FRANCE; 5Hôpital de Bicêtre, Le Kremlin-Bicêtre, Le Kremlin-Bicêtre
Correspondence: Hélène Beringuer -

Annals of Intensive Care 2019, 9(Suppl 1):CO-35

Introduction: Fluid expansion is the first therapeutic option in patients presenting acute circulatory failure but its hemodynamic effects (persistency and time of maximal increase in cardiac output) are unknown. We sought to describe the time course of cardiac output over a 2-hours period after a fluid expansion. Our objectives were 1) to identify patterns of fluid responsiveness and 2) to determine the time of maximal increase in cardiac output during and after fluid expansion.

Patients and methods: It was a prospective multicentre observational study conducted in four intensive care units. To be included, mechanically ventilated patients with acute circulatory failure (infusion of norepinephrine) had to be equipped with a transpulmonary thermodilution device (PiCCO 2) and a decision of fluid expansion (500 ml of saline over a standardized 10-minutes period) had to be made. Calibration was achieved before and at two hours after fluid expansion. Transpulmonary thermodilution derived indices were collected over the two-hours period of observation. Fluid responsiveness was defined as an increase in cardiac index 15% from the start of the fluid expansion. Four patterns of fluid responsiveness were predefined- never responders, ultrafast responders (before the end of the fluid expansion), short-term responders (at the end of the fluid expansion) and persistent responders (over the 2-hours observation period). No change in drugs dosage nor ventilatory settings were allowed during the study.

Results: Fifty-eight patients (79 cases) were included. Septic shock was the main reason of acute circulatory failure (46 58, 79%). Patterns of fluid responsiveness were the followings- non-responders (36 79, 46%), ultrafast responders (3 79, 4%), short-term responders (13 79, 16%) and persistent responders (6 79, 8%). In addition, 21 (27%) patients had hemodynamic instability and required a second fluid expansion during the protocol. No significant difference was found between patterns in terms of baseline hemodynamic characteristics nor drugs administrated. The increase in cardiac output was maximal after 300 [189–411] seconds in never responders, after 516 [396–672] seconds in ultrafast responders, after 456 [360–660] seconds in short-term responders and after 348 [315–372] seconds in persistent responders.

Conclusion: Persistent fluid responsiveness occurred in a minority of patients presenting acute circulatory failure. Maximal increase in cardiac output could be observed during fluid expansion between 5 and 10 min.

CO-36 Molar Sodium Lactate Improves Mesenteric MicroCirculation, Cardiac Function, Capillary Leakage and Inflammation in a Rat Sepsis Model

Emmanuel Besnier (speaker)1, David Coquerel2, Geoffrey Kouadri1, Mathieu Soulié3, Nicolas Perzo3, Raphael Favory4, Thibault Duburcq4, Olivier Lesur2, Soumeya Bekri5, Vincent Richard3, Paul Mulder3, Fabienne Tamion6

1Department of Anesthesia and Critical Care, Rouen University Hospital, Rouen, FRANCE; 2Université de Sherbrooke, Division of Intensive Care Units, Centre de Recherche Clinique du CHUS, Sherbrooke, CANADA; 3Normandie Univ, Unirouen, Inserm U1096 EnVi, Rouen, FRANCE; 4Intensive care unit, Lille University Hospital, Lille, FRANCE; 5Institute of Clinical Biology, Rouen University Hospital, Rouen, FRANCE; 6Department of Medical Intensive Care, Rouen University Hospital, Rouen, FRANCE
Correspondence: Emmanuel Besnier -

Annals of Intensive Care 2019, 9(Suppl 1):CO-36

Introduction: Fluids composed of molar sodium lactate have recently been identified as beneficial in endotoxinic animal models1. The objective of our work was to evaluate the effects of molar sodium lactate on micro- and macro-circulation, capillary leakage and different biological parameters in a sepsis rat model of caecal ligation and puncture (CLP).

Patients and methods: We realized CLP in 30 rats randomized in 3 groups (n = 10 per group)- Sham + CLP-NaCl 0.9% + CLP-Lactate 11.2%. Immediately after CLP, fluids were infused intravenous (2.5 mL kg h) during 18 h. Then, we evaluated mesenteric microcirculation (laser speckle imager), cardiac function (echocardiography) and inflammation (uremia, albuminemia, VEGF-A, IL-1β, IL-10, TNFα). Assay of capillary leakage using Blue Evans extravasation in the lung and gut were realized on additional rats (n = 5 group). Results are expressed as medians with interquartiles and comparisons versus CLP-NaCl were realized using Kruskall-Wallis or ANOVA test.

Results: Mesenteric microcirculation was lower in CLP-NaCl vs. Sham (240.6 [209.3–390.8] vs. 935.9 [855.0–1067.0] pixels unit, p < 0.0001) and CLP-Lactate (735.5 [407.4–878.8], p = 0.0006). CLP-NaCl rats presented a lower cardiac output vs. Sham (0.14 [0.10–0.18] vs. 0.30 [0.26–0.34] mL/min g, p = 0.004) and CLP-lactate (0.34 [0.28–0.43], p < 0.0001) and lower left ventricular shortening fraction vs. CLP-Lactate (39.1 [32.9–51.8] vs. 55.2 [46.2–73.2]  %, p = 0.009). There was no difference concerning albuminemia, left ventricular diastolic diameter, central venous pressure or E A mitral flow ratio between CLP-NaCl and CLP-lactate, suggesting comparable volemia. Mean arterial pressure between CLP-NaCl and CLP-Lactate was similar at the end of infusion. Evans Blue diffusion was reduced in the gut and the lung for CLP-lactate (37.2 [31.0–43.3] vs. 112.7 [63.3–141.6] and 107.5 [82–174.3] vs. 272.7 [221.8–444.5] ng EB mg of tissue). Plasma levels of lactate and 3OH-butyrate were higher in CLP-lactate vs. CLP-NaCl (6.03 [3.08–10.3] vs. 3.19 [2.42–5.11] mmol/L, p = 0.04 + 400 [174–626] vs. 189 [130–301] µmol L, p = 0.03), but no difference for plasma pyruvate or acetoacetate. Inflammatory response was reduced in CLP-lactate (IL-1β- 172.2 [119.0–446.3] vs. 927.7 [244.8–1470] pg mL, p = 0.004 + TNFα- 17.9 [12.5–50.3] vs. 53.9 [30.8–85.6] pg mL, p = 0.005 + IL-10- 351.6 [267.0–918.6] vs. 904.5 [723.1–1243] pg mL) as well as VEGA-A plasma levels (198.2 [185.3–250.0] vs. 260.7 [249.8–268.9] pg mL, p = 0.009). No difference was observed for syndecan plasma level.

Conclusion: We demonstrate for the first time that molar lactate fluid perfusion protects against CLP-induced cardiovascular dysfunction, mesenteric microvascular alteration and capillary leakage, in association with a significant reduction in inflammatory process. These results suggest that molar lactate fluid perfusion may be an attractive target for the treatment of sepsis.

CO-37 Severe pneumocystis jiroveci pneumonia in human immunodeficiency virus negative patients

Julien Grouille (speaker)1, Jean Baptiste Lascarrou2, Delphine Chatellier3, Mickael Landais4, Pierre Fillatre5, Thomas Daix6, Virginie Verrier 7, Jean Claude Lacherade8, Grégoire Muller9, Marc Feller10, Michel Hira11, Maud Jonas12, Denis Garot1

1CHRU Bretonneau, Tours, FRANCE; 2CHU, Nantes, FRANCE; 3CHU, Poitiers, FRANCE; 4CH Le Mans, FRANCE; 5CH, Saint Brieuc, FRANCE; 6CHRU, Limoges, FRANCE; 7CH, La Rochelle, FRANCE; 8CH, La Roche Sur Yon, FRANCE; 9CHR, Orleans, FRANCE; 10CH, Blois, FRANCE; 11CH, Chateauroux, FRANCE; 12CH, Saint Nazaire, FRANCE
Correspondence: Julien Grouille -

Annals of Intensive Care 2019, 9(Suppl 1):CO-37

Introduction: Pneumocystis jiroveci pneumonia (PCP) is one of the main opportunistic complications in patients infected with the human immunodeficiency virus (HIV), but it also affects non-HIV-infected patients, with mortality rates of 30 to 60%. We have described the population of HIV-negative patients hospitalized in intensive care units (ICU) for PCP.

Patients and methods: A French multicenter retrospective study performed in 12 ICUs from ARCO (Association des Réanimateurs du Centre Ouest) between January 1st 2012 and December 31st 2017.

Discussion: 117 patients were included, mean age 63.9 years. The main underlying diseases were hematological diseases (46 cases, 39.3%). 99 patients (83.9%) received chemotherapeutic treatments, 77 (65.8%) were on steroids. Of 44 patients undergoing steroids potentially**eligible for PCP prophylaxis, 43 did not received it (97.7%).

105 patients (89.7%) had dyspnea as their main symptom. A biological inflammatory syndrome was present in 116 cases**(98.8%). Chest CT revealed an interstitial lung disease for 53 patients (65.4%). The diagnosis was made during bronchial fibroscopy with bronchoalveolar lavage (BAL) in 106 cases (90.6%) and by non-invasive method (specific sputum PCR, B D glucan assay) in 11 cases (9.4%) with a mean diagnostic delay after admission of 4.5 days.

65 subjects (55.5%) were mechanically ventilated (average duration of 15.5 days). 35 patients (30.3%) developed septic shock with vasopressor support (average duration of 6.4 days). 39 patients (33.3%) had acute renal failure according to the KDIGOcriteria. The average length of stay in ICU was 14.8 days. 36 patients (30.7%) were treated for a possible PCP upon admission and the average treatment time for the other patients was 3.9 days. 112 (97.5%) received curative treatment with cotrimoxazole and 26 (23.3%) developed side effects. Steroids were associated with anti-infective treatment for 96 patients (82%) with no impact on evolution. The mortality rate in ICU was 46 (39.3%).

Conclusion: PCP is an increasingly infection over immunosuppressed non HIV infected patients. Our study confirms the severity of this pathology and the main populations at risk. PCP should be mentioned in cases of pulmonary involvement in at-risk patients and a probabilistic treatment should be implemented quickly. The interest of an adjuvant steroid treatment remains to be demonstrated. Recommendations on prophylactic treatment should be more widely distributed and probably further refined.

No authors have any conflicts of interest.

CO-38 Candidemia in critically ill immuno-compromised patients

Etienne Ghrenassia (speaker)1, Djamel Mokart2, Julien Mayaux3, Alexandre Demoule3, Imène Rezine1, Lionel Kerhuel1, Laure Calvet1, Audrey De Jong1, Elie Azoulay1, Michael Darmon1

1Hôpital Saint Louis, Paris, FRANCE; 2Insititut Paoli-Calmettes, Marseille, FRANCE; 3Hôpital Pitié Salpétrière, Paris, FRANCE
Correspondence: Etienne Ghrenassia -

Annals of Intensive Care 2019, 9(Suppl 1):CO-38

Introduction: Candidemia is a major threat for patients with cancer. This study sought to identify outcomes of critically ill immunocompromised patients with candidemia. Our secondary objectives were to describe the clinical phenotype of patients, Intensive Care Units (ICU) features, Candida ecology, and factors associated with hospital mortality.

Patients and methods: Adult critically ill immunocompromised patients with candidemia were retrospectively included from 3 ICUs. Immune defect was defined as an underlying solid tumor, hematological malignancy or autoimmune disease.

Results: Overall, among 31792 patients admitted in 3 ICUs over a 15-y period (2002 to 2017), 219 developed a Candidemia (0.7%). Of these, 121 had an underlying immune defect and were ultimately included. Two third of the patients had hematological malignancy, chiefly non-Hodgkin’s lymphoma (37%), and acute leukemia (21%). Respectively, 10% and 7% of included patients underwent autologous or allogeneic stem cell transplantations. Most patients had one or more risk factors of candidemia. During ICU stay, 91 (75%) patients needed invasive mechanical ventilation, 71 (61%) patients needed renal replacement therapy, and 91 (75%) patients needed vasopressors. One out of five patients (21%) was considered as emergency surgery patients. Candida albicans (54%) was the predominant specie, followed by glabrata (19%), tropicalis (11%), parapsilosis (7%) and krusei (7%). One third of identified Candida was resistant to Fluconazole (30%) and 8% to Echinocandins. ICU mortality was 52% and hospital mortality was 60%. Candida specie or susceptibility was not associated with outcome (figure). Median time from feature of candidemia onset to first antifungal therapy was 3 (1–3) days. After adjustment for confounders, need for vasopressors (HR- 1.8 + CI95%- 1.1–3.1), need for mechanical ventilation (HR- 2.0 CI95%- 1.1–3.8) and allogeneic stem cell transplantation (HR- 2.5 + CI95%- 1.1–6.0) were independently associated with hospital mortality. Half the patients (58%) developed candidemia more than 24 h after ICU admission and 42% were admitted in ICU for severe candidemia. There was no difference in clinical or Candida features or outcomes between these two kinds of patients.

Conclusion: Candidemia in critically ill immunocompromised patients is associated with high mortality rates. Despite high prevalence of non-albicans Candida, only severity and preexisting allogeneic stem cell transplantation were independently associated with poor outcome. Last, antifungal therapy proved to be started late as regard to symptoms. Then, development of strategies to reduce this delay are warranted.

CO-39 Dynamic risk factors of invasive Candida infection in intensive care unit- a landmark method

Clément Le Bihan (speaker)1, Sébastien Bailly1, Maité Garrouste-Orgeas1, Elie Azoulay2, Lila Bouadma3, Carole Schwebel4, Claire Dupuis1, Guillaume Marcotte5, Yves Cohen6, Bertrand Souweine7, Laurent Papazian8, Jean Reignier10, Stéphane Ruckly11, Jean-Francois Timsit1

1UMR 1137 - IAME Team 5 - DeSCID- Decision SCiences in Infectious Diseases, control and care Inserm University Paris Diderot, Sorbonne Paris Cité, Paris, FRANCE; 2Medical ICU, Saint-Louis Hospital, APHP, Paris, FRANCE; 3AP-HP, Bichat Hospital, Medical and infectious diseases Intensive Care Unit, Paris Diderot university, Paris, FRANCE; 4Medical ICU, Albert Michallon University Hospital, Grenoble, FRANCE; 5Surgical ICU, Edouard Herriot University Hospital, Lyon, FRANCE; 6AP-HP, Avicenne Hospital, Intensive Care Unit, Paris and Medicine University, Paris 13 University, Bobigny, FRANCE; 7Medical ICU, Gabriel Montpied Hospital, Clermont-Ferrand, FRANCE; 8Respiratory and infectious diseases ICU, APHM Hôpital Nord, Aix Marseille University, Marseille, FRANCE; 10Medical ICU, Hôtel-Dieu University Hospital, Nantes, FRANCE; 11ICU REsearch, Department of Biostatistics, Paris, FRANCE
Correspondence: Clément Le Bihan -

Annals of Intensive Care 2019, 9(Suppl 1):CO-39

Introduction: Invasive candidiasis (IC) is associated with a high incidence of mortality in intensive care unit (ICU). Delayed antifungal therapy for IC contributes to poor outcomes in ICU. None of the predictive published risk scores are able to accurately predict IC. Therefore, new scores introducing previous anti-infective therapy, elapsed time between admission, risk evaluation and differentiating candidemia and other invasive candidiasis should be developed.

Patients and methods: Retrospective analysis of data from the OUTCOMEREA prospective multicenter cohort. Critically ill patients with an ICU length of stay of at least 3 days were included. Multivariable sub-distribution survival models with competing risk were used at each day, from day 1 to day 10 using a landmark method with a sliding window and an horizon of 10 days. The impact of every identified risk was analysed for candidemia and invasive visceral candidiasis (IVC- Candida peritonitis or pleural candidiasis) separately. Competing risks were ICU-death and ICU-discharge. Predicted probability of infection during the next 10 days was calculated at day 1, 4 and 7 based on cumulative incidence function of every combination of risk.

Results: 16901 patients were included. Candidemia occurs in 96 patients and IVC in 87 patients. Antifungal therapy was used in 2019 (12%) patients. In multivariate analysis, 10 risk factors for candidemia and 12 for IVC were identified (figure). We confirmed the impact of some previously described risk factors (i.e. septic shock, multifocal colonization, SOFA score, hospital stay before ICU, parenteral nutrition, corticosteroids, blood transfusion, mechanical ventilation and surgery). Those risks were differently associated with candidemia and IVC. Some variables such as corticosteroids therapy, blood transfusion and mechanical ventilation were only associated with candidemia. Risks were time dependent with some having more impact on early state (landmark time < 5) or on late onset (landmark time > day 5). Antibiotherapy (broad spectrum, piperacillin-tazobactam), number of PPI days of use and a medical history of cancer were only associated with IVC after landmark time 5. Prediction model found maximum predicted probability of 38, 17 and 16% at landmark 1, 4 and 7 respectively for candidemia and 58, 52 and 41% for IVC. A shiny app was developed to compute the predicted risk of candidemia or IVC depending on risk factors combination.

Conclusion: Risk factors of Candida infection are different for candidemia and IVC and must be analysed as distinct diseases. Moreover, those risks are highly time-related with an early and late pattern during ICU stay.

CO-40 Zygomycosis in Intensive care unit- surgery as a major prognostic factor

Johanna Claustre (speaker)1, Thomas Jouve1, Romaric Larcher3, Saad Nseir4, Julien Cadiet5, Yoann Zerbib6, Alexandre Lautrette 7, Jean-Michel Constantin7, Pierre-Emmanuel Charles8, Cédric Daubin9, Remy Coudroy10, Jean Dellamonica11, Laurent Argaud12, Pierre Phelouzat13, Damien Contou14, Juliette Pocquet15, Guillaume Voiriot16, Jean-Christophe Navellou17, Pierre Lavagne18, Michel Durand19, Carole Schwebel20, Nicolas Terzi20

1CHU Grenoble Alpes, Grenoble, FRANCE; 3CHU, Médecine intensive et Réanimation, Montpellier, FRANCE; 4CHU, Urgence respiratoire et réanimation médicale, Lille, FRANCE; 5Service de médecine intensive Réanimation, Nantes, FRANCE; 6Réanimation Médicale, Amiens, FRANCE; 7Réanimation Médicale, Clermont-Ferrand, FRANCE; 8Médecine intensive Réanimation, Dijon, FRANCE; 9Réanimation Médicale, Caen, FRANCE; 10Réanimation Médicale, Poitiers, FRANCE; 11Réanimation Médicale, Nice, FRANCE; 12Réanimation Médicale, Lyon, FRANCE; 13Réanimation Médicale, Rennes, FRANCE; 14Réanimation polyvalente, Argenteuil, FRANCE; 15Médecine intensive Réanimation, Tours, FRANCE; 16Réanimation médico-chirurgicale, Paris, FRANCE; 17Réanimation Médicale, Besançon, FRANCE; 18Réanimation polyvalente chirurgicale, Grenoble, FRANCE; 19Réanimation cardiovasculaire et thoracique, Grenoble, FRANCE; 20Médecine intensive Réanimation, Grenoble, FRANCE
Correspondence: Johanna Claustre -

Annals of Intensive Care 2019, 9(Suppl 1):CO-40

Introduction: Zygomycosis is an invasive fungal infection, with an increasing incidence especially in patients with hematological malignancy. Its prognosis is poor because of its high invasive power, leading to fungal dissemination and tissular necrosis. No study aimed to describe patients with zygomycosis admitted in intensive care units (ICU). We aimed to describe epidemiology of zygomycosis in French ICU and evaluate outcomes.

Patients and methods: We performed a retrospective multi-center study in 16 French ICUs (universitary hospitals or general hospitals) between 2008 and 2017. During this period, 74 patients were diagnosed as having probable or proven zygomycosis. We reported demographic baseline and evolutive clinical data and compared the patients who survived in ICU (21 patients) and the patients who did not (53 patients) with univariate and multivariate analyses to identify factors associated with ICU survival. Then, we also focus on the subgroup of patients with hematological malignancy.

Results: Sixty of the 74 patients admitted in ICU and having zygomycosis were deeply or mildly immunocompromised- 41 had an hematological malignancy, 4 had a solid malignancy, 9 were solid organ transplant recipients, 2 had an inflammatory disease, 31 had long-term steroids, 11 had a diabetes, 24 had a malnutrition. Among the 74 patients, only 21 patients survived to ICU stay (28.4%) with a median survival of 22 days (Q1-Q3 = 9–106). In the univariate and multivariate analyses, survivors were significantly younger (p < 0.001), hematological malignancy was less frequent (p = 0.003), as well as malnutrition (p = 0.02). Trauma was associated with better survival (p = 0.02) in univariate analysis, while all patients with a disseminated infection (2 or more non-contiguous infected organ) died in ICU (p = 0.007). **Regarding patients with hematological malignancy, median survival (n = 41) was 15 days (Q1-Q3 = 5–23.5 days), worse than 22 days (Q1-Q3 = 9–106) in the whole cohort. In multivariate analysis, surgery was the only factor associated with survival (p = 0.02)- no long-term survival was observed in the subgroup of patients without surgical management (Log-rank p < 0.001, Figure). The association of two antifungal treatment did not demonstrate any benefit in survival (p = 0.21).

Conclusion: Overall prognosis of zygomycosis in ICU remains poor, especially in patients with hematological malignancy. Despite an increased frailty, an aggressive management including early surgery was the only factor improving survival in this ICU population.

CO-41 Clinical features and outcome of life-threatening Adult Onset Still Disease- a French multi center retrospective study

Lucie Mortier (speaker)1, Marc Pineton de Chambrun2, Eliane Albuisson3, Nathalie Lerolle4, Matthieu Groh5, Philippe Guiot6, Mathilde Neuville 7, Jérémie Joffre8, Damien Du Cheyron9, Roland Jaussaud10, Pierre-Edouard Bollaert1, Jérémie Lemarié1

1Service de Médecine Intensive Réanimation, Nancy, FRANCE; 2Hôpital de la Pitié Salpêtrière - AP-HP, Paris, FRANCE; 3Unité de Méthodologie, Datamanagement & Statistiques, Vandoeuvre-Les-Nancy, FRANCE; 4Service de Médecine Interne Hôpital Bicêtre, APHP, Le Kremlin-Bicêtre, FRANCE; 5Service de Médecine Interne Hôpital St Louis, Paris, FRANCE; 6Grpe Hosp Region Mulhouse & Sud Alsace, Hopital Emile Muller, Mulhouse, FRANCE; 7Réanimation Médicale et Infectieuse Hôpital Bichat-Claude Bernard, Paris, FRANCE; 8Service de réanimation médicale, Paris, FRANCE; 9Service de réanimation médicale, Caen, FRANCE; 10Service de Médecine Interne Hôpital St Louis, Vandoeuvre-Les-Nancy, FRANCE
Correspondence: Lucie Mortier -

Annals of Intensive Care 2019, 9(Suppl 1):CO-41

Introduction: Adult onset Still disease (AOSD) is a rare and complex auto-inflammatory disorder with several systemic life-threatening manifestations and empirical treatment. Our objectives were to describe of clinical features, management and outcome of AOSD patients who required intensive care support.

Patients and methods: We undertook a 17-year retrospective multicenter cohort study in 44 French academic ICUs from January 2001 to June 2017 including all cases of ICU admission related to life-threatening AOSD. All patients older than 18 years were considered for inclusion if they met the following criteria- (1) admission to ICU, (2) related to systemic AOSD flare (according to Yamaguchi and or Fautrel criteria). The study methodology and the data privacy policy (French methodology of reference MR-003) were approved by our institutional ethics committee (application no. 156-2017) and registered on (NCT03276650).

Results: Forty-five patients were included (median age 34.5 (1st-3rd quartiles 24–42.5) years). Men were more affected (54%) than women. It was the first AOSD flare for 60% of population. Median SAPS II was 32.5 (21.5–44). Cardio-circulatory (shock, 42% and pericarditis, 40%), respiratory (Acute Respiratory Distress Syndrome, 22%) and hematologic (Reactive Hemophagocytic Syndrome, 16%) disorders were the most frequent life-threatening manifestations. Multiple organ failure was frequent (24%) and associated with bad prognosis if presented at admission (p = 0.021). Nine patients died (20%, median age of 24 (21–41) years) among whom 8 with previously diagnosed AOSD, versus 10 without previous diagnosis (p = 0.001). Comparison between survivors and non-survivors during ICU stay are shown in Table 1. Corticosteroids (CS) were used as first line therapy in 43 (96%) patients with efficacy of CS monotherapy in 27 (63%). Three patients were successfully treated by anakinra, with 43% global efficacy. Four patients with more severe organ failures at admission died despite treatment with anakinra, introduced after several lines of other immunosuppressive agents (cyclosporine, etoposide, intravenous immunoglobulins).

Conclusion: Herein, we report the largest cohort of patients admitted to ICU for life-threatening AOSD. It can lead to life-threatening presentation with miscellaneous presenting clinical spectrum and high mortality rate among young people with previous diagnosis of AOSD. Because of the lack of specific clinical and biological signs and its frequent cardiovascular collapse, the key point is to quickly eliminate septic diagnosis and add immunosuppressive treatment to empiric antibiotherapy.

CO-42 Septic shock among patients with systemic lupus erythematosus- short and long-term outcome and associated costs. Analysis of a French nationwide database

Arthur Mageau (speaker)1, Karim Sacre2, Anne Perozziello1, Stéphane Ruckly1, ClaireDupuis1, Lila Bouadma1, Thomas Papo3, Jean-François Timsit1

1IAME UMR 1137 Team 5 Descid, Paris, FRANCE; 2Hôpital Bichat, Service de Médecine Interne, Paris, FRANCE; 3Service de Médecine Interne, Hôpital Bichat, Paris, FRANCE
Correspondence: Arthur Mageau -

Annals of Intensive Care 2019, 9(Suppl 1):CO-42

Introduction: We aimed to assess the characteristics, outcome and associated costs of septic shock complicating Systemic Lupus Erythematosus (SLE).

Patients and methods: Characteristics of SLE patients experiencing a septic shock in France from 2010 to 2015 were analyzed through the French nationwide medico-administrative database. The factors associated with the 1-year post-admission mortality were analyzed, the crude 1-year survival of SLE patients experiencing septic shock was compared to those admitted for another reason, and we compared the 1-year outcome and costs associated with SLE septic shock survival to a matched SLE ICU control population.

Results: Among 28.522 SLE patients, 1,068 experienced septic shock. Septic shock was severe (mean SAPSII score- 47.0 [± 21.6]) + 369 (34.6%) patients had invasive mechanical ventilation and 342 (32.0%) required renal replacement therapy. The 1-year mortality rate was 43.4% (n = 463). The hospital stay mean cost was € 25,327 [± 23,396]. Factors assocaited with 1-year mortality are on Table 1. Independently of the severity of comorbidities and acute illness, an associated Sjögren syndrome was the only specific SLE phenotype significantly associated with 1-year mortality (HR [CI95]- 1.392 [1.021–1.899]). Within one year, post-septic shock survivors (n = 738) were re-admitted 6.42 [17.3] times for 64.1 [48.9] days with a total cost of € 14,431 [20,444]. Unmatched analysis showed that the outcome of patients admitted in ICU for septic shock was poorer than that of patients admitted in ICU or hospital for another disease. Post-ICU 1-year hospital outcome and costs of septic shock survivors were not different from the other ICU survivors when matched on the severity of acute illness.

Conclusion: Associated Sjögren syndrome was the main determinant of the 1-year mortality among SLE patients experiencing septic shock. Although septic shock was a turning point in SLE history, survivors did not use more healthcare than survivors of another life-threatening disease.

CO-43 A 15 year retrospective audit of plasma exchange in the critically ill- Predictive factors of associated infections

Delphine Daubin (speaker)1, Florian Rissel1, Fanny Garnier1, Vincent Brunot1, Romaric Larcher1, Laura Platon1, Noemie Besnard1, Tarik Kanouni2, Pascal Latry2, Boris Jung1, Kada Klouche1

1Intensive Care Unit Department, Montpellier, FRANCE; 2Hematology Department, Montpellier, FRANCE
Correspondence: Delphine Daubin -

Annals of Intensive Care 2019, 9(Suppl 1):CO-43

Introduction: Plasma exchange (PE) often leads to adverse effects. Sepsis resulting from impaired immunity caused by the removal of antibodies and related to catheter-associated infections represents the most severe complications. The study evaluate the infectious risk associated with the current use of PE and identify its predictive factors.

Patients and methods: All major patients admitted to our tertiary ICU and received PE therapy between 2002 and 2017 were included. Patients who developed secondary infection were compared to those who did not by the use of univariate and multivariate analysis.

Results: Eighty-eight patients underwent 616 PE procedures. 56% patients were in mechanical ventilation, 44% received constrictive drugs and 40% were in renal replacement therapy. Indications were Thrombotic Thrombocytopenic Purpura 48 (55%), Neurologic diseases 18 (20%), Vasculitis 14 (16%), autoimmune diseases 4 and other indications 4.

Adverse effects per-procedure were observed in 22% of patients in witch half of them were life-threathening. PE-associated infections occurred in 39 patients (44%). Median time elapsed from ICU admission and EP initiation to infection diagnosis were respectively 8 [5–14] and 4 [2–10] days. Infectious complications included 19 ventilation-associated pneumonia, 6 nosocomial pneumonia, 16 viral reactivations, 5 bacteremia, 4 catheter-associated infections, 4 urinary infections. Most of them were severe with a septic shock in 22 39 patients (58%) and 8 patients deceased (20%).

The comparison by univariate analysis showed that patients with associated infection had a significantly higher age, need more frequently a mechanical ventilation and vasoactive drugs. The number of PE and ICU and in-hospital length of stay were significantly higher in the infectious group. No significant differences were observed between infectious and no infectious groups regarding to severity score at admission, corticoids, immunosuppressive therapy, type of substitute liquid, PE volume. In multivariate analysis, age ([OR] = 1.05, CI95% 1.01–1.09, p = 0.02), number of session ([OR] = 1.22, CI 95% 1.08–1.39, p = 0.001) and requirement of mechanical ventilation ([OR] = 4.19, CI95% 1.27–13.81, p = 0.02) were predictive of the infection occurrence. ICU mortality was higher in the infectious group (28% versus 16% p = 0.28).

Conclusion: PE was associated with an infection, in particular pneumonia, in 44% of patients and lead to a life-threatening condition in 25% of them. Age, number of PE session and requirement of mechanical ventilation were predictive of the occurrence of infection.

CO-44 A reappraisal of short and long-term outcome for patients with systemic rheumatic diseases admitted to intensive care unit and reassessment of prognostic factors- a multicenter retrospective study of 415 patients

Romaric Larcher (speaker)1, Marc Pineton de Chambrun2, Emma Rubenstein1, Laura Platon1, Kevin Chalard1, Delphine Daubin1, Noemie Besnard1, Philippe Corne1, Jonathan Charbit1, Charles-Edouard Luyt2, Zahir Amoura2, Samir Jaber1, Boris Jung1, Kada Klouche1

1Montpellier University Hospital, Montpellier, FRANCE; 2La Pitie-Salpetriere Hospital, APHP, FRANCE
Correspondence: Romaric Larcher -

Annals of Intensive Care 2019, 9(Suppl 1):CO-44

Introduction: Recent advances in critical and rheumatic care may have impacted the outcome for patients with systemic rheumatic diseases (SRD) admitted to the intensive care unit (ICU). We aimed to reassess the short and long-term outcome and prognostic factors for SRD patients admitted to the ICU.

Patients and methods: We conducted a multicenter retrospective observational study in seven French ICU. All SRD patients admitted between 2006 and 2016 were included. In-hospital mortality and one-year survival after discharge were estimated through the Kaplan–Meier method and prognostic factors through Cox models. We compared survived and non-survived patients by univariate and multivariate analysis in order to identify prognostic factors of in-hospital mortality and one-year survival.

Results: In the 415 patients (female- 58%, mean age- 57 years) included, predominant SRD were vasculitides, rheumatoid arthritis and systemic lupus erythematosus. Main causes of admission were shock (39%) and acute respiratory failure (33%) and predominant diagnoses in ICU were infections (39%) and SRD flare-up (36%). At admission, mean SOFA score was 7 ± 4 and mean SAPS II was 46 ± 20. 276 patients (67%) required mechanical ventilation, 211 (51%) vasoactive drugs, 122 (29%) renal replacement therapy and 37 (9%) extracorporeal membrane oxygenation. In-hospital mortality rate was 29.4%, 95%CI [25–33.8] and one-year survival probability was 91.7%, 95%CI [88.3–95.2] (figure 1). Prognostic factors were age (HR 1.02, 95%CI [1.01–1.04], p = 0.002), previous corticosteroid treatment (HR 1.71, 95%CI [1.13–2.58], p = 0.01), SOFA score (HR 2.03, 95%CI [1.31–3.15], p = 0.002) and SAPS II (HR 1.63, 95%CI [1.09–2.45], p = 0.02) for in-hospital mortality. Besides age (HR 1.05, 95%CI [1.01–1.09], p = 0.007) no other variable was associated with one-year mortality after discharge.

Conclusion: Our study showed that most of two third of SRD patients admitted to the ICU survived at hospital discharge and a probability of one-year survival at 91.7% for these patients. This satisfactory survival rate may be related to the recent improvement in critical care and rheumatic care. We also found that SOFA score, SAPS II and previous corticosteroid treatment were predictive factors of in-hospital outcome.

CO-45 Chest CT scan and alveolar procollagen III to predict lung fibroproliferation in acute respiratory distress syndrome

Annabelle Hamon (speaker)1, Ugo Scemama2, Jeremy Bourenne3, Florence Daviet2, Benjamin Coiffard2, Nicolas Persico2, Mélanie Adda2, Christophe Guervilly2, Sami Hraiech2, Kathia Chaumoitre2, Antoine Roch2, Laurent Papazian2, Jean-Marie Forel2

1Chiap, Marseille, FRANCE; 2Hôpital Nord AP-HM - CHU, Marseille, FRANCE; 3Hôpital Timone AP-HM - CHU, Marseille, FRANCE
Correspondence: Annabelle Hamon -

Annals of Intensive Care 2019, 9(Suppl 1):CO-45

Introduction: Lung fibroproliferation in ARDS patients is associated with mortality. Alveolar procollagen III (NT-PCP-III) is a biomarker of lung fibroproliferation. Chest CT scan could be useful for the diagnosis of fibroproliferative radiologic lesions. The main objective of the present study was therefore to quantify chest CT scan lesions in ARDS patients with a high alveolar level of NT-PCP-III indicating an active fibroproliferation.

Patients and methods: This retrospective cohort study over 6 years (January 2011-January 2017), included ARDS who had at least one dosage of alveolar NT-PCP-III obtained by performing a bronchoalveolar lavage (BAL) and a chest CT scan within 3 days before or after BAL. According to a previous study alveolar level of NT-PCP-III > 9µG L indicated a histological lung fibroproliferation (1). CT scan was scored on interstitial (honeycombing) and alveolar (ground-glass) abnormalities (2). Crude score and corrected score (related to the number of scored lobes in case of important lobar condensation) were used.

Results: One hundred ninety-two patients were included, either 228 “couples” alveolar NT-PCP-III and CT scan. Using a threshold of NT-PCP-III > 9µG L, crude and corrected fibrosis score was highest when lung fibroproliferation are present. Values of lung fibroproliferation CT scan score according to alveolar NT-PCP-III level appear in table 1. The median delay between the onset of ARDS and CT scan was of 5 [0–14] days, between the onset of ARDS and BAL 5 [0–14] days, between BAL and CT scan 0 [− 1–1] days.

Conclusion: When alveolar NT-PCP-III is used as a surrogate of lung fibroproliferation, CT scan fibrosis score is significantly higher in patient with active pulmonary fibroproliferation.

  1. 1.

    Forel J-M, Guervilly C, Hraiech S, Voillet F, Thomas G, Somma C, et al. Type III procollagen is a reliable marker of ARDS-associated lung fibroproliferation. Intensive Care Med. 1 janv 2015 + 41(1): 1-11.

  2. 2.

    Kazerooni EA, Martinez FJ, Flint A, Jamadar DA, Gross BH, Sp


CO-46 Ultra-protective ventilation without extracorporeal circulation in moderately severe and severe ARDS patients

Jean-Christophe Richard (speaker)1, Sophie Marqué2, Antoine Gros3, Michel Muller4, Gwenaël Prat5, Gaëtan Beduneau6, Jean-Pierre Quenot 7, Jean Dellamonica8, Romain Tapponnier9, Edouard Soum10, Bitker Laurent11, Jack Richecoeur12

1Hospices Civils de Lyon, FRANCE; 2Centre Hospitalier Sud-Francilien, Corbeil-Essonnes, FRANCE; 3Hôpital André Mignaud, Le Chesnay, FRANCE; 4Centre Hospitalier Annecy Genevois, Pringy, FRANCE; 5CHU de la Cavale Blanche, Brest, FRANCE; 6CHU Charles Nicolle, Rouen, FRANCE; 7Hôpital François Mitterrand, Dijon, FRANCE; 8CHU Nice- Hôpital Archet 1, Nice, FRANCE; 9Hospices Civils de Lyon- Centre Hospitalier Lyon Sud, Pierre-Benite, FRANCE; 10CHU Gabriel-Montpied, Clermont-Ferrand, FRANCE; 11Hospices Civils de Lyon- Hôpital de la Croix-Rousse, Lyon, FRANCE; 12Centre Hospitalier, Beauvais, FRANCE
Correspondence: Jean-Christophe Richard -

Annals of Intensive Care 2019, 9(Suppl 1):CO-46

Introduction: Ultraprotective ventilation with tidal volume (VT) reduction below 6 ml kg predicted body weight (PBW) in severe ARDS may reduce alveolar strain, driving pressure and hence ventilator-induced lung injury, with main drawback of worsening respiratory acidosis. We hypothesized that VT could be reduced up to 4 ml kg, with clinically significant decrease in driving pressure (DeltaP), without the need for extracorporeal CO2 removal, while maintaining pH in the range targeted in recent ARDS trials.

Patients and methods: We conducted a non-experimental before-and-after multi-center study on 35 ARDS patients with PaO2 FiO2 <= 150 mm Hg, within 24 h of ARDS diagnosis. After inclusion (H0), VT was reduced to 4 ml kg PBW and further adjusted to maintain pH > = 7.20, respiratory rate was increased up to 40 min and PEEP was set using a PEEP-FiO2 table favoring high PEEP. This strategy was applied until positivity of a PEEP weaning trial. The primary judgment criterion was DeltaP on day 2, as compared to study inclusion.

Results: Patients’ age was 62 ± 14 year, SAPS II amounted to 47 ± 15, 29 patients (85%) had pneumonia as ARDS risk factor, SOFA at inclusion was 13 ± 3, and PaO2 FiO2 at inclusion was 107 ± 35 mm Hg under a PEEP of 10 ± 4 cmH2O. From inclusion to day 2, DeltaP decreased significantly from 12.1 ± 4.3 to 8.6 ± 3.1 cmH2O, while VT decreased from 6.1 ± 0.6 to 4.4 ± 0.7 ml kg. On day 2 of the study (table 1), VT was below 4.2 ml kg in 22 patients (65% [IC95% 48%-79%]), and below 5.25 ml kg in 30 patients (88% CI95% [74%-95%]). Time with VT below 4.2 ml kg averaged 1.7 ± 1.4 days. Sedation drugs were not significantly modified. **Two patients (6%) developed acute cor pulmonale after inclusion. Right ventricle left ventricle ratio increased non-significantly from 0.59 ± 0.16 at H0 to 0.69 ± 0.22 on day 2. Eleven patients (32%) presented with severe respiratory acidosis (pH < 7.15). Fourteen patients (41%) died before day 90.

Conclusion: Ultraprotective ventilation may be applied in approximately 2 3 of moderately severe to severe ARDS patients, with a mean reduction of DeltaP approximating 3.5cmH2O, at the price of severe respiratory acidosis in 1 3 of the patients.

Table 1








Driving pressure (cmH2O)

12.1 ± 4.3

9.6 ± 3.2†

9.2 ± 3.9†

9.0 ± 3.8†

8.6 ± 3.1†

9.0 ± 3.5†

VT (−1 PPT)

6.1 ± 0.6

4.2 ± 0.5†

4.3 ± 0.5†

4.3 ± 0.7†

4.4 ± 0.7†

4.7 ± 1.1†

Respiratory rate (min−1)

27 ± 5

35 ± 5†

36 ± 5†

36 ± 5†

37 ± 5†

35 ± 6†

PEEP (cmH2O)

10 ± 4

15 ± 3†

15 ± 4†

15 ± 4†

14 ± 4†

13 ± 4†


7.32 ± 0.10

7.23 ± 0.09†

7.25 ± 0.09†

7.28 ± 0.07

7.28 ± 0.08

7.33 ± 0.09

PaCO2 (mm Hg)

47 ± 11

62 ± 15†

59 ± 12†

56 ± 12†

55 ± 12†

55 ± 15†

PaO2/FiO2 (mm Hg)

107 ± 35

152 ± 61†

188 ± 69†

189 ± 54†

197 ± 50†

197 ± 53†

Midazolam (mg.h−1)

8 ± 7

6 ± 3

7 ± 3

Morphine (mg.h−1)

15 ± 12

15 ± 12

17 ± 15

Vasopressor (µ−1)

0.57 ± 0.64

0.53 ± 0.75

0.29 ± 0.25†

†p < 0.05 vs H0.

CO-47 Improved mortality associated with prone positioning during ECMO for ARDS: a retrospective study in 169 patients

Eloi Prud’Homme (speaker)1, Christophe Guervilly1, Vanessa Pauly1, Florence Daviet1, Jeremie Bourenne2, Benjamin Coiffard1, Jean-Marie Forel1, Sami Hraiech1, Melanie Adda1, Antoine Roch3, Nicolas Persico3, Laurent Papazian1

1Medical Intensive Care Unit, North Hospital, APHM, Marseille, FRANCE; 2Medical and Emergency Resuscitation, la Timone 2 University Hospital, Aix-Marseille University, Marseille, FRANCE; 3Emergency Department, Hôpital Nord, APHM, Marseille, FRANCE
Correspondence: Eloi Prud’Homme -

Annals of Intensive Care 2019, 9(Suppl 1):CO-47

Introduction: Early, prolonged and repeated sessions of prone positioning (PP) have proven substantial outcome benefit in ARDS patients. In the most severe forms of ARDS, veno venous extracorporeal membrane oxygenation (vvECMO) is considered as a therapeutic option performed on a case by case. About 50% of patients treated with vvECMO have previously been turned prone. However, only few studies have evaluated the interest of maintaining PP during ECMO. Therefore, the objective of this study was to compare the 90-day survival rate of patients supported by vvECMO with concomitant PP (prone ECMO group) to those not turned prone during the ECMO period (ECMO alone group).

Patients and methods: We included all adults treated with vvECMO for ARDS between January 2012 and April 2017 in a university teaching hospital in Marseille, France. Patients were either cannulated in our ICU or in a referring hospital then transported by the ECMO mobile team. Before considering patients for ECMO, a PP session was systematically performed in our center if feasible, and strongly encouraged for patients from referral centers. Criteria for vvECMO have been previously published. During ECMO, patients were considered for prone position in case of 1 persistent hypoxemia defined by SpO2 < 88% or PaO2 < 55 mmHg despite a 100% FdO2 and FiO2 with a maximal ECMO blood flow, 2 failure of attempt to wean ECMO after at least 10 days of ECMO and the presence of lung consolidations on chest X-ray or ultrasounds echography, 3 according to the physician in charge. Technical issues have been previously described. We compared the prone ECMO group to the ECMO alone group. We performed Kaplan–Meier 90-day survival curves and Coxmodel for 90-day mortality.

Results: 169 patients were included in the analysis, 92 in the prone group and 77 in the ECMO alone group. The prone group has a 57.6% 90-day survival rate as compared with a 37.7% 90-day survival rate of the ECMO alone group, p = 0.01. SAPS2 [HR = 1.040 (95% CI 1.024–1.056) and use of iNO before ECMO [HR = 1.86 (95% CI 1.156–2.992)] were independently associated with increase in 90-day mortality whereas the use of PP during ECMO was independently associated with decrease in 90-day mortality [HR = 0.439 (95% CI 0.270–0.714)].

Conclusion: The continuation of PP during vvECMO could be of clinical interest. Further prospective controlled study is warranted.

CO-48 Mechanical ventilation management on ECMO for ARDS An International Multicenter Prospective Cohort - The ventiLatIon management oF patients with Extracorporeal membrane oxyGenation for Acute Respiratory Distress Syndrome - LIFEGARDS study

Matthieu Schmidt (speaker)1, Tai Pham2, Christophe Guervilly3, Mathilde Neuville4, Elie Zogheib5, Jean-Paul Mira6, Hadrien Rozé 7, Marc Pierrot8, Daniel Brodie9, Alain Combes10

1Hopital Pitié Salpetrière Medical ICU, Paris, FRANCE; 2Interdepartmental Division of Critical Care Medicine, University of Toronto, CANADA; 3CHU Hopital Nord-Service de Medecine Intensive et Reanimation, Marseille, FRANCE; 4AP-HP, Bichat Hospital, Medical and Infectious Diseases Intensive Care Unit, Paris, FRANCE; 5Cardiothoracic and Vascular Intensive Care Unit, Amiens, FRANCE; 6Cardiothoracic and Vascular Intensive Care Unit, APHP, Groupe Hospitalier Universitaire de Paris Centre, Paris, FRANCE; 7South Department of Anesthesiology and Critical Care, Pessac, FRANCE; 8Service de Réanimation Médicale, Angers, FRANCE; 9Department of Medicine, New York, UNITED STATES; 10Service de Réanimation Médicale, Paris, FRANCE
Correspondence: Matthieu Schmidt -

Annals of Intensive Care 2019, 9(Suppl 1):CO-48

Introduction: The objectives were to report current practices regarding mechanical ventilation and use of adjunct therapies in patients treated with extracorporeal membrane oxygenation (ECMO) for severe acute respiratory distress syndrome (ARDS) and to describe major ECMO-related complications and 6-month outcomes.

Patients and methods: International, multi-center, prospective cohort study of patients undergoing ECMO for ARDS during a one-year period in a convenience sample of 23 international intensive care units (ICUs).

Results: We collected demographics, daily pre- and per-ECMO mechanical ventilation settings and use of adjunctive therapies, ICU- and 6-month–outcome data for 350 patients (median ± standard deviation age 47 ± 17 years, pre-ECMO PaO2 FiO2 71 ± 34 mmHg). Pre-cannulation use of prone positioning and neuromuscular blockers were 26% and 62%, respectively. Tidal volume, plateau pressure, and minute ventilation were significantly reduced during ECMO. The driving pressure was significantly reduced from 18.6 ± 8.3 to 11.6 ± 5.3 cm H2O. Median ECMO duration and ICU stay were respectively 10 and 24 days and six-month survival was 61%. Multivariable analyses retained immunodeficiency, older age, extra-pulmonary sepsis, lower pre-ECMO pH, higher pre-ECMO respiratory rate, greater delay from intubation to ECMO start, and higher mean fluid balance in the first ECMO days as being independently associated with 6-month mortality.

Conclusion: Pre-ECMO management of patients with severe forms of ARDS in high case-volume centers was notable for low use of prone positioning, suggesting the potential for improvement in pre-ECMO ARDS management that might impact the number of cases considered for ECMO. Once receiving ECMO, the use of least damaging lung ventilation strategies was largely adopted.

CO-49 Extracorporeal vs. Conventional Cardiopulmonary Resuscitation Comparison- A matched pairs retrospective study

Daniel Patricio (speaker)1, Lorenzo Peluso1, Alexandre Brasseur1, Olivier Lheureux1, Mirko Belliato2, Jean-Louis Vincent1, Creteur Jacques1, Fabio Silvio Taccone1

1Hôpital Erasme, Université Libre de Bruxelles, Bruxelles, BELGIUM; 2Policlinico San Matteo, Pavia, Italy, Pavia, ITALY
Correspondence: Daniel Patricio -

Annals of Intensive Care 2019, 9(Suppl 1):CO-49

Introduction: The potential benefit of extracorporeal cardiopulmonary resuscitation (ECPR) for patients with refractory cardiac arrest (CA) remains unsettled.

Patients and methods: Retrospective analysis of a database from our prospective observational cohort of CA patients including all consecutive adult patients referred to the Department of Intensive Care with CA between January 2012 and December 2017. The decision to initiate ECPR was dependent on the attending physician and initiated by the ECPR team, which is composed of ICU physicians. A propensity score was derived using a logistic regression model, including characteristics that varied between groups with a p < 0.10 and others that potentially related to outcome. The primary outcomes were survival to ICU discharge and 3-month favorable neurologic outcome, assessed by the Cerebral Performance Categories (CPC) of 1–2.

Results: On a total of 635 CA patients admitted over the study period (ECPR, n = 112), 80 patients with ECPR were matched to 80 patients who underwent CCPR (median age- 57 years—out-of-hospital CA, 62%). Time from arrest to termination of CPR (i.e. return of spontaneous circulation—ROSC, ECMO initiation or death) was 54 ± 22 and 54 ± 19 min in the ECPR and CCPR groups, respectively. ROSC rates were 77 80 (96%) and 30 (38%) in the two groups (p < 0.001). Survival at ICU discharge was 18 80 (23%) vs. 14 80 (18% + p = 0.46), respectively, and 3-month favorable outcome (17 80, (21%) and 9 80, 11%), receptively (p = 0.13). Kaplan–Meier survival analysis showed a better neurological outcome in the matched ECPR group than the CCPR group (log-rank p = 0.01).

Conclusion: ECPR after cardiac arrest can be associated with an improved long-term neurological outcome.

CO-50 Unexpected cardiac arrest occurring in the ICU- preliminary results of a French multicenter study (ACIR)

Maxime Leloup (speaker), Isabelle Briatte, Alice Langlois, Alexandre Herbland, Olivier Lesieur, Study Group Acir

Hôpital Saint Louis, La Rochelle, FRANCE
Correspondence: Maxime Leloup -

Annals of Intensive Care 2019, 9(Suppl 1):CO-50

Introduction: To our knowledge, no study mapped the epidemiology of in-ICU cardiac arrest (CA) in France. Our survey aims to describe demographics, management, vital and neurological prognosis of patients concerned.

Patients and methods: “ACIR” (in French- Arrêt Cardiaque Inattendu en Réanimation) is a prospective observational study implemented in 45 French ICUs throughout 2017. All victims of unexpected CA with resuscitation attempt (chest compression, adrenaline and or electric shock) once admitted to the ICU were included in the “in-ICU CA” chart. The data collected comprises medical history, circumstances and ongoing treatments at the time of the event, resuscitation maneuvers that were performed, survival rate and Cerebral Performance Category (CPC) score.

Results: Of the approximately 30000 patients admitted over the study period, 677 (69% men, 68 ± 13 years, SAPS2 71 ± 25) endured an unexpected CA. Among them, 36% had been hospitalized for shock, 25% for respiratory distress and 15% for out-of-ICU CA. One quarter had 3 or more organ failures. Half of the CA occurred within the first 24 h from admission and 58% during procedures deemed at risk (mainly endotracheal intubation 15%, dialysis 9%, nursing care 6%, and intra-hospital transport 4%). A recognizable etiology was identified in 79% of CA- 18% were attributed to current life-sustaining interventions (mechanical ventilation 7%, circulatory assistance 7%, drug-related complications 4%), and 82% to life-threatening conditions (mainly hypoxemia 33%, metabolic disorder 23% and hypovolemia 18%).

The cardio pulmonary resuscitation was started without delay. A shockable rhythm (ventricular tachycardia or fibrillation) was diagnosed in 18% of cases, with the first shock delivered within 2 [0–3] minutes. In cases of non-shockable rhythm (asystole, pulseless bradycardia), the first adrenaline injection was given at 1 [0–2] minute. Return of spontaneous circulation (ROSC) was achieved in 477 677 (70%) patients, of whom 146 677 (21.6%) were discharged alive from hospital. The number of CA survivors with a CPC score of 1 or 2 (considered satisfactory) was 135 146 (92%) at hospital discharge and 117 128 (91%) 6 months after CA.

Conclusion: In-ICU CA occurred in severely ill patients, during the first few days of hospitalization and more than half of cases during at-risk procedures. One-fifth was attributed to the life-sustaining intervention(s) in place. Despite a high rate of ROSC (70%), only 21.6% of victims were discharged alive from the hospital. However, survivors maintained a good neurological status over time.

CO-51 Temporal trends in TTM use after cardiac arrest and association with outcome in a large registry

Jean-Baptiste Lascarrou (speaker)1, Florence Dumas2, Richard Chocron2, Wulfran Bougouin3, Stephane Legriel3, Nadia Aissaoui3, Nicolas Deye3, Eloi Marijon3, Xavier Jouven3, Alain Cariou3

1Service de Médecine Intensive Réanimation, Nantes, FRANCE; 2Emergency department, Cochin Hotel Dieu Hospital, APHP, Paris, FRANCE; 3Paris Sudden Death Expertise Center, Paris, FRANCE
Correspondence: Jean-Baptiste Lascarrou -

Annals of Intensive Care 2019, 9(Suppl 1):CO-51

Introduction: Emphasis regarding the importance of early cardiopulmonary resuscitation (CPR) and defibrillation led to an improvement in outcome after cardiac arrest (CA). Meanwhile, recent data raised some concerns regarding the benefit of targeted temperature management (TTM) at 33 °C for comatose cardiac arrest survivors. This may have resulted in a loss of interest in TTM in these patients, with uncertain consequences on outcome. We use data from the Sudden Death Expertise Center (SDEC) of the Great Paris area to assess the relationship between changes over time in the use of TTM and patients’ outcome after adjustment for potential confounders.

Patients and methods: We used data prospectively collected in SDEC registry between May 2011 and December 2016. Variables were collected according to Utstein style. All non-traumatic OHCA patients with stable return of spontaneous circulation (ROSC) were included in the analysis. We compared patients’ characteristics and outcome according to TTM use. We performed a multivariable logistic regression using survival at ICU discharge as the main endpoint.

Results: During study period, 3667 patients were retained in the analysis, of whom 1794 received TTM. As compared with controls, patients with TTM were significantly more frequently male and younger + CA occurred more frequently in a public place + a shockable rhythm was more frequent and the proportion of patients with a short no-flow (< 3 min) was greater (all P < 0.001). Regarding time trends, gender, age and location of CA did not change according to the year of inclusion in the registry. Bystander CPR increased from 56% in 2011 to 82% of patients in 2016 (P < 0.001), the rate of shockable rhythms increased from 40% to 51% (P = 0.005). The rate of patients with a no-flow > 3 min decreased from 58% to 43% (P < 0.001) (Figure 1). TTM use decreased over time, from 55% in 2011 to 48% in 2016 (P < 0.001). Meanwhile, survival rate at ICU discharge increased significantly from 20% to 28% of patients (P = 0.018) (Figure). In multivariate analysis, year of CA occurrence was associated with outcome at ICU discharge. The TTM use decrease over years was not driven by any particular Utstein criteria.

Conclusion: We report an improvement over time in ICU survival of post-CA patients, which appears mostly associated with an increase in bystander CPR and higher rate of shockable rhythms. TTM use declined over years in all subgroups of CA patients with no evident consequences regarding outcome.

CO-52 Incidence and outcome of mesenteric ischemia after cardiac arrest

Marine Paul (speaker)1, Paul Jaubert1, Guillaume Savary1, Stephane Andre1, Pierre Jaquet1, Arnaud Roccabianca1, Jean François Llitjos1, Sarah Benghanem1, Wulfran Bougouin2, Frederic Pene1, Jean Paul Mira1, Florence Dumas3, Alain Cariou1

1Medical icu cochin hospital, Paris, FRANCE; 2INSERM U970 (team 4), Paris, FRANCE; 3Emergency department cochin hospital, Paris, FRANCE
Correspondence: Marine Paul -

Annals of Intensive Care 2019, 9(Suppl 1):CO-52

Introduction: Post-resuscitation shock has a high incidence after cardiac arrest (CA) due to the whole-body ischemia–reperfusion and may be complicated by multiple organ failure (MOF) and death. Mesenteric ischemia (MI) may occur in this setting but incidence and outcome are poorly documented.

Patients and methods: We performed a retrospective monocentric study aiming to describe characteristics and outcome of MI after CA. All consecutive patients admitted in a tertiary CA center between 2007 and 2017 with a confirmed MI during their ICU stay were studied. MI was confirmed by endoscopy, abdominal CT-scan or surgical laparotomy.

Results: Among 1503 CA post-CA patients, MI was clinically suspected in 59 patients (4%) and confirmed in 28 59 patients. Median age was 62 years, median time from CA to cardio-pulmonary resuscitation (CPR) was 3 min (0–5), and time from CPR to return of spontaneous circulation was 22 min (18.3–29.5). Patients had an initial shockable rhythm in 57%, they received epinephrine during CPR in 82% and 93% of patients developed a post-resuscitation shock. Median initial blood lactate was 8.5 mmol/L. Delay between CA and the diagnosis of MI was 2 days (2–4), mostly because of clinical abdominal symptoms (79%) including hemorrhage and feeding intolerance, confirmed with CT scan abnormalities (54%) and lower digestive endoscopy (39%). Despite surgical intestinal resection performed in 7 28 patients, ICU mortality was 96%, as only 1 patient survived with good neurologic outcome. Cause of death was the MOF itself (41%) or post-anoxic brain injury for 37% despite a recovery after the initial shock.

Conclusion: In this cohort, we observed that MI after CA is an uncommon complication associated with a very poor outcome. Efforts should be made in order to obtain an earlier recognition of MI in patients with preserved brain function. These preliminary results need to be confirmed in a multicentric study.

CO-53 BOTUREA STUDY - A descriptive retrospective multicenter study of severe adult cases of botulism between 2000 and 2017 in French intensive care units

Nathalie Courtois (speaker)1, Laurent Argaud2, Pierre Asfar3, Paul Mercury4, Juliette Pocquet5, Hervé Dupont6, Julien Charpentier 7, Djillali Annane8, Cédric Bretonniere9, Jean-Christophe Callahan1, Mickael Landais1, Christophe Guitton1

1Réanimation Médico-Chirurgicale et USC, Le Mans, FRANCE; 2Service de Réanimation Médicale, Lyon, FRANCE; 3CHU Angers-Service de Médecine Intensive- Réanimation et Médecin Hyperbare, Angers, FRANCE; 4Service de Réanimation Médico-Chirurgicale et USC, Bastia, FRANCE; 5Service de Médecine Intensive - Réanimation, Tours, FRANCE; 6Service de Réanimation Polyvalente, Amiens, FRANCE; 7Service de Médecine Intensive - Réanimation Cochin, Paris, FRANCE; 8Service de Médecine Intensive - Réanimation Garches, Paris, FRANCE; 9Service de Médecine Intensive - Réanimation, Nantes, FRANCE
Correspondence: Nathalie Courtois -

Annals of Intensive Care 2019, 9(Suppl 1):CO-53

Introduction: Botulism is a rare potentially severe neuroparalytic disease caused by neurotoxins produced in most cases by Clostridium botulinum. Treatment mainly consists in life support techniques, and some specific medications such as antitoxins. In France, there are no guidelines regarding the use of these antitoxins, and they are only available with a Temporary Authorization for Use. The primary goal of this study was to describe and evaluate morbidity and mortality of severe adult botulism in ICUs in France. Our secondary goal was the evaluation of current patient care in French ICUs, especially concerning administration of antitoxins.

Patients and methods: We conducted a retrospective observational multicenter study in French ICUs. Among 197 adult ICUs, 93 units agreed to participate. Patients’ data were collected through a standardised CRF. We included patients aged above 15 years of age who were admitted in an ICU between January 2000 and June 2017 for an acute botulism. Study protocol was approved by the SRLF ethics committee. Data collection was conducted according French law and CNIL recommendations.

Results: Among the participating units, 59 didn’t identify any botulism cases and 6 were unable to collect patient’s data. Twenty-eight ICUs identified 52 patients. At admission, median (IQR) age was 49.2 years (37–65) and 51% of patients were male. 94.1% of the patients had foodborne botulism. Median SAPS II was 21 (17–26). For clinical presentation, ocular symptoms (especially diplopia 84% and ptosis 67%) and oro-pharyngeal symptoms (especially swallowing impairment 78%) were the most frequent, and were usually the first to be described. 65% of patients were intubated. Median duration of mechanical ventilation was 13 days (0–53) with a maximum of 333 days. Tracheotomy was performed for 36% of patients. Median ICU hospitalisation duration was 27 days (8–47). The current botulism episode was associated with sequelae evaluated by the modified Rankin Scale. At ICU discharge, hospital discharge and last follow-up, 43, 22 and 10% of patients respectively had a score above 3. Two patients died during their hospitalisation (1 during ICU stay). Twenty-five patients (48.1%) were treated with botulism antitoxin. We were unable to show any significant difference in outcomes between patients treated or not treated with antitoxin.

Conclusion: Although botulism rarely causes death nowadays, it is a potentially severe disease, with a high morbidity and possible extended hospitalisation and mechanical ventilation. Recent studies are in favour of the use of antitoxins, and clear recommendations are needed.

CO-54 First-in patients TREM-1 Pathway Inhibitory Peptide in Septic Shock- The MOT-C-201 Clinical Trial Results

Bruno Francois (speaker)1, Xavier Wittebole2, Ricard Ferrer3, Jean-Paul Mira4, Thierry Dugernier5, Sébastien Gibot6, Marc Derive 7, Peter Pickkers8, Jean-Jacques Garaud7, Miguel Sanchez9, Margarita Salcedo-Magguilli7, Pierre-François Laterre2

1Medical-Surgical ICU department and Inserm CIC1435, CHU, Limoges, FRANCE; 2Department of Critical Care Medicine, St Luc University Hospital, Université Catholique de Louvain, Brussels, BELGIUM; 3ICU department, Vall d’Hebron University Hospital, Barcelona, SPAIN; 4Medical ICU, Cochin Hotel-Dieu, AP-HP, Paris, FRANCE; 5ICU department, Clinique St. Pierre, Ottignies-Louvain-La-Neuve, BELGIUM; 6Medical ICU department, Hospital Central, CHU, Inserm U1116, Nancy Medical Faculty, Nancy, FRANCE; 7Inotrem SA, Paris, FRANCE; 8ICU department, Radboudumc Hospital, Nijmegen, THE NETHERLANDS; 9ICU department, Hospital Clínico San Carlos, Madrid, Spain, Madrid, SPAIN
Correspondence: Bruno Francois -

Annals of Intensive Care 2019, 9(Suppl 1):CO-54

Introduction: Nangibotide peptide is a specific TREM-1 inhibitor. In preclinical septic shock models, nangibotide was able to restore appropriate inflammatory response, vascular function, and improved survival. In phase I, nangibotide was found to be safe and well tolerated up to the highest dose.

Patients and methods: International, multi-center phase IIa randomized, double-blind, two-stage, placebo-controlled study (NCT03158948). Main inclusion criteria were septic shock according to Sepsis 3 definition and nangibotide to be initiated within 24 h of shock onset. Patients were randomized to receive either placebo, 0.3, 1 or 3 mg kg h of nangibotide. Study drug was infused until end of vasopressors + 12 h or up to 5 days. Safety data were reviewed by an independent Data Safety Monitoring Board (DSMB). Primary endpoint was safety and tolerability. Patient follow-up period was 90 days.

Results: 50 patients were randomized and 49 treated (1 patient died before dosing). All groups were well balanced in terms of baseline characteristics, except for APACHE II score which tend to be non-significantly lower in placebo group. Primary infection source was 40% abdominal, 50% pulmonary and 10% urinary.

Nangibotide was safe and well tolerated in all groups. The DSMB did not raise any safety concern. Number of SAEs AEs and number of patients with SAEs AEs was comparable between all groups. Most frequent AEs were atrial fibrillation, anemia, pleural effusion and thrombocytopenia.

A trend toward a decrease in circulating levels of pharmacodynamic markers was observed in nangibotide-treated patients. All-cause mortality at day-28 was 14% (5/37) in pooled nangibotide groups and 25% (3/12) in placebo group. In the subgroup with sTREM-1 levels above median, the day-5 mortality was calculated as 40% (2/5) and 20% (4/20) in placebo and nangibotide groups respectively. In this group with high sTREM-1 levels, a trend towards an increase in organ support free days alive was seen for nangibotide-treated patients versus placebo.

Conclusion: Nangibotide was shown to be safe and well tolerated in septic shock patients. Although this small exploratory study was not powered to conclude on efficacy, a non significant lower mortality was observed in the nangibotide group. These results support the need of a larger study to demonstrate the role of nangibotide in the treatment of septic shock and the exploration of sTREM-1 as potential efficacy predictive biomarker for nangibotide.

CO-55 Anxiety and depression in patients admitted in ICU for more than 48 h- results of the multicentre, observational, prospective MOOD-ICU study

Camille Vissac (speaker)1, Saad Nseir2, Françis Schneider3, Jean-Francois Timsit4, Alexandre Boyer5, Jean-Paul Mira6, Vincent Peigne 7, Carole Schwebel8, Kada Klouche9, Ferhat Meziani10, Walter Picard11, Emmanuelle Mercier12, Emmanuel Pontis13, Charles-Edouard Luyt14, Daniel Silva15, Aurélie Le Thuaut16, Jean Reignier17, Group Mood-Icu18

1Médecine Intensive-Réanimation, Nantes, FRANCE; 2Médecine Intensive Réanimation, Lille, FRANCE; 3Chu-Médecine Intensive Réanimation, Strasbourg, FRANCE; 4Médecine Intensive Réanimation, Paris, FRANCE; 5Médecine Intensive Réanimation, Bordeaux, FRANCE; 6Médecine Intensive Réanimation, Paris, FRANCE; 7Médecine Intensive Réanimation, Chambéry, Chambéry; 8Médecine Intensive Réanimation, La Tronche, FRANCE; 9Médecine Intensive Réanimation, Montpellier, FRANCE; 10Médecine Intensive-Réanimation, Strasbourg, FRANCE; 11Médecine Intensive-Réanimation, Pau, FRANCE; 12Médecine Intensive-Réanimation, Tours, FRANCE; 13Médecine Intensive-Réanimation, Rennes, FRANCE; 14Médecine Intensive-Réanimation, Paris, FRANCE; 15Médecine Intensive-Réanimation, Saint-Denis, FRANCE; 16Maison de la recherche en santé, Nantes, FRANCE; 17CHU, Nantes, FRANCE; 18Médecine Intensive-réanimation, Nantes, FRANCE
Correspondence: Camille Vissac -

Annals of Intensive Care 2019, 9(Suppl 1):CO-55

Introduction: Several studies have been published on psychiatric disorders after intensive are unit (ICU) and risk factors associated with these disorders. No studie have evaluated the frequency of anxiety and depression disorders during the ICU stay.

Patients and methods: The main objective of this prospective, multicentre, observational study was to assess rates of patients with anxiety and or depression in ICU. The secondary objectives were to assess rates of patients with anxiolytic and antidepressant drugs and to identify factors associated with anxiety and depression in ICU. All patients hospitalized since more than 48 h in ICU at study day were eligible. The evaluation of patients anxiety and depression symptoms was performed by the bedside physician and the nurse caring for the patient at the day of the study.

Results: 81 French ICU participated to the study, and 799 patients were included. 389 patients were reported as having anxiety and or depression by caregivers (51.0%, 95% CI [47.3–54.6]). 21.2% and 5.1% of the patients included in the study received drugs to treat anxiety and depression, respectively. Risk factors independently associated with anxiety and or depression were- a history of anxiety or depressive syndrome (OR 1.75, 95% CI [1.19–2.57], p = 0.004), announcement of severe illness and or loss of autonomy during ICU stay (OR = 2.12, 95% CI [1.32–3.43], p = 0.002), having tracheostomy performed during ICU stay (OR = 2.04, 95% CI [1.08–3.85], p = 0.03), administration of antipsychotic medication in ICU (OR = 1.94, 95% CI [1.20–3.14], p = 0.003), and reading of a book by cargivers or relatives during the ICU stay (OR = 2.01, 95% CI [1.08–3.74], p = 0.03). Organ failure was independently associated with less anxiety and or depression (OR = 0.57, 95% CI [0.33–0.99], p = 0.03).

Conclusion: Anxiety and depression symptoms were common in critically ill patients during their ICU stay. History of anxiety or depression, announcement of severe illness and or loss of autonomy, tracheostomy, antipsychotic medication and reading of book by caregivers were associated with increased risks of anxiety and depression disorders in ICU. However, severity of critical illness was associated with decreased risk of anxiety and depression. Future studies are required to assess the relationship between anxiety and depression disorders during and after ICU stay.

CO-56 Assessment of care-givers’ experience of end of life in the ICU

Florence Boissier (speaker)1, Valérie Seegers2, Stéphane Legriel3, Alain Cariou4, Samir Jaber5, Jean-Yves Lefrant6, Bernard Floccard 7, Anne Renault8, Isabelle Vinatier9, Armelle Mathonnet10, Danielle Reuter11, Olivier Guisset12, Christophe Cracco13, Amélie Seguin14, Jacques Durand-Gasselin15, Beatrice Eon16, Marina Thirion17, Jean-Philippe Rigaud18, Bénédicte Philippon-Jouve19, Laurent Argaud7, Renaud Chouquer20, Mélanie Adda21, Laurent Papazian21

1Réanimation médicale, CHU, Poitiers, FRANCE; 2Departement of Clinical Research, Integrated Center of Oncology, Angers, FRANCE; 3Service de réanimation, CH de Versailles André Mignot, Le Chesnay, FRANCE; 4Service de réanimation médicale, hôpital Cochin APHP, Paris, FRANCE; 5Service de réanimation, CHU Saint Eloi, Montpellier, FRANCE; 6Service de réanimation chirurgicale, CHU Carémeau, Nîmes, FRANCE; 7Service d’anesthésie réanimation, Hospices civils de Lyon Edouard Herriot, Lyon, FRANCE; 8Service de réanimation, CHU Cavale Blanche, Brest, FRANCE; 9Service de réanimation, CH Vendée, La Roche-Sur-Yon, FRANCE; 10Service de réanimation, hôpital de la Source, Orléans, FRANCE; 11Groupe de recherche FAMIREA, Hôpital Saint Louis APHP, Paris, FRANCE; 12Service de réanimation, CHU de Bordeaux Saint André, Bordeaux, FRANCE; 13Service de réanimation, CH, Angoulême, FRANCE; 14Service de réanimation, CHU, Caen, FRANCE; 15Service de réanimation, Hopital Sainte Musse, Toulon, FRANCE; 16Service de réanimation, Hopital La Timone APHM, Marseille, FRANCE; 17Service de réanimation, CH, Argenteuil, FRANCE; 18Service de réanimation, CH, Dieppe, FRANCE; 19Service de réanimation, CH, Roanne, FRANCE; 20Service de réanimation, CH Annecy, Annecy, FRANCE; 21Service de réanimation, Hopital Nord APHM, Marseille, FRANCE
Correspondence: Florence Boissier -

Annals of Intensive Care 2019, 9(Suppl 1):CO-56

Introduction: An increasing number of deaths occur in the ICU. In this setting, research has focused on describing, understanding, and improving end-of-life care as well as improving the patients and the families’ experience. The aim of this study was to describe and evaluate the experience of physicians and nurses involved in end-of-life situations in the ICU and describe the factors associated with a poor or good score.

Patients and methods: A 15-item questionnaire was validated in a multicenter prospective study conducted from July 2011 to July 2013 in 41 French ICUs. In each ICU, consecutive adults who died after at least 48 h after admission in the ICU were included. The physician and the nurse in charge of the patient at the time of death were asked to complete the questionnaire within 24 h after the death. The psychometric validation was conducted using two datasets- a learning cohort and a validation cohort.

Results: Among 475 patients, 398 nurse scores and 417 physician scores were analyzed. The median global CAESAR score for nurses was 62 75 and 64 75 for physicians. Lower nurse scores were significantly associated with the following factors- conflict with physicians, pain managed by nurses alone rather than with physicians, death disclosed to the family by phone or upon arrival rather than at the bedside and use of invasive treatment at the end of life. Higher physician scores were significantly associated with the following factors- availability of a dedicated information room, joint family information with the nurse, total rather than partial family information, joint implementation with the nurse of the decision to withdrawing withholding therapies, open visiting policy. Higher scores were also associated with patients’ characteristics- McCabe ≤ 1, not immunocompromised, no dementia, no liver failure, no haematological disease and no use of psychotropic medication. Last, higher score were associated with End Of Life conditions- decision to withdrawing withholding therapies, no cardiopulmonary resuscitation, death disclosed to the family at the bedside rather than by phone or upon arrival at the ICU.

Conclusion: We described and validated a new instrument for assessing the experience of physicians and nurses involved in end-of-life situations in the ICU. This study shows important areas for improving practices, including adapted level of intensive care, quality teamwork, quality communication and implication of family members.

Flash Communications

F-01 Efficiency of a combined geriatric and intensivist evaluation to identify clusters of elderly patients admitted to the intensive care unit - the SENIOREA study

Julien Demiselle (speaker)1, Guillaume Duval1, Jean François Hamel1, Anne Renault4, Dominique Perrotin5, Laurent Martin-Lefèvre6, Dominique Vivier 7, Daniel Villers8, Montaine Lefèvre9, René Robert10, Philippe Markowicz11, Sylvain Lavoué12, Anne Courte13, Eddy Lebas14, Stéphanie Chevalier15, Cédric Annweiler1, Nicolas Lerolle1

1CHU, Angers, FRANCE; 4CHU, Brest, FRANCE; 5CHU, Tours, FRANCE; 6CHD Vendée, La Roche Sur Yon, FRANCE; 7CH, Le Mans, FRANCE; 8CHU, Nantes, FRANCE; 9CH, Morlaix, FRANCE; 10CHU, Poitiers, FRANCE; 11CH, Cholet, FRANCE; 12CHU, Rennes, FRANCE; 13CH, Saint-Brieuc, FRANCE; 14CH, Vannes, FRANCE; 15CH, Saint-Malo, FRANCE
Correspondence: Julien Demiselle -

Annals of Intensive Care 2019, 9(Suppl 1):F-01

Introduction: Improving the care for older patients in the ICU is a challenge both in terms of survival and of quality of life (QoL). We set up a prospective observational to assess the effectiveness of a combined geriatric and intensivist initial evaluation in determining clusters of older patients and the one-year outcome of these clusters.

Patients and methods: Patients aged 75 years and older admitted to 13 ICUs in Western France and requiring mechanical ventilation from September 2012 to December 2013 were included. Proxy members were consulted at ICU admission using a predefined comprehensive geriatric assessment. Survival was assessed at one year and comprehensive geriatric assessment was performed on place of living in survivors. Clustering was performed based on baseline characteristics and ICU admission parameters using Jaccard similarity measure.

Results: Five hundred and one patients were included in the analysis. Age was 80 ± 4 years, 53.5% were male. Ninety-two percent of patients were living at home before ICU admission, mean comorbidity burden assessed with CIRS-G score was 8.3 ± 4.3, 78.9% of patients had ADL score ≥ 5 and SOFA at ICU admission was 7 ± 3.7. Clustering resulted in three groups. Group 1 (n = 166) was characterized by older age, lower comorbidity burden and intermediate acute severity. Group 2 (n = 111) was characterized by lower age, intermediate comorbidity, and higher acute severity. Group 3 (n = 98) had higher age, higher comorbidity, and lower acute severity. 126 patients could not be clustered due to incomplete data. Clusters experienced different trajectories throughout ICU stay with higher ICU mortality for group 2, and the highest amount of organ support. Moreover, clusters evolution was different after hospital discharge. Overall, one-year survival was 53.8%, mortality was higher in group 2 and 3 when compared to group 1 (p = 0.03, see figure). 163 patients (62% of survivors) underwent one-year at-home evaluation. If a moderate decrease in independency was observed, perceived QoL remains satisfactory with 88% of patients indicated that they were happy or very happy. No difference in one-year geriatric evaluation was observed between clusters.

Conclusion: Comprehensive geriatric assessment and acute severity parameters are useful to identify three different patterns of patients with different outcomes. Such patterns may be useful for design future interventional studies. Long term survival of older patients after an ICU stay has improved compared to previous studies.

F-02 Unique blood culture in ICU is associated with a dramatic reduction of blood culture contamination rates: an interrupted time-series study

Rafael Mahieu (speaker)1, Matthieu Eveillard1, Valérie Seegers1, Carole Lemarié1, Vincent Dubée2, Alain Mercat1, Achille Kouatchet1

1Médecine Intensive et Réanimation, Médecine Hyperbare, CHU, Angers, FRANCE; 2Infectious and Tropical Diseases Department, Angers University Hospital, Angers, FRANCE
Correspondence: Rafael Mahieu -

Annals of Intensive Care 2019, 9(Suppl 1):F-02

Introduction: Blood cultures (BC) remain the first-line tool for identification of the causative agent in patients with severe infection. However, several parameters affect diagnostic performances of BC. Sensibility depends on volume of collected blood + false-positive BC may arise from blood contamination during venipuncture. Our objective was to assess the effect of a unique high-volume blood culture (UBC) strategy on the rate of BC contamination and pathogen detection.

Patients and methods: An interrupted time-series study was conducted in a medical 24-bed ICU from January 2013 to December 2016. After the UBC protocol was implemented on January 2015, all patients with suspected bacterial infection had 40 mL of blood collected by a single phlebotomy. Ten milliliters of blood were inoculated into two BacT Alert FA aerobic bottles and two BacT Alert FN anaerobic bottles. Performing additional blood culture within 48 h was strongly discouraged unless a new bloodstream infection was suspected. Nurses’ and physicians’ knowledge on BC collection and practice skills were reinforced through an educational program on December 2014. Protocol adherence was evaluated at 2 and 12 months through monitoring of BC volume.

Results: During the study period, there were 3497 ICU stays with at least one BC. No difference was observed between the two periods regarding severity (IGS2 and SOFA at admission) or in-ICU death. After the educational program, the median volume of blood culture per bottle increased from 3.2 mL [95% confidence interval (CI) 3–3.5] to 9.1 mL [95% CI- 8.7–9.5] at 2 months and decreased to 7.1 mL [95% CI- 6.3–8] at 12 months (p < 0.001). During the UBC period, the rate of bloodstream infection per patient (bacteremia with a true pathogen) was 14.3%, as compared with 14.7% before (p = 0.1). The rate of BC contamination amongst positive BC decreased by 55.8% (from 43% before protocol implementation to 19% after; p < 0.001). The median number of BC bottles collected per week decreased by 58.5% (from 188 [interquartile range 149–235] before UBC to 78 [64–90] after protocol implementation), corresponding to a saving of 37,000 € per year.

Conclusion: Implementation of a UBC protocol was associated with an important decrease in rate of BC contamination and BC-associated costs, without affecting pathogen detection rate.

F-03 Risk factors of post-traumatic stress disorder (PTSD) among ICU survivors

Marwa Zghidi (speaker), Imen Ben Saida, Said Kortli, Hend Zorgati, Abdelbaki Azouzi, Khaoula Meddeb, Ahmed Khedher , Mohamed Boussarsar

Farhat Hached University Hospital, Medical Intensive Care Unit, Sousse, TUNISIA
Correspondence: Marwa Zghidi -

Annals of Intensive Care 2019, 9(Suppl 1):F-03

Introduction: Extended follow-up of ICU survivors has shown that many patients suffer from long-term physical and psychological sequalae. PTSD is an increasingly psychiatric disorder reported in ICU survivors. The aim was to determine the frequency of PTSD and its predictors.

Patients and methods: It is a mixed method study conducted in a medical ICU from January 2017 to January 2018. Data were obtained from medical records. At 3 months post-ICU discharge, patients were contacted by phone to complete the Impact Event Scale- Revised questionnaire (IES-R). Univariate and logistic regression analyses were used to identify variables independently associated with IES-R (≥ 33).

Results: 393 patients were admitted during the study period. 191 (48.6%) were discharged alive, 56(29.3%) were never successfully contacted and 21(11%) died within the 3 months’ period. 114 patients fulfilled the inclusion criteria. Patients’ characteristics were - mean age 56.29 ± 17.88 years; male, 66(57.9%); median Charlson comorbidity index, 1 [1–2]; mean SAPSII, 25.04 ± 12.1; invasive mechanical ventilation (IMV), 47(41.2%) and vasopressors use, 30(26.3%). The median duration of IMV and length of stay were respectively 0 days [0–4] and 6 days [4–10]. 25 (21.9%) patients met diagnostic criteria for PTSD. In univariate analysis, risk factors of PTSD were respectively for PTSD group and controls - age (42.7 ± 19 vs 60.1 ± 15.6, p = 0.000), female sex (64% vs 36%, p = 0.012); Alcoholism (24% vs 7.9%, p = 0.035); IMV (76% vs 31.5%, p = 0.000); Sedative use (76% vs 31.5%, p = 0.000); polyneuromyopathy (28% vs 2.2%, p = 0.000); delirium (16.7% vs 1.1%, p = 0.008); catheterization (56% vs 21.6%, p = 0.001); duration of IMV (6.58 ± 7 vs 2.32 ± 6.3, p = 0.005) and physical restraint (72% vs 23.6%, p = 0.000). On multivariable logistic regression, age (OR, 0.95 + 95%CI, [0.92 -0.98]; p = 0.001), female sex (OR, 3.88; 95%CI, [1.08 -13.9]; p = 0.038), physical restraint (OR, 6.27; 95%CI, [1.66–23.67]; p = 0.007) and polyneuromythy (OR, 11.15; 95%CI, [1.5–80.1]; p = 0.01) were identified as risk factors independently associated to PTSD.

Conclusion: PTSD is common in ICU survivors. Physical restraint, female sex, polyneuromyopathy and younger age were the only factors independently associated to PTSD.

F-04 Help, my hospital is burning - Experience return on fire in the Guadeloupe University Hospital (GUH)

Frédéric Martino (speaker), Pascale Piednoir, Bertrand Pons, Benjamin Madeux, Khalid Elkoun, Elain Elie, Roland Lawson, Rémi Lazdunsk, Rémi Malhomme, Cyril Marimoutou, Juliette Masse, Alessia Napoleone, Eric Paris, Jean-Baptiste Putegnat, Asma Benguerrah, Morgane De Oliveira, David Fedida, Philippe Le Noach, Zakaria Mahi, Mathilde Pibarot, Mathieu Tournay, Killian Yao, Mohamed Zidani

CHU - Abymes, Pointe-À-Pitre, FRANCE
Correspondence: Frédéric Martino -

Annals of Intensive Care 2019, 9(Suppl 1):F-04

Introduction: The GUH fire causing its total evacuation is an unprecedented event in France. We report factual elements of the first days after the disaster involving intensive care unit (ICU) activities with specific proposed imposed responses, as it is the sole ICU of Guadeloupe island (near half million people).

Patients and methods: Data are extracted from ICU medical files, experience reports, and Emergency department softwares.

Results: The fire started on 28.11.2017 around 13:45 on the 2nd floor of the building, between the ICU (1st floor) and operating room (3rd floor). At 14:00, the ICU (25 occupied beds over 26, including 12 ventilated patients) was invaded by a thick smoke, not triggering the fire alarm. The evacuation of the ICU was decided by doctors before the activation of the official evacuation plan (> 1 h) in absence of any administrative instructions.

The 12 ventilated patients were relocated at the radiology level (access to medical gases), 8 non-ventilated patients evacuated to the outdoor car park, 5 patients to other services. At 17:00 all patients were extracted from the ICU. Patients were then evacuated to other centers- 3 ventilated patients to the Basse Terre public hospital (Post Anesthesia Care Unit), 9 ventilated patients and 8 non-ventilated patients to a private hospital located at 11 km, into a unit of 12 ambulatory surgery beds, thus the same day re-affected as ICU with 2 to 3 patients per room.

At 22:53, all ICU patients were evacuated from the GUH. On 29.11.2017, 11 patients were prioritized for medical evacuation to the Martinique University Hospital (MUH) thanks to military vectors - 8 patients simultaneously (CASA military aircraft), and 3 by medical helicopter (SAMU). No worse initial progression was noted, except for 1 patient (septic shock needing intubation and norepinephrine) at the radiology level. ICU activities were heavily impacted during this period by difficulties related to emergency services, medico-surgical services and support structures (laboratories, radiology, pharmacy) rearrangement.

Conclusion: This disastrous never experienced situation had a limited immediate impact (no deaths) due to the fast management of patient’s transfer, related to ICU doctors initiatives. Regional support (MUH and military vector) helped to propose an immediate response for the most severe ICU patients. Mid- and long-term consequences are significant for evacuated patients as well as for the Guadeloupe care provision, in term of ICU and other hospital components, related to the distance away from France.

F-05 Evaluation and description of alarms in an Intensive Care Unit, Improvement of professional practices- optimizing monitoring and reducing noise pollution

Pierre-Yves Delannoy (speaker)1, Lydie Martel2, Nicolas Boussekey2, Damien Thellier2, Hugues Georges2, Olivier Robineau2, Olivier Leroy2

1CH, Marcq En Baroeul, FRANCE; 2CH, Tourcoing, FRANCE
Correspondence: Pierre-Yves Delannoy -

Annals of Intensive Care 2019, 9(Suppl 1):F-05

Introduction: The excess of alarms in intensive care has multiple consequences- the noise generated alterate the quality of patient’s stay and participate in the genesis of the phenomenon called alarm fatigue. This, by altering the quality of surveillance, directly impacts patient safety in a pejorative term. The objective of this project is to improve the quality of care and working conditions in intensive care unit in Tourcoing by optimizing the monitoring of our patients through a collaboration between the medical team and the company providing the monitoring products.

Patients and methods: An analysis of alarms from Philips Intelivue monitors for one month is performed, then corrective measures to reduce their number are implemented through a collaborative DMAIC approach (Define, Measure, Analyze, Improve, Control). These corrective measures include a personalization of the monitoring with a medical reflection on the alarm’s thresholds and creation of specific profile patient, training of caregivers on the good use of the monitoring, creation of an alarm culture via these trainings and the creation of communication tools. To these actions are added technical interventions on intensive care ventilators and monitors. The impact of alarms on caregivers and patients present during the analysis period are evaluated using questionnaires. A second alarm analysis assess an evaluation of the effectiveness of the implemented measures.

Results: 127.927 alarms (equivalent to 243 alarms day patient) including 40.118 vital alarms, 58.647 critical and 29.162 techniques are found in the first period. The analysis of the patient questionnaire is not very contributive with 20 usable questionnaires. Among the responders, the majority did not feel any significant discomfort during their stay. The health questionnaire has a response rate of 91% (n = 51). Responders believe that there is a decrease in attention and an increase in reaction time due to false alarms. They also believe that irrelevant alarms disrupt the quality of the patient’s stay and that better alarm management could prevent incidents. Nearly 86% of caregivers say that alarms could not be heard and were missed. In the second period, a 30% reduction in the total number of alarms (90,096 alarms, equivalent to 202 alarms day patient, p = 0.002) is obtained.

Conclusion: By creating and implementing targeted corrective measures with DMAIC approach, a reduction in the number of alarms is achieved. This reduction probably enhance patient safety and improve healthcare environement for patients and caregivers.

F-06 Can we routinely use electronic medical data management systems to drive the prescription of red blood cell transfusion in Canadian and British pediatric intensive care units?

Camille Jutras (speaker)1, Geneviève Du Pont-Thibodeau1, Marisa Tucci1, Simon Stanworth2, Samiran Ray3, Barney Scholefield4, Samuel Kadoury 5, Philippe Jouvet1, Patricia Fontela6, Jacques Lacroix1

1CHU Sainte Justine, Montreal, CANADA; 2NHS Blood & Transplant Oxford John Radcliffe Hospitals, Oxford, UNITED-KINGDOM; 3Great Ormond Street Children’s Hospital, London, UNITED-KINGDOM; 4Birmingham Children’s Hospital, Birmingham, UNITED-KINGDOM; 5Centre de recherche, CHU Sainte-Justine, Montreal, CANADA; 6Montreal Children’s Hospital, Montreal, CANADA
Correspondence: Camille Jutras -

Annals of Intensive Care 2019, 9(Suppl 1):F-06

Introduction: There is evidence that electronic medical data management systems (eMDMS) may help practitioners to improve the appropriateness of red blood cell (RBC) transfusion in hospitalised adults, but there is a lack of data supporting this in children. It is unclear if eMDMS are used in paediatric intensive care units (PICUs). This survey aims to document the availability of eMDMS in British and Canadian PICUs and characteristics of the data recorded.

Patients and methods: An electronic self-administered questionnaire was sent out thrice, using Survey Monkey, from November 2017 to February 2018, to the director of 16 Canadian and 27 British PICUs. Respondents were asked to describe their PICU and to indicate what eMDMS was used in their institution.

Results: Among the 43 units studied, 12 Canadian and 25 British PICUs (37/43 = 86%) answered to the survey. Three of them were excluded because they exclusively took care of premature infants and newborns. Some eMDMS were used in 24 out of the 34 remaining PICUs (71%). The table below details the data collected by these eMDMS in PICUs. Seventeen PICUs (71%) used Windows as their eMDMS operating system. Two types of Electronic Medical Records (EMR) were used to document patient care- IntelliSpace Critical Care and Anesthesia (ICCA, Philips, 9 PICUs) and Allscripts Professional (1 PICU) (total- 42%). EMR name was unknown in 14 instances. Electronic medical data were known to be stored in a Structured Query Language (SQL) server in seven PICUs (29.17% %).

Conclusion: Data management by eMDMS is common in British and Canadian PICUs. Computerized physician order for RBC transfusion is reported to be available in 79.17% of these PICUs. Using eMDMS to enable an electronic medical decision support tool for RBC transfusion should be feasible in most PICUs, and may facilitate efficient data collection for clinical trials.

Table 1. Data management by eMDMS in 24 PICUs





Data collected by eMDMS


 Main diagnosis

22 (92%)

2 (8%)



22 (92%)

2 (8%)


 Symptoms and signs

21 (88%)

3 (13%)


 Laboratory data

24 (100%)



 Data on red blood cell (RBC) transfusions

24 (100%)



 Plasma and/or platelet transfusions

24 (100%)




20 (83%)

4 (17%)


 Dates when events happen†

23 (96%)

1 (4%)


Computerized physician order (CPO) entry


 Prescription of RBC transfusion

19 (79%)

5 (21%)


 Prescriptions other than RBC transfusion

19 (79%)

5 (21%)


 Example of events: transfusion, adverse outcomes, death…

Table 1. Availability of physiologic measures for RBC transfusions in 34 PICUs

Physiologic measures

Not available

Some beds

All beds

Systemic physiologic measures

 Point of care PaO2

4 (12%)

2 (6%)

28 (82%)

 Pulse oximetry (SpO2)



34 (100%)

 Mean airway pressure (MAP) *



34 (100%)

 Swan-Ganz catheter

29 (85%)

3 (9%)

2 (6%)

 Systemic O2 consumption †

27 (79%)

5 (15%)

2 (6%)

 Point of care lactate measurement

4 (12%)

1 (3%)

29 (85%)

 Heart rate variability

14 (41%)


20 (59%)

 Plethysmographic variability

23 (68%)

1 (3%)

10 (29%)

 Near infrared spectroscopy (NIRS)

13 (38%)

15 (44%)

6 (17%)

 Microvascular blood flow

33 (98%)

1 (3%)


Organ specific or local physiologic measurements


2 (6%

2 (6%)

30 (88%)

 Pro-brain natriuretic protein (BNP)

17 (50%

2 (6%)

15 (44%)

 Brain PO2 measured in situ ‡

32 (94%)

2 (6%)


 Jugular bulb O2 saturation

27 (79%)

5 (15%)

2 (6%)

 Gastric tonometry (pHi)

31 (91%)

1 (3%)

2 (6%)

 Peripheral O2 extraction ¶

31 (91%)

1 (3%)

2 (6%)

 Microvascular blood flow

34 (100%)



* MAP can be used to calculate oxygenation index (OI) and oxygenation saturation index (OSI), two physiologic markers advocated by some experts, They can be calculated using the following equations given that FiO2 is always available in mechanically ventilated patients:

OI = (FiO2 x MAP×100)/PaO2); OSI = (FiO2 x MAP x 100)/SpO2).

 Measured by equipment like Deltatrac metabolic monitor.

 Brain PO2 measured by intracerebral device like Licox® Brain Tissue Monitoring System.

 Peripheral O2 extraction measured by pulse oximetry.

F-07 Short and long term outcome of patients with systemic rheumatic disease related interstitial lung disease admitted to intensive care unit- a multicentre retrospective study

Romaric Larcher (speaker), Lorrain Banuls, Fanny Garnier, Matthieu Amalric, Laura Platon, Jonathan Charbit, Kevin Chalard, Samir Jaber, Boris Jung, Kada Klouche

Montpellier University Hospital, Montpellier, FRANCE
Correspondence: Romaric Larcher -

Annals of Intensive Care 2019, 9(Suppl 1):F-07

Introduction: Owing their high mortality rate, admission to the intensive care unit (ICU) of patients with interstitial lung disease (ILD) is questioning. Systemic rheumatic diseases related ILD (SRD-ILD) seem to be associated with a better outcome but data are still lacking. We aimed therefore to evaluate short and long-term outcome and prognostic factors for SRD patients admitted to the ICU.

Patients and methods: This multicenter retrospective study was conducted, between 2006 and 2016, in five French ICUs and included patients with SRD-ILD admitted for acute respiratory failure. In-hospital and one-year crude mortalities were assessed and potential prognostic factors were identified using logistic regression.

Results: Seventy-one patients (female- 55%, mean age- 65 years [58–74], median SAPS2 43 [32–59], median SOFA Score 7 [4–9]) entered the study. ILD was related to connective tissue disease (43%), vasculitis (28%), myositis (13%), sarcoidosis (13%), and ankylosing spondylitis (3%). Causes of acute respiratory failure were sepsis (48%), pulmonary flare-up (28%) and miscellaneous (24%). Fifty-five patients (77%) required mechanical ventilation, 38 patients (53%) vasoactive drugs and 24 (34%) renal replacement therapy. One patient had an extracorporeal membrane oxygenation. Occurrence of severe, moderate and mild ARDS was 39%, 21%, 28% respectively. Thirty-seven survived at hospital discharge and 24 one year later. By univariate analysis, the following factors were significantly associated with mortality- ICU admission SAPS II (p = 0.009), PaO2/FiO2 ratio (p = 0.007), vasoactive drugs (p = 0.01) and mechanical ventilation requirement (p = 0.02). Among these factors, multivariate analysis showed that only low PaO2 FiO2 ratio (OR 3.63, 95% CI [3.59–298.74]) was associated with one year- mortality.

Conclusion: In our study, SRD-ILD patients admitted to the ICU have an in-hospital and one-year crude mortality rate at 48% and 66% respectively. Though this mortality rate remains high, it should not discourage intensivists to admit these patients. We found also that a low PaO2/FiO2 ratio during ICU stay was associated with one-year mortality. Further studies are however needed to best help physicians in the ICU management of ILD patients.

F-08 Impact of Advance Directives on the decision-making in Intensive Care Unit

Margot Smirdec (speaker), Bruno Pereira, Alexandre Lautrette

CHU Clermont-Ferrand, Clermont-Ferrand, FRANCE
Correspondence: Margot Smirdec -

Annals of Intensive Care 2019, 9(Suppl 1):F-08

Introduction: Physicians do not know how to follow the patients’ wishes when they are unable to express themselves. We assessed the impact of Advances Directives (AD) and how they have been written on the physician’s decisions.

Patients and methods: A multicentre, prospective, interventional, simulation study was carried out. Eight patients were recruited and wrote AD after receiving clear and complete information by video and interview with one ICU physician. Two simulation scenarios including ten questions about ICU admission and situations of withholding withdrawing therapies using the patients’ characteristics were submitted to ICU physicians from 28 French ICU, in three rounds (R)- simulation without knowledge of the patient’s AD (R1), with these AD (R2) and with these AD and the knowledge of how they were carried out (R3).

Results: The qualitative analysis of these 8 AD highlights a form of living will or end-of-life will and not formal guidelines on medical care. The results were performed on complete data of 102 physicians [Figure 1]. The variability between physicians themselves was high- among the 80 questions of R1, there were 37, 26 and 17 questions with an agreement > 80%, 80–60% and < 60% respectively. The AD significantly decreased the number of questions with an agreement > 80%, and increased the number of questions with an agreement < 60% (p = 0.02). There was no difference between the rounds 2 and 3 (p = 0.84). Few physicians’ characteristics were associated with the inter-individual variability. The intra-individual variability between R1 and R2 was very high (52 questions on 80 with kappa coefficient k < 0.4) and the AD were significantly associated with this variability after adjustment on characteristics of the physicians and the patients (p < 0.001). It is relevant to notice that the knowledge of how AD were carried out has a very low impact on the physicians’ decisions.

Discussion: We were expecting find some help in the AD of patients to diminish this variability among physicians and respect the patients’ autonomy. But the variability between the physicians themselves was increased in our study. We assume that this variability reflects a difference in the interpretation of these AD among physicians themselves.

Conclusion: The AD have a major impact on the physicians’ decisions for admission, withholding and withdrawing therapies decisions and increase the inter-individual variability but not the knowledge of how these AD have been written. It might be a limit of the autonomous model.

F-09 End of life decisions in acute respiratory failure immunocompromised patients

Gaston Burghi (speaker)1, Victoria Metaxa2, Peter Pickkers3, Marcio Soares4, Anders Perner5, Jordi Rello6, Philippe Bauer 7, Andry Van de Louw8, Pleun Hemelaar9, Virginie Lemiale10, Fabio Silvo Taccone11, Ignacio Martin Loeches12, Tine Sylvest Meyhoff5, Jorge Salluh4, Peter Schellongowski15, Katerina Rusinova16, Nicolas Terzi17, Geeta Mehta18, Massimo Antonelli19, Achille Kouatchet20, Andreas Barrat Due21, Miia Valkonen22, Pearl Landburg23

1Terapia Intensiva, Hospital Maciel, Montevideo, Montevideo, URUGUAY; 2Frank Stansil Critical Care Unit King’s College Hospital, Londres, UNITED-KINGDOM; 3The Department of Intensive Care Medicine (710), Radboud University Medical Center, Nijmegen, THE NETHERLANDS; 4Department of Critical Care and Graduate Program in Translational Medicine, D’Or Institute for Research and Education, Programa de Pós-Graduação em Cl, Rio de Janeiro, BRAZIL; 5Department of Intensive Care, Rigshospitalet, University of Copenhagen, DENMARK; 6Universitat Autonòma de Barcelona, European Study Group of Infections in Critically Ill Patients (ESGCIP), Barcelona, SPAIN; 7Pulmonary and Critical Care Medicine, Mayo Clinic, Rochester, UNITED STATES; 8Pulmonary and Critical Care Medicine, Mayo Clinic, Rochester Division of Pulmonary and Critical Care, Penn State University College of Medicine, Hershe, Hershey, UNITED STATES; 9Pulmonary and Critical Care Medicine, Mayo Clinic, Rochester Division of Pulmonary and Critical Care, Penn State University College of Medicine, Hershe, Nijmegen, THE NETHERLANDS; 11Department of Intensive Care, Hôpital Erasme, Université Libre de Bruxelles (ULB), Brussels, BELGIUM; 12Department of Intensive Care Medicine, Multidisciplinary Intensive Care Research Organization (MICRO), St. James’s Hospital, Dublin, IRELAND; 15Department of Medicine I, Medical University of Vienna, Vienna, Vienne, AUSTRIA; 16Department of Anesthesiology and Intensive Care Medicine and Institute for Medical Humanities, 1st Faculty of Medicine, Charles University in Prague a, Prague, THE CZECH REPUBLIC; 17CHU Grenoble Alpes, Service de réanimation médicale, Faculté de Médecine, INSERM, U1042, Université Grenoble-Alpes, Grenoble, FRANCE; 18Department of Medicine and Interdepartmental Division of Critical Care Medicine, Sinai Health System, University of Toronto, CANADA; 19Agostino Gemelli University Hospital, Università Cattolica del Sacro Cuore, Rome, ITALY; 20Department of Medical Intensive Care Medicine, University Hospital of Angers, FRANCE; 21Department of Emergencies and Critical Care, Oslo University Hospital, Oslo, NORWAY; 22Division of Intensive Care Medicine, Department of Anesthesiology, Intensive Care and Pain Medicine, University of Helsinki and Helsinki University Ho, FINLAND; 23Department of Critical Care, University Medical Center, Groningen, FINLAND
Correspondence: Gaston Burghi -

Annals of Intensive Care 2019, 9(Suppl 1):F-09

Introduction: In recent years, there has been an increase in immunocompromised patients admitted to the intensive care unit (ICU). End-of life decisions are challenging in these severe patients with often life-threatening underlying disease. The main aim of this study was to identify patient and organizational factors associated to decisions to forgo life-sustaining therapies (DFLSTs) among immunocompromised ICU patients.

Patients and methods: A secondary analysis of the EFRAIM study including 1611 immunocompromised patients with acute respiratory failure enrolled in 68 ICUs from 16 countries between October 2015 and June 2016 was performed. A multivariate logistic analysis was performed to identify independent predictors of DFLSTs.

Results: In the 1382 patients with data about DFLSTs available, 485 (35%) had a DFLST. Figure 1 presents the code status on ICU admission and the final decision about DFLSTs for each code status. The following variables were independently associated with increased incidence of DFLST: age (OR 1.024 per one year increase, IC 95% 1.012–1.035 + p < 0.0001), SOFA at ICU admission (OR 1.065, IC 95% 1.025–1.107, p = 0.0013), need for orotracheal intubation (OR 1.478, IC 95% 1.068–2.004, p = 0.0183), and performance status measured by the Eastern Cooperative Oncology Group scale (ECOG) (OR 1.7 per point, IC 95% 1.504–2.028, p < 0.0001), units without protocoled admission criteria (OR 1.701, IC 95% 1.176–2.462, p = 0.0022), palliative care involvement (OR 1.688, IC 95% 1.216–2.343, p = 0.0018) and units with frequent admission of transplant patients (OR 1.662, IC 95% 1.200–2.302, p = 0.0022). Two variables were independently associated with a decreased incidence of DFLST, namely acute respiratory failure with an easy aetiology identification (OR 0.913, IC 95% 0.860–0.968, p = 0.0025) and the use of checklists for global ICU care (OR 0.531, IC 95% 0.387–0.728, p < 0.0001). Centre effect was not significant after multivariate analysis.

Conclusion: The current study suggests that, in addition to patient-related factors, ICU characteristics and critical-care organization also influence decisions to forgo life sustaining therapies. Palliative care involvement, frequent admission of transplant patients, protocoled admission criteria at admission and the use of checklists are organizational variables associated with DFLSTs.

F-10 Practices assessment ten years after setting up a procedure of collegial decision for Withholding or Withdrawing Treatment (WWT) in Intensive Care

Marc Amouretti (speaker), Martha Gomis, Fabien Merrina, Sarah Benhamida, Nicolas Lau, Matthieu Le Meur, Remy Paulet, Nicolas Roucaud, Jordane Lebut, Martial Thyrault

Groupe Hospitalier Nord Essonne, Longjumeau, FRANCE
Correspondence: Marc Amouretti -

Annals of Intensive Care 2019, 9(Suppl 1):F-10

Introduction: Since the 2002 “Kouchner law” and the 2016 “Leonetti Claeys law”, WWT decisions have become essential in discussions within intensive care units. Our Intensive Care Unit (ICU) has conducted a work on this topic in 2008, leading to the implementation of WWT procedural forms and follow-ups. We assessed our practices 10 years after having implemented this traceability.

Patients and methods: Patient Limitation Records (PLR) are initiated and reassessed during weekly ICU meetings bringing together medical and paramedical staff. They can also be initiated at any other time during exceptional meetings. PLR are initiated at the request of any member of the paramedical team, medical team or patient’s entourage as soon as the meaningfulness of continuing care is contested. They include advanced medical instructions, an estimation of the patient’s quality of life before and after the ICU stay, patient’s wellbeing during the stay, interviews with the patient, his relatives and his practitioners. They also mention the conditions and time of application for the limitation. Analyzed records extend over a period of 18 months ending in June 2018. In 2008, the analysis period lasted 10 months from September to June.

Results: 106 PLR were established over the 2017–2018 period compared to 32 in 2008. 70% of patients are limited during the first ethical meeting during the 2017–2018 period. 18% have no limitation of life sustaining treatments. Patient’s will is unknown in 72% of cases. Among patients whose will is known, 21% express their opposition to therapeutic relentlessness. Patient’s general practitioner interview before the meeting is made for 42% of the patients. The most frequent therapeutic limitations are cardiac arrest resuscitation (70%) and renal replacement therapy (57%). Invasive mechanical ventilation is limited in 35% of cases. Nutrition and hydration are never limited.

Conclusion: Ten years after setting up a WWT procedure in our ICU, limitations are more frequent, appear earlier, concern younger patients, and are more often at the paramedic initiative, which reflects a stronger involvement of caregivers in end of life issues.

F-11 Assessment of the psychological consequences of hospitalization in intensive care unit- Patients and their relatives are both concerne

Ghada Sbouii (speaker)1, Nesrine Baili2, Rabia Atig2, Olfa Beji2, HoussemHmouda2

1Yasminet hospital, Kairouan, ABKHAZIA; 2Hopital sahloul sousse TUNISIA, Sousse, TUNISIA
Correspondence: Ghada Sbouii -

Annals of Intensive Care 2019, 9(Suppl 1):F-11

Introduction: The aim of our study was to evaluate the psychological impact of critical illness on patients and their relatives in a Tunisian medical ICU (MICU).

Patients and methods: In this study, included the patients who were admitted in our MICU from January 2015 and December 2016. We used the Impact of Events Scale (IES) 22-items for patients. Scores above 30 on the IES indicate severe psychological trauma symptoms, and individuals scoring in the range between 24 and 35 are likely to meet diagnostic criteria for PTSD. The Burden (Mini-Zarit) 7-items was used for relatives. All patients and relatives were interviewed by telephone.

Results: We contacted 46 families, 18 patients (39%) died, 28 patients (60.9%) participated with their relatives to this study, sex ratio was F M = 1.09, mean age was 75 years [65 to 90 years], average length of stay (LOS) was 11.89 days [1 to 60 days], 45.7% of patients had mechanical ventilation with a mean duration of 4.84 days, 8.7% of the patients had a tracheostomy, dementia was diagnosed in 6.5% of patients, 78.3% of the patients were autonomous, and the main reason of ICU admission was respiratory distress in 30.4%. The average of IES on patients was 22.7 [6 to 35], the IES was significantly correlated with male gender (p = 0.039), LOS in the ICU (p = 0.043), oro-tracheal intubation (OTI) (p = 0.03), tracheostomy (p = 0.013), and use of vasopressor agents (p = 0.045). The average of Mini-Zarit on relatives was 3.35 [1.5 to 5.5], the Mini-Zarit was significantly correlated with LOS (p‹0.001), OTI (p = 0.001), duration of OTI (p = 0.045), tracheostomy (p = 0.012), duration of tracheostomy (p = 0.024), use of vasopressor agents (p = 0.034) and the APACHE II score (p = 0.047).

Conclusion: Patients recovering from critical illness can be left with significant physical and cognitive problems that can deeply affect quality of life of both patients and relatives. Understanding the nature of the relationship between critical illness and PTSD is a challenge that demands attention, particularly in an era when mental health professionals are beginning to recognize the profound costs associated with this psychiatric syndrome.

F-12 Impact of frailty on elderly patients (≥ 80 years) admitted in French Intensive Care Units: a post hoc analysis from the international VIP study

Jérémy Rosman (speaker)1, Aurélien Cordonnier2, Xavier Forceville3, Guillaume Besch4, Hervé Mentec5, Philippe Michel6, Philippe Michel 7, Lucie Vettoretti8, Jérémy Bourenne9, Nathalie Marin10, Max Guillot11, Nadia Aissaoui12, Cyril Goulenok13, Nathalie Thieulot-Rolin14, Jonathan Messika15, Lionel Lamhaut16, Cyril Charron17, Bertrand Guidet18, Philippe Mateu1

1Service de Médecine Intensive Réanimation, CH, Charleville-Mézières, FRANCE; 2Département de l’Information Médicale, CH, Charleville-Mézières, FRANCE; 3Service de réanimation médico-chirurgicale, CH, Meaux, FRANCE; 4Département d’Anesthésie Réanimation Chirurgicale, CHR, Besançon, FRANCE; 5Service de Réanimation Polyvalente, CH Victor Dupouy, Argenteuil, FRANCE; 6Service de Réanimation médico-chirurgicale, CH de Carnelle - Portes de l’Oise, Saint-Martin-Du-Tertre, FRANCE; 7Service de Réanimation médico-chirurgicale, CH René Dubos, Pontoise, FRANCE; 8Service de réanimation médicale, CHU, Besançon, FRANCE; 9Service de Réanimation des Urgences et Médicale, CHU - Timone, Marseille, FRANCE; 10Service de réanimation médicale, Centre Hospitalier Cochin, Paris, FRANCE; 11Service de Réanimation médicale, Hôpital de Hautepierre, Strasbourg, FRANCE; 12Service de Réanimation médicale, Hôpital Européen Georges Pompidou, Paris, FRANCE; 13Service de Réanimation Médicale, Hôpital Privé Jacques Cartier, Massy, FRANCE; 14Service de réanimation, Centre Hospitalier de Melun, FRANCE; 15Service de Réanimation Médico-Chirurgicale, Centre Hospitalier Louis Mourier, Colombes, FRANCE; 16Service de réanimation polyvalente, Hôpital Necker, Paris, FRANCE; 17Service de réanimation médico-chirurgicale, Hôpital Ambroise Paré, Boulogne-Billancourt, FRANCE; 18Service de réanimation médicale, Hôpital Saint-Antoine, Paris, FRANCE
Correspondence: Jérémy Rosman -

Annals of Intensive Care 2019, 9(Suppl 1):F-12

Introduction: Very elderly Intensive care Patients (VIP1) is an international multicentric prospective study endorsed by the European Society of Intensive Care Medicine assessing prognosis of elderly patients (≥ 80 years) admitted in the intensive care unit (ICU). Our study is a subgroup analysis of French patients.

Patients and methods: All elderly patients (≥ 80 years) admitted in 18 French ICUs and included in VIP1 study from October 2016 to May 2017 were analyzed in a post hoc subgroup study. Frailty was defined as a Clinical Frailty Scale ≥ 5/9. Multivariable analysis [with adjusted odd ratio (aOR) and 95% confidence interval (IC 95%)] was performed to determine factors associated with in-ICU and 30-days mortality.

Results: Among 368 patients admitted in French ICUs, 38% were frail (44% in the entire cohort). Compared to non-frail patients, frailty was associated with the similar in-ICU (34% vs 35%, p = 0.93) and 30-days mortality (51% vs 44%, p = 0.22), despite a lower severity (SAPS2 46 vs 54, p = 0.04), lower use of mechanical ventilation (46% vs 62%, p = 0.003), and lower use of vasoactive drugs (39% vs 55%, p = 0.004). However, withholding was more frequent (57% vs 35%, p < 0.001), but not withdrawing (20% vs 21%, p = 0.76). In multivariable analysis, frailty was not an independent risk factor for mortality, contrary to withdrawing [aOR 18.8 (IC95% 8.1–43.8)], withholding [aOR 3.9 (IC95% 2.0–7.4)], use of mechanical ventilation [aOR 2.3 (IC95% 1.1–5.0)] and SOFA at admission (per one point increase) [aOR 1.2 (IC95% 1.1–1.3)].

Conclusion: In elderly patients admitted in French ICUs, frail patients have the same mortality than non-frail patients, despite a lower initial severity and higher frequency of withholding decisions.





(n = 241)


(n = 126)


Adjusted OR

(CI95 %)


Age, years

85 [81–89]

83 [81–86]



Age 80–89 years

212 (88%)

114 (91%)



Sex, male


118 (49%)

91 (38%)

66 (52%)

47 (37%)




Cause of admission

 Respiratory failure

108 (45%)

36 (29%)





27 (11%)

36 (29%)

< 0.001



 Elective surgery

9 (4%)

4 (3%)



 Urgent surgery

11 (5%)

3 (2%)



 Neuro (non trauma)

17 (7%)

11 (9%)



 Medical admission

218 (91%)

113 (90%)




3 (1%)

6 (5%)



Prior hospital stay > 7 days

33 (14%)

23 (18%)



Admission SOFA

5 [3–7]

12 [8–14]

< 0.001

1.2 (1.1–1.3)

< 0.001


43 [37–50]

65 [57–82]

< 0.001

not analysed

ICU interventions

 Non invasive ventilation

91 (38%)

36 (29%)



 Invasive ventilation

106 (44%)

99 (79%)

< 0.001

2.3 (1.1–5.0)


 Vasoactive drugs

89 (37%)

91 (72%)

< 0.001



 Renal replacement therapy

22 (9%)

22 (18%)





72 (30%)

87 (69%)

< 0.001

3.9 (2.0–7.4)



10 (4%)

65 (52%)

< 0.001

18.8 (8.1–43.8)


F-13 Evaluation of patients refused admission to intensive care unit

Alix Leurent (speaker)1, Matthieu Jamme1, Claire Pichereau1, Siu-Ming Au1, Christophe Barbier1, Yann Loubieres1, Jan Hayon1, Renaud Getti2, Hervé Outin1, Omar Ben Hadj Salem3

1Reanimation, CHI Poissy, Poissy, FRANCE; 2SAU, CHI Poissy, Poissy, FRANCE; 3 Cochin, Paris, FRANCE
Correspondence: Alix Leurent -

Annals of Intensive Care 2019, 9(Suppl 1):F-13

Introduction: Some patients referred to intensive care unit (ICU) are refused by intensivist because they are considered to be too well or too ill to benefit from intensive care treatment. This study aims to evaluate factors associated with decision to refuse ICU admission and to assess the outcome of excluded patients.

Patients and methods: All inpatients referred to our ICU between February and August 2018 were included in the analysis. All patients were evaluated in emergency unit or in hospitalization areas by a senior intensivist. Refused patients were considered to be too well (inappropriate referral (IR)) or too ill (futility) for ICU admission. The main outcome was survival 28 days after first referral to ICU. Patient characteristics, number of beds available at the time of referral, quality of life 28 days after ICU referral were also assessed.

Results: Out of 379 patients, 267 were admitted and 112 were refused (29.6%). Reasons for refusal were futility (“too ill”) (n = 68 (60.7%)) and IR (“too well”) (n = 44 (39.3%)). The median mortality probability model (MPMII) score for IR group and for futility group at day 0 was -1.65 and -0.4. Admission request came from emergency unit (65.2% (n = 73)), medical department (24% (n = 27)) or surgery department (10.8% (n = 12)). The main outcome for futility and IR groups were 52.9% and 4.5% respectively. In the futility group, modified Rankin score increased from 3.52 to 4.15 between inclusion and final evaluation. Lack of autonomy, elderly people and dementia were the main causes for refusal in the futility group (n = 57, 42 and 31 respectively).

Conclusion: Refusal of ICU admission occurs in 29.6% of cases. Refusal for futility is strongly but not invariably associated with deaths whereas patients considered to be too well for intensive care treatment rarely died. Strategy should be developed to create admission criteria for patients.

F-14 Carbapenem- sparing regimens for infections caused by ESBL-producing Enterobacteriaceae in ICU patients

Matthieu Bosset (speaker)1, Anne Gaelle Si Larbi1, Solen Kerneis2, Julien Charpentier2, Mathilde Phillips Houlbracq1, François Parquin1, Jean Paul Mira2, Charles Cerf1, Rémy Gauzit2, Philippe Lesprit2

1Hôpital Foch, Suresnes, FRANCE; 2Hôpital Cochin, Paris, FRANCE
Correspondence: Matthieu Bosset -

Annals of Intensive Care 2019, 9(Suppl 1):F-14

Introduction: The objectives of this study were, in ICU patients with an ESBL-producing Enterobacteriaceae infection- (i) to identify predictors of de-escalation of the empirical treatment with a carbapenem (CP) to a CP-sparing regimen (CPS) and (ii) to compare outcomes of patients de-escalated to those maintained on CP (CPM).

Patients and methods: Retrospective study performed in ICUs from 2 hospitals in which all CP prescriptions were reviewed within 72 h by the antimicrobial stewardship team and the ICU physicians. Demographics, clinical and therapeutic data were compared between CPS and CPM patients. Unfavorable clinical outcome (defined as mortality or relapse) was evaluated at day 60 of follow-up. Results are presented as median (IQR) or n (%).

Results: Ninety patients were included (54 CPM and 36 CPS), mainly treated for pneumonia (n = 35) or urinary tract infection (UTI, n = 30). Septic shock was found in 20 CPM (37%) and 19 CPS (53%), p = 0.14. The main CP-sparing antibiotics were piperacillin tazobactam (n = 11), cefoxitin (n = 11) and temocillin (n = 7). Median CP duration was 3 days (2–5]) in CPS patients. Cohort patients received a total of 1263 days of antibiotic (CP 827 days, CPS 436 days). Therefore, 35% of CP daily doses were saved. Factors associated with the use of a CPS were- older age (CPS 71 years [66–77], CPM 64 years [51–76] + p = 0.02), UTI (CPS- n = 24 (69%), including 19 male UTI + CPM- n = 6 (11%) + p < 0.001), absence of immunosuppressive therapy (CPS- n = 6 (21%), CPM- n = 17 (39%), p = 0.002) and total duration of antibiotic therapy (CPS 14 d [9–21], CPM- 9 d [7–15], p = 0.014). An unfavorable outcome was observed in 9 CPS (25%) and 22 CPM (41%), respectively (p = 0.5). Factors associated with unfavorable outcome were UTI (Hazard ratio [HR] 0.4, p = 0.045), respiratory tract infection (HR 2.76, p = 0.01) and immunosuppressive therapy (HR 3.5, p < 0.001). Use of a CPS had no effect (HR 0.6, p = 0.20).

Conclusion: CP de-escalation was performed in 40% of the patients, mainly those who were immunocompetent and treated for an UTI. This strategy seemed not to be associated with a worse outcome and reduced CP exposure. However, total duration of antibiotic therapy was longer with the CPS, because of the high rate of male IU treated.

F-15 Influence of the intervention of an infectiologist on the strategy of antibiotic therapy in a intensive care unit

Marie Bénistand (speaker)1, Arnaud Salmon Rousseau2, Stéphanie Honoré3, Thomas Rogier2, Mylène Herrera2, René-Gilles Patrigeon3

1CHU Dijon, Nîmes, FRANCE; 2CHU, Dijon, FRANCE; 3CH, Auxerre, FRANCE
Correspondence: Marie Bénistand -

Annals of Intensive Care 2019, 9(Suppl 1):F-15

Introduction: Impact of the discussion with bacteriologists and an infectiologist on antibiotic therapy in the 14-bed intensive care unit of the Auxerre hospital.

Patients and methods: Observational and retrospective study was conducted from 01/01/2014 to 31/12/2017. It was a before after study. First period - a weekly infectiology meeting of one hour composed of intensive care pratician, bacteriologist and pharmacist. Second period - intervention of an infectiologist. The primary outcome was the change in antibiotic consumption in daily doses per 1,000 days of hospitalization (DDD/1000 D). The secondary outcome was the change in the number of multi-resistant bacteria (BMR) isolated by weekly screenings after 72 h of hospitalization. The various proportions were compared using a Chi2 test.

Results: 1712 patients were hospitalized in intensive care unit (833 in 2014-15 vs 879 in 2016-17). Populations were comparable in age, sex, average length of stay, simplified seriousness index, and mortality. Antibiotic consumption differs between two periods.

Discussion: There is a statistically significant impact in antibiotic consumption. Fluoroquinolones and Amoxicillin-clanulanic acid which are known to induce resistance have decreased during the second period. Prescriptions of some antibiotics have improved such as Cefepime for Enterobacter, Ceftazidime and Ciprofloxacin for Pseudomonas thus saving Colimycin and Imipenem Cilastatin. There is a non-significant trend to lower multi-resistant broad-spectrum betalactamase-producing bacteria, cephalosorinases and carbapenemases in screening tests.

Conclusion: The intervention of an infectiologist in our service allowed an effective adaptation of the strategy of the antibiotic therapy in the hope that, in the coming years, the selection of BMR will decrease.






Total Beta lactam



Amoxicillin clavulanic acid









Total Quinolones















F-16 Appropriateness of empiric antimicrobial therapy with imipenem colistin in severe septic patients- An observational cohort study

Ahlem Trifi (speaker)1, Cyrine Abdennebi2, Sami Abdellatif2, Foued Daly2, Yosr Touil2, Rochdi Nasri2, Salah Ben Lakhal2

1Faculty of Medicine of Tunis, TUNISIA; 2Medical intensive care unit. La Rabta Hospital, Tunis, TUNISIA
Correspondence: Ahlem Trifi -

Annals of Intensive Care 2019, 9(Suppl 1):F-16

Introduction: empiric antimicrobial therapy (EAMT) using imipenem colistin is commonly prescribed as a first line therapy in critically ill patients with severe sepsis. We aimed to assess the appropriateness of prescribing imipenem colistin as EAMT in ICU patients.

Patients and methods: a 3-years observational prospective study including ICU patients that required imipenem colistin as EAMT. EAMT was assessed according to microbiological and clinical outcome. Outcomes were: delay to apyrexia, delay to decrease biological inflammatory parameters (BIP), requirement for vasoactive agents, bacteriological eradication, length of stay (LOS), ventilator days and 30-day mortality.

Results: 79 administrations of EAMT in 70 patients were studied. EAMT was appropriate in 52% of studied cases. An ICU stay > 6 days was related to inappropriateness and chronic respiratory failure was associated to appropriateness. In the appropriate EAMT group, we showed- earlier apyrexia, shorter delay to decrease biological inflammatory parameters (BIP) and less significant vasopressors requirement (attached fig). Furthermore, EAMT improved survival with a median gain of 4 days. Inappropriate EAMT increased the mortality risk by six. Acquisition of NI in ICU was also an independent factor of mortality.

Conclusion: EAMT using imipenem-colistin was appropriate in half of cases and inappropriateness was associated with increased ICU mortality risk.

F-17 Treatment of infections with extended spectrum beta-lactamase producing Enterobacteriaceae in ICU patients

Lev Volkov (speaker)1, Elisabeth Baux2, Sandrine Henard2, Bruno Levy2, Sebastien Gibot1, Pierre-Edouard Bollaert1

1Réanimation médicale de l’Hôpital Central, Nancy, FRANCE; 2Maladies infectieuses et tropicales Hôpital Brabois, Vandoeuvre-Les-Nancy, FRANCE
Correspondence: Lev Volkov -

Annals of Intensive Care 2019, 9(Suppl 1):F-17

Introduction: The incidence of extended-spectrum beta-lactamase producing-enterobacteriaceae (ESBL-PE) is growing, with an emerging worldwide distribution. The aim of this study was to describe the outcome of patients hospitalized in an ICU with an extended spectrum beta-lactamase producing Enterobacteria (ESBL-PE) infection treated with carbapenems and alternative carbapenem-sparing therapy.

Patients and methods: We performed a retrospective, descriptive, monocentric study in a Public University Hospital. Patients over 18 years, hospitalized in one of the ICUs of the hospital and presenting at the admission or during the stay an infection with an ESBL-PE between January 1st 2016 and March 31st 2018 were included. Patients colonized but not infected by ESBL-PE were not included. Primary study outcome was patient’s vital status at discharge of the ICU. Secondary outcomes were vital status at discharge of the ICU in patients treated with carbapenems and in patients treated with an alternative therapy, relapse and antibiotic change after reception of the antibiotic susceptibility test results.

Results: In our study, 109 patients were included. Among them, 93.6% had at least one risk factor of ESBL-PE acquisition, mostly a recent hospitalization within 3 months and a use of antibiotics within 3 months. Overall, 32 patients (29.4%) died during the ICU stay. The mean SAPS II score at admission was 49.8 ± 22.1. Invasive mechanical ventilation was required for 66% of the patients. Catecholamine use was required for 64% of patients. Escherichia coli was the most frequent bacteria identified (41%), followed by Enterobacter cloacae complex (33%) and Klebsiella pneumonia (19%). Distribution of the infection sites is summarized in Figure 1. Relapse occurred in 13 (12.1%) patients. In our study 84 of 109 patients (77.1%) received an empirical antimicrobial treatment, of which 22 (26.2%) received a carbapenem and 62 (73.8%) a non-carbapenem treatment. Among the 95 patients with documented treatment, carbapenems were used in 55 (57.9%) patients and non-carbapenems in 40 (42.1%) patients. Piperacillin tazobactam was the most used non-carbapenem treatment. There was no statistically significant difference in mortality between patients treated with a carbapenem and patients treated with a non-carbapenem therapy.

Conclusion: Infections with ESBL-PE in critically ill patients are associated with a great mortality. Because of the sample size of this retrospective study, no difference between the groups is observed. Therefore, a larger study should be drawn.

F-18 Risk factors in intravascular catheter-related infections

Hayfa Fazzeni (speaker), Sahar Habacha, Ameni Sghaier, Ines Fathallah, Eya Seghir, Asma Mehdi, Ghada Sboui, Khaoula Ben Ismail, Emna Ennouri, Nadia Kouraichi

Ben Arous Regional Hospital, Intensive care Unit, Ben Arous, TUNISIA
Correspondence: Hayfa Fazzeni -

Annals of Intensive Care 2019, 9(Suppl 1):F-18

Introduction: Catheter-related infection (CRI) is common in critically ill patients. Different factors are incriminated in developing CRI. Our study aimed to identify CRI risk factors.

Patients and methods: We conducted retrospective study including all patients admitted in our intensive care unit (ICU) during the period from October 2016 to August 2018. Catheter related infections were identified using the criteria of Infectious Diseases Society of America of 2009. Only patients with valid bacteriological data were included.

Results: During the study period, 164 patients were collected with a mean age of 55 ± 20 years and a sex ratio of 1.92. Median SOFA score was 4 [2-7] The most frequent causes of admission in ICU were acute respiratory failure (48.2%), coma (15.1%) and septic shock (7.9%). Diabetes mellitus was the main medical antecedent (34.1%). The median length of stay was 14 days [1-117]. The overall mortality rate was 29.9% (49 cases).

Among the 96 central venous catheters inserted, only twenty-two catheter related infections were diagnosed. SOFA score superior to 5 on admission was related to a higher risk of developing a CRI (OR, 1.75, 95%CI, [0.44–4.66], p = 0.033). In addition, time delay to develop a ventilator-associated pneumonia (VAP) was an independent risk factor for developing a CRI, for patients who developed a CRI median time to develop a VAP was 6 days versus 14 days for patients who did not develop a CRI (p = 0.04).

Conclusion: SOFA score at admission superior to five was a predictive factor of developing a catheter-related infection.

F-19 Cannula-related infection (Ca-RI) and insertion site colonization in patients supported by peripheral veno-arterial extracorporeal membrane oxygenation (VA-ECMO) managed with a standardized dressing procedure

Faiza Sayagh (speaker)

Réanimation médicale, hôpital bichat, Clichy, FRANCE
Correspondence: Faiza Sayagh -

Annals of Intensive Care 2019, 9(Suppl 1):F-19

Introduction: VA-ECMO is a life support technique used in patients with cardiac failure. High rates of Ca-RI have been reported in patients supported by ECMO. Based on evidences from Ca-RI prevention, we developed a standardized procedure- maximal sterile barrier precautions, antiseptic skin preparation using with 2% chlorhexidine-70% isopropyl alcohol solution, semi-permeable transparent and highly adhesive chlorhexidine dressing. Dressings were changed 24 h after catheter insertion and then every 7 days, but leaking or soiled dressings were changed immediately.

Patients and methods: Study design - We retrospectively review charts of all consecutive patients who underwent VA-ECMO support for > 48 h from January 2015 to December 2017. Peripheral blood samples were collected for culturing each VA-ECMO day and the insertion site was sample at each dressing by pressing a nutritive trypticase-soy agar plate (Count-tact + Biomerieux, France) on the skin for 5 s, centering the plate on the insertion site. Evaluation criteria were cannula-related bloodstream infection (Ca-RBSI) (defined as a combination of 1 or more positive peripheral blood cultures without any other infection explaining the positive blood culture result), Ca-RI (defined as local infections signs with or without Ca-RBSI), and skin colonization defined as neither Ca-RBSI nor Ca-RI but a positive insertion-site skin culture at catheter dressing.

Results: Over the study period, 131 patients were supported by 133 VA-ECMO (femoro-femoral VA-ECMO- n = 47, femoro-axillary- n = 86)- age- 56.4 ± 13.8; male- 97(73.6%); SAPS II- 56.7 ± 19.3; SOFA at cannulation- 9 ± 3.7; ECMO duration- 9 ± 8.8, corresponding to 1524 ECMO-days. Primary ECMO indications were post-cardiac surgery- 32 (24.1%), myocardial infarction- 30 (22.5%), primary graft dysfunction- 26 (19.5%), cardiac arrest- 20 (15%). Ca-RBSI and Ca-RI occurred in 13 (9.8%) and 31 (23.3%) patients respectively corresponding to an incidence rate of 8.5 1000 ECMO-days and 20.3 1000ECMO-days. Colonization of insertion site was frequent and mainly related to Coagulase-negative staphylococci, however, we found no Ca-RBSI and no Ca-RI related to this microorganism. Skin colonization was more frequent for cannulas in femoral position than for cannulas in axillary position.

Conclusion: Cannula-related infection and insertion site colonization were frequent despite a standardized procedure including semi-permeable transparent and highly adhesive Chlorhexidine dressing in patients supported by VA-ECMO.

F-20 Impact of the implementation of a multimodal and multidisciplinary care bundle for necrotizing skin and soft tissue infections

Tomas Urbina (speaker)1, Camille Hua2, Emilie Sbidian2, Romain Bosc3, Françoise Tomberli4, Raphael Lepeule5, Jean-Winoc Decousser6, Armand Mekontso Dessap7, Olivier Chosidow2, Nicolas De Prost7

1APHP, Paris, FRANCE; 2Service de Dermatologie, Hôpitaux Universitaires Henri Mondor – Albert Chenevier, AP-HP, Créteil, FRANCE; 3Service de Chirurgie Plastique et Reconstructrice, Hôpitaux Universitaires Henri Mondor – Albert Chenevier, AP-HP, Créteil, FRANCE; 4Service d’anesthésie et de réanimation chirurgicale, Hôpitaux Universitaires Henri Mondor – Albert Chenevier, AP-HP, Créteil, FRANCE; 5Service d’immunologie clinique et maladies infectieuses, Hôpitaux Universitaires Henri Mondor – Albert Chenevier, AP-HP, Créteil, FRANCE; 6Service de bactériologie-virologie, Hôpitaux Universitaires Henri Mondor – Albert Chenevier, Assistance Publique – AP-HP, Créteil, Créteil; 7Service de Réanimation Médicale, Hôpitaux Universitaires Henri Mondor – Albert Chenevier, AP-HP, Créteil, FRANCE
Correspondence: Tomas Urbina -

Annals of Intensive Care 2019, 9(Suppl 1):F-20

Introduction: Prompt antibiotic administration and surgical debridement of infected tissues are the main modifiable prognostic factors for necrotizing skin and soft tissue infections (NSTI). A multidisciplinary and multimodal care bundle was implemented in our center in order to standardize patient management. The aim of this work was to evaluate its impact on patient recruitment, management, and outcomes.

Patients and methods: This retrospective cohort study included all NSTI cases admitted in our center between 2006 to 2017. A multimodal care bundle was progressively implemented through 2012 to 2013. It consisted mainly in (1) the creation of an NSTI multidisciplinary team involving intensive care and infectious diseases physicians, dermatologists available 24/7 for patient referral, plastic surgeons and microbiologists + (2) the use of a triage algorithm including a multidisciplinary bedside assessment to accelerate access to the operating room + (3) the prospective identification of all NSTI cases + (4) and an active communication policy towards the medical community about the existing bundle. We compared patients from the pre-implementation period (2006–2011) to patients from the post-implementation period (2014–2017) regarding clinical features, management and outcomes. The primary endpoint was 60-day survival.

Results: Overall, 224 patients were admitted during the study period (pre-implementation, n = 60; per-implementation, n = 35, and post-implementation, n = 129). The number of yearly NSTI cases increased significantly between the pre- and the post-implementation periods (median [IQR] 9 [8–13] vs 30 [24–42]); p = 0.014). There was no significant difference between these periods regarding age, comorbidities, NSTI location or potentially modifiable prognostic factors, including time to surgery (0 [0–1] vs 0 [0–1] days; p = 0.281), the proportion of patients undergoing surgery within 24 h of admission (79% vs 78%; p > 0.99), antibiotic administration within 48 h (98% vs 100% + p = 0.683) and the adequation of antibiotherapy to guidelines (93% vs 94%; p = 0.960). Inclusion during the post-implementation period was associated with an increase in 60-day survival in univariable analysis (70% vs 85%; p = 0.020 by log-rank analysis) (Figure 1). Yet, this association was not statistically significant after adjustment for patient comorbidities and severity in a multivariable Cox model (adjusted hazard ratio = 0.88; 95% CI [0.41–1.87]; p = 0.741).

Conclusion: The implementation of a multidisciplinary and multimodal care bundle for NSTI was associated with higher patient recruitment and better 60-day censored survival in univariable but not multivariable analysis.

F-21 Does prior home ventilation positively impact the outcome of ICU-conducted Non Invasive ventilation?

Syrine Maatouk (speaker), Manel Lahmar, Zeineb Hamouda, Islem Ouanes, Fahmi Dachraoui, Wiem Nouira, Fekri Abroug, Lamia Besbes

CHU fattouma Bourguiba Monastir, Monastir, TUNISIA
Correspondence: Syrine Maatouk -

Annals of Intensive Care 2019, 9(Suppl 1):F-21

Introduction: NIV failure is associated with increased morbi-mortality in the ICU. NIV failure prediction could help improving overall outcome by prompting earlier intubation, and paying more attention to the management of patients at risk. In addition to the type of disease leading to acute respiratory failure, technical considerations might account for NIV failure. The aim of the current study is test the hypothesis that patients who are under home ventilation do experience a lower rate of failure when ventilated non-invasively in the ICU.

Patients and methods: This retrospective study with prospective data collection included all patients admitted between January 2011 and December 2017 to the ICU of CHU F.Bourguiba Monastir, for acute hypercapnic respiratory failure. In included patients we collected demographic and clinical data, details of diagnosis workup, ventilatory support, and its outcome (success vs failure). Variables usually considered with impact on ventilatory outcome were compared between patients with successful NIV course and those with NIV failure. Continuous variables are presented as mean ± SD, and p < 0.05 was considered statistically significant.

Results: During the study period, 355 patients were admitted in the ICU for hypercapnic respiratory failure. Of these 341 (96%) had NIV as the primary mode of ventilatory support. 94/341 had home ventilation, and the main cause of decompensation overall, was cardiac dysfunction (54%). ICU-conducted NIV failed in 50 patients (14.6%) with similar rates in patients with prior home ventilation (n = 16/94, 17%) compared to patients without prior home ventilation (n = 34/247, 14%). Table 1 depicts the risk factors of NIV failure disclosed by statistical analysis.

Conclusion: Prior home ventilation is not associated with a lower rate of NIV failure suggesting that failure from technical reasons was not highly prevalent in our series. NIV failure risk-factors disclosed pertain to clinical severity of the decompensation and its cause.

Table 1 Depicts the risk factors of NIV failure disclosed by statistical analysis


NIV Failure

NIV Success


7.26 ± 0.9

7.28 ± 0.4

PaCO2 (kPa)

9.9 ± 2.8

8.9 ± 2.2


6.6 ± 2

8 ± 2.6

ProBNP (pg/ml)

1680 ± 1462

2146 ± 1911

CRP (mg/l)

101 ± 92

83 ± 97

F-22 Physiological study of minimally invasive ECCO2R in exacerbation of COPD requiring invasive mechanical ventilation (EPHEBE study)

Jean-Luc Diehl (speaker)1, Lise Piquilloud2, Damien Vimpere3, Nadia Aissaoui3, Emmanuel Guérot3, Jean-Loup Augy3, Marc Pierrot4, Delphine Hourton3, Armelle Arnoux3, Christian Richard5, Jordi Mancebo6, Alain Mercat7

1Paris Descartes Faculty, Paris, FRANCE; 2CHU Vaudois, Lausanne, SWITZERLAND; 3Hopital Européen georges Pompidou, Paris, FRANCE; 4CH, Angers, FRANCE; 5Hopital de Bicètre, Le Kremlin-Bicètre, FRANCE; 6Hospital de la Santa Creu i Sant Pau, Barcelona, SPAIN; 7Angers University, Angers, FRANCE
Correspondence: Jean-Luc Diehl -

Annals of Intensive Care 2019, 9(Suppl 1):F-22

Introduction: Mechanical ventilation for exacerbations of COPD (AE-COPD) aims to provide adequate gas exchanges and to reduce the work of breathing (WOB). Extracorporeal CO2 removal (ECCO2R) can be a valuable additional modality for AE-COPD patients requiring invasive mechanical ventilation (IMV), by improving gas exchanges without deleterious consequences in terms of dynamic hyperinflation. However, little is known about the quantification of such ECCO2R-induced benefits.

Patients and methods: Open phase II-III prospective crossover (fixed order) study performed in 12 deeply sedated IMV AE-COPD patients. Dynamic hyperinflation and gas exchanges were compared without and with ECCO2R (Hemolung, Alung, Pittsburgh, USA) and adjustment of the respiratory rate (Carescape R860, GE Healthcare). The adjustment algorithm (either positive or negative) aimed to normalize arterial pH value and was based mainly on native lungs VCO2 values. When possible, WOB (Campbell’s method) with and without ECCO2R was measured at the end of the weaning process. Results are expressed in median [IQR]. Non-parametric tests were used.

Results: Patients (SAPS2- 33[28.5–38.5] were included in 2 centers during an 18-months period. Table indicates the main results obtained without and with ECCO2R (using the higher permitted value of sweep gas flow) and adjustments of respiratory rate. WOB measurements (Joules/min and Joules/L) were possible in 5 patients, indicating near-significant higher values when stopping the sweep gas flow for a 1 h. period- 11.7[7.5 ± 15.0] versus 22.6[13.9 ± 34.7] J/min, p = 0.0625 and 1.1[0.8 ± 1.4] versus 1.5[0.9 ± 2.8] J/L, p = 0.0625.

Three patients died in-ICU. The other patients were successfully hospital-discharged. The total duration (days) of tracheal intubation was 8 [6–18]. The duration of tracheal intubation after ECCO2R initiation was 6 [4; 16.5]. The duration of ICU stay was 14.5 [8; 22.5]. There were 4 VAP episodes, 3 hemorrhagic and 3 thrombotic complications.

Conclusion: Significant PaCO2, pH and SatHbO2 improvements were observed. The algorithm for pH normalization performed rather well, but without reduction in respiratory rate in the whole group and accordingly improvements in dynamic hyperinflation. This can be due to (i) mixed respiratory-metabolic acidosis in some patients and to (ii) intrinsic Hemolung properties (low to middle extracorporeal blood flow and membrane surface area). Results in terms of WOB are in line with previous publications.

F-23 Hyperoxia is toxic in patients with septic shock according to the Sepsis-3 criteria

Julien Demiselle (speaker)1, Martin Wepler2, Clair Hartmann3, Peter Radermacher2, Frédérique Schortgen4, Ferhat Meziani5, Mervyn Singer6, Valérie Seegers1, Pierre Asfar1

1CHU, Angers, FRANCE; 2Institut für Anästhesiologische Pathophysiologie und Verfahrensentwicklung, Universitätsklinikum, Ulm, GERMANY; 3Klinik für Anästhesiologie, Abteilung Klinische Anästhesiologie, Universitätsklinikum, Ulm, GERMANY; 4Service de Réanimation adulte, Centre Hospitalier Intercommunal de Créteil, FRANCE; 5Université de Strasbourg (UNISTRA), Faculté de Médecine, Hôpitaux universitaires de Strasbourg, Service de Réanimation, Strasbourg, FRANCE; 7Bloomsbury Institute of Intensive Care Medicine, London, UNITED-KINGDOM
Correspondence: Julien Demiselle -

Annals of Intensive Care 2019, 9(Suppl 1):F-23

Introduction: The Sepsis-3 definition of septic shock includes vasopressor treatment to maintain a mean arterial pressure over 65 mmHg and a lactate concentration over 2 mmol/L. The impact of hyperoxia in patients fulfilling these criteria is unknown.

Patients and methods: We conducted a post hoc analysis of the HYPER2S trial, including patients requiring vasopressor therapy with an available plasma lactate value at study inclusion. We compared the effect of hyperoxia and normoxia treatment on mortality of patients with hyperlactatemia (> 2 mmol/L) and of patients requiring vasopressor for hypotension without hyperlactatemia.

Results: 397 patients were enrolled in this analysis, in whom 230 had lactate over 2 mmol/L and 167 had lactate lower or equal to 2 mmol/L. Among patients with lactate > 2 mmol/L, 108 and 122 were “hyperoxia”- and “normoxia”-treated, respectively. Patients with lactate > 2 mmol/L had significantly less coronary artery disease, more cirrhosis and required surgery more frequently. They also had higher illness severity (SOFA 10.6 ± 2.8 vs 9.5 ± 2.5, p = 0.0001), required more renal replacement therapy (RRT), and received vasopressor and mechanical ventilation for longer time. Mortality rate at day 28 was higher in the “hyperoxia”-treated patients with lactate > 2 mmol/L as compared to “normoxia”-treated patients (57.4% vs 44.3%, p 0.054), despite similar RRT requirements as well as vasopressor and mechanical ventilation-free days. A multivariate analysis showed an independent association between hyperoxia and mortality at day 28 and 90. In patients with lactate ≤ 2 mmol/L, hyperoxia had no effect on mortality, nor on other outcomes.

Conclusion: This study suggests that hyperoxia may be associated with a higher mortality rate in patients with septic shock using the Sepsis-3 criteria, but not in patients with hypotension requiring vasopressor without hyperlactatemia.

F-24 Acute respiratory failure in obesity-hypoventilation syndrome managed in the intensive care unit

Nader Chebib (speaker)1, Pascale Nesme2, Nathalie Freymond3, Laurent Argaud4, Thomas Rimmele4, Julien Bohe4, Gilles Devouassoux2, Pierre Jean Souquet3, Claude Guerin1

1Réanimation Médicale Hôpital de la Croix Rousse, Hospices Civils, Lyon, FRANCE; 2Pneumologie Hôpital de la Croix Rousse, Hospices Civils de Lyon, FRANCE; 3Pneumologie Centre Hospitalier Lyon Sud, Hospices Civils de Lyon Pierre-Bénite, Lyon, FRANCE; 4Réanimation Médicale Hôpital Edouard Herriot, Hospices Civils de Lyon Pierre-Bénite, Lyon, FRANCE
Correspondence: Nader Chebib -

Annals of Intensive Care 2019, 9(Suppl 1):F-24

Introduction: Obesity-hypoventilation syndrome (OHS) is an increasing cause of acute hypercapnic respiratory failure (AHRF) in intensive care unit (ICU). Our objective was to describe the epidemiology, the ventilatory management and the outcome of patients with OHS admitted to the ICU for AHRF.

Patients and methods: We retrospectively built a cohort of OHS patients staying in 2 pneumology wards and admitted for AHRF in the 4 ICUs taking part of the tertiary University teaching hospital in Lyon, France, between January 1st 2013 and September 30th 2017. Clinical, functional and biological characteristics of patients at the time of OHS diagnosis were used as covariates. The main end-point was the rate of success of noninvasive ventilation (NIV) in the ICU. The secondary end-points were patient survival from OHS diagnosis to last follow-up, risk factors for ICU admission and for long-term survival including the role of ICU stay. Data were expressed as median (1st-3rd quartiles) and counts (percentage) and compared with nonparametric tests. Survival was measured by the Kaplan–Meier method and the difference in restricted mean survival time (RMST). Multivariate logistic regression analysis was used to assess risk factors of ICU admission. Cox proportional hazard model and multistate model were used to determine the risk factors of long-term survival.

Results: One-hundred and fifteen prevalent patients with OHS were included. Over a median follow-up period of 3.5 (1.4–6) years, 37 patients (32.1%) were admitted to the ICU for AHRF. The most frequent cause of AHRF was congestive heart failure (54%). Fourteen patients (37.8%) were treated for OHS with continuous or bilevel positive airway pressure prior to ICU admission. NIV was used as a first-line ventilatory support in 36 patients (97.2%) and was successful in 33 patients (89.2%). ICU mortality was low (2.7%). Patients admitted to ICU had significantly higher age, lower forced expiratory volume in 1 s and lower vital capacity (VC) at the time of OHS diagnosis. Difference in RMST was significant with a gain in survival of 663 days for patients not admitted to ICU. Multivariate analysis showed that lower VC at OHS diagnosis was significantly associated with higher risk of ICU admission. No factor was independently associated with long-term overall mortality in multivariate analysis.

Conclusion: AHRF is frequent in OHS diagnosis and is generally responsive to NIV. Lower VC is associated with a higher risk of ICU admission.

Table 1. Clinical, biological and treatment characteristics of patients with OHS admitted to the ICU



Age at ICU admission (years)

72 (66–79)

ABG at ICU admission



7.06 (7.22–7.31)

 PaCO2 (mmHg)

70 (61–76)

 PaO2 (mmHg)

79 (66–83)

 Bicarbonate (mmol/L)

30 (26–35)

 Lactate (mmol/L)

1.2 (0.8–1.7)

Vital signs


 MAP (mm Hg)

81 (66–95)

 HR (beats per minute)

91 (80–95)

 RR (cycles per minute)

25 (20–30)

 Glasgow coma score

13 (13–15)

 IGSII score

36 (23–49)

AHRF cause n (%)


 Congestive heart failure

20 (54)


11 (29.7)

 Pulmonary embolism

4 (10.8)


1 (2.7)

 Neoplastic pleural effusion

1 (2.7)

 Prior OHS diagnosis n (%)

14 (37.8)

 Delay OHS diagnosis-ICU admission (months)

11.5 (− 2.4 to 25)



 NIV first-line n (%)

36 (97.2)

 Invasive ventilation first-line n (%)

1 (2.7)

 Invasive ventilation after NIV failure n (%)

3 (8.1)

 Length invasive ventilation (days)

4.4 (1–7)

 Length NIV (days)

6 (3–8)

 Time to correct acidosis with NIV (days)

2.9 (1–3)

 NIV at ICU discharge n (%)

23 (62.1)

 Renal dialysis n (%)

3 (8.1)

 Vasopressor agents n (%)

8 (21.6)

Length of stay (days)



7.2 (4–8)


14.2 (10–16.5)

 Death in ICU n (%)

1 (2.7)

Quantitative values are expressed as median with first and third quartile; MAP: mean arterial pressure; HR: heart rate; RR: respiratory rate; IGSII: simplified severity index II; NIV: non invasive ventilation

F-25 Critical Illness-Related Corticosteroid Insufficiency during difficult weaning from mechanical ventilation

François Bagate (speaker), Alexandre Bedet, Françoise Tomberli, Keyvan Razazi, Nicolas De Prost, Guillaume Carteaux, Armand Mekontso Dessap

CHU Henri Mondor, Créteil, FRANCE
Correspondence: François Bagate -

Annals of Intensive Care 2019, 9(Suppl 1):F-25

Introduction: Critical illness-related corticosteroid insufficiency (CIRCI) is common during critical illness and is usually associated with poor outcomes, as prolonged duration of mechanical ventilation (MV) and higher mortality. CIRCI may alter cardiac and vascular functions. Weaning-induced pulmonary oedema (WiPO) is a major mechanism of weaning failure. The aim of this study was to evaluate the role of CIRCI in patients with difficult ventilator weaning and its possible relation with WiPO.

Patients and methods: Prospective study conducted in the intensive care of a university hospital in France. Patients under MV for more than 24 h, meeting weaning criteria and having failed the first spontaneous breathing trial (SBT) underwent a corticotropin stimulation test, with assessment of total blood cortisol levels immediately before (T0) 0.25 mg iv of tetracosactrin and 30 and 60 min afterward. Δmax was defined as the difference between the maximal value after the test and T0. CIRCI was defined as T0 < 10 μg/dL (276 nmol/L) and or Δmax < 9 μg/dL (248 nmol/L) and inadequate adrenal reserve as Δmax < 9 μg/dL. Biomarkers (natriuretic peptide and protidemia) sampling and echocardiograms were performed during the second SBT and were used to diagnose WiPO, which was defined according to two definitions (one liberal and one conservative) derived from recent publications on the topic. Successful extubation was defined as patient alive without reintubation 7 days after extubation.

Results: Seventy-six consecutive patients (63 ± 14 years; 49 men) with difficult weaning were enrolled. CIRCI and inadequate adrenal reserve occurred in 25 (33%) and 17 (22%) patients, respectively. The probability of successful extubation (alive was significantly decreased in patients with CIRCI or inadequate adrenal reserve (Figure 1A and 1B), as compared to their counterparts, and this association persisted after adjustment on severity (SOFA score at first SBT). WiPO occurred in 44 (58%) and 8 (11%) patients, according to the liberal and conservative definition, respectively. WiPO was not associated with CIRCI or with inadequate adrenal reserve, and did not influence weaning outcome, whatever the definition used.

Conclusion: CIRCI was common during difficult weaning and was associated with its prolongation. We did not find a significant association between CIRCI and WiPO.

F-26 Bedside predictors of successful weaning from high-flow nasal cannula in ICU

Rémi Coudroy (speaker), Maeva Rodriguez, René Robert, Jean-Pierre Frat, Arnaud W. Thille

Service de Médecine Intensive et Réanimation, CHU, Poitiers, FRANCE
Correspondence: Rémi Coudroy -

Annals of Intensive Care 2019, 9(Suppl 1):F-26

Introduction: High-flow nasal cannula oxygen therapy (HFNC) is a recent technique that can provide high FiO2, decrease the work of breathing, and recruit lungs. Despite promising clinical results, its weaning has never been investigated. Our objective is to describe factors associated with successful weaning of HFNC.

Patients and methods: In a retrospective monocenter study over a 2-year period, all patients admitted to ICU and treated with HFNC for acute respiratory failure were included. Patients who were never weaned from HFNC during ICU stay, those who were treated with HFNC associated to NIV, and those who received preventive post-extubation HFNC were excluded. Our primary outcome was to compare patients according to the outcome of the first HFNC weaning attempt. Our secondary outcome was to compare the evolution of patients who failed HFNC weaning at first attempt, and then who were successfully weaned from HFNC.

Results: From the 190 patients included, 168 (88%) were successfully weaned from HFNC at the first attempt. At time of the first attempt, those who were successfully weaned had lower FiO2 (39 vs. 48%, p = 0.02), were more likely to have a SpO2/FiO2 > 235 (corresponding to a PaO2/FiO2 > 200 mm Hg) (74% vs. 41%, p = 0.005), and had a higher ROx index (SpO2/FiO2/RR) than those who failed (p = 0.002). The 19 patients who failed the first attempt significantly increased the ROx index at time of successful weaning (p = 0.04).

Conclusion: If confirmed in further studies, SpO2/FiO2 and ROx index could be 2 useful noninvasive predictors of success of HFNC weaning available at bedside.

F-27 Is immunosuppression status a risk factor for noninvasive ventilation failure in acute hypoxemic respiratory failure?

Rémi Coudroy (speaker)1, Tài O. Pham2, René Robert1, Jean-Pierre Frat1, Arnaud W. Thille1

1Service de Médecine Intensive et Réanimation, CHU de Poitiers, Poitiers, FRANCE; 2Interdepartmental Division of Critical Care Medicine, Toronto, CANADA
Correspondence: Rémi Coudroy -

Annals of Intensive Care 2019, 9(Suppl 1):F-27

Introduction: Recent European American clinical practice guidelines recommend first-line noninvasive ventilation (NIV) to manage acute hypoxemic respiratory failure in immunocompromised patients. In contrast, in non-immunocompromised patients, the experts were unable to offer a recommendation given the uncertainty of evidence. Immunocompromised patients have particularly high mortality rates when they require invasive ventilation. However, it is not clear whether immunosuppression status is a risk factor for NIV failure. We aimed to assess the influence of immunosuppression status on outcomes of patients treated with NIV for acute respiratory failure.

Patients and methods: We performed a post hoc analysis pooling 2 prospective studies on acute hypoxemic respiratory failure. Patients treated by NIV were included. Those with cardiogenic pulmonary edema, acute-on-chronic respiratory failure or hypercapnia were excluded.

Results: Among the 208 patients analyzed, 71 (34%) were immunocompromised whose main reason was cancer (n = 37, 52%). Immunocompromised patients had higher severity scores upon ICU admission, were more likely to have bilateral lung infiltrates and received higher pressure-support levels under NIV than non-immunocompromised patients. Intubation and in-ICU mortality rates were higher in immunocompromised patients than in the others- 61% (43 out of 71 patients) vs. 43% (p = 0.02), and 38% (27 patients) vs. 15% (p < 0.001), respectively. Using multivariate analysis, immunosuppression was independently associated to intubation with an adjusted OR of 2.18 (95% CI 1.18–4.14, p = 0.01) and to ICU mortality (OR 3.44, 95%CI 1.73–7.03, p < 0.001).

Conclusion: Immunosuppression status influences outcomes in patients with acute hypoxemic respiratory failure treated with NIV. Studies in this specific population are mandatory.

F-28 Diagnostic tools used and pathogens involved in Ventilatory Acquired Pneumonia in PICU: a one-year prospective multicenter database (the INCIPAVE study)

Stephane Dauger (speaker)1, Yves Gallien2, Marcel Tinnevelt3, Maryline Chomton1, Alexandra Binoche5, Olivier Brissaud6, Capucine Didier 7, Pierre-Louis Leger8, Sonia Pelluau9, Laure De Saint-Blanquat10, Jérôme Naudin1, Matthieu Resche-Rigon12

1Réanimation pédiatrique, hôpital Robert Debré, Paris, FRANCE; 2Service de Biostatistique et Information Médicale, Paris, FRANCE; 3PICU, Utrecht, FRANCE; 5Réanimation pédiatrique, Hôpital Jeanne de Flandres, Lille, FRANCE; 6Réanimation pédiatrique, Hôpital Pellegrin, Bordeaux, FRANCE; 7Réanimation pédiatrique, Lyon, FRANCE; 8Réanimation pédiatrique, Hôpital Armand Trousseau, Paris, FRANCE; 9Réanimation pédiatrique, Hôpital des enfants, Toulouse, FRANCE; 10Réanimation pédiatrique, Hôpital Necker enfants malades, Paris, FRANCE; 12Service de Biostatistique et Information Médicale, Hôpital Saint-Louis, Paris, FRANCE
Correspondence: Stephane Dauger -

Annals of Intensive Care 2019, 9(Suppl 1):F-28

Introduction: Ventilatory Acquired Pneumonia (VAP) is one of the main nosocomial infection in adult ICU. To date, only one prospective multicenter study performed during six months in 16 PICUs of the US has prospectively described pediatric VAP. We design the INCIPAVE study to report the occurrence of VAP in european PICUs. One of the aim of the INCIPAVE study was to precisely describe the diagnostic tools used and the pathogens involved in pediatric VAP.

Patients and methods: Multicenter prospective cohort study from 03/04/2017 to 03/04/2018 including all patients mechanically ventilated (MV) at least once in eight PICUs, one in the Netherlands and seven in France. VAP was defined using the 2015 CDC criteria, applied during PICU stays, excluding the 48 h preceeding and following PICU. Patients were described on admission and main risk factors ever tested in the medical litterature were daily included by a pediatric intensivist of each PICU in an electronic database on a securized dedicated website. The Ethics Committee of the French Society of Intensive Care approved the study, which has been declared to the CNIL and recorded on Clinical-Trials. All parents or legal representatives were individually informed by a dedicated sheet. Descriptive data are reported as number (%) or medians [first-third quartiles].

Results: These results are based on declarative information reported in the INCIPAVE database during its first opening on September 2018, before cleaning. A total of 158 VAP (76.8% according to CDC criteria) was declared during 11685 days of MV. At least one pathogen was identified in 58.7% case of VAP, with more than one in 11.6%. Diagnosis was made on Tracheal Aspirates, with or without quantitative cultures (51% and 12.9% respectively), and on Blind Protecting Specimen Brush with or without quantitative cultures (7.7% and 2.6% respectively). Identified pathogens were Pseudomonas aeruginosas (PA, 12.1%), Haemophilus influenzae (HI, 12.1%), Staphylococcus aureus (SA, 7.4%), Streptococcus pneumoniae (SP, 5.4%), Stenotrophomonas maltophilia (SM, 3.4%). HI and SP were involved in “early VAPs” (< 6 days) and PA, SM and SA were reported in “late onset” VAPs.

Conclusion: Tracheal aspirate with quantitative culture is still the method of choice to diagnose VAP in PICU. Nosocomial pathogens are mainly involved after 6 days of MV. A more precise analysis is planned after cleaning of the database, including a special reading of each case of VAP.

F-29 Management of pleural infection in children at a University Hospital: A retrospective study over 6 years

Ombeline Roignot (speaker), Anne-Sophie Guilbert, Charlie De Melo, Amélie Stern

CHU de Strasbourg - Hôpital de Hautepierre, Strasbourg, FRANCE
Correspondence: Ombeline Roignot -

Annals of Intensive Care 2019, 9(Suppl 1):F-29

Introduction: Parapneumonic effusions and pleural empyema are common complications of community-acquired bacterial pneumonia in children. There is not yet a consensus concerning the management of this pathology, and expert’s recommendations are scarce. The objective of this study was to compare whether the evolution of our patients benefited or not from drainage procedures.

Patients and methods: This retrospective single-center study collected data from 79 children treated for pleural infection at the Strasbourg University Hospital from May 2010 to May 2016. 2 groups of children were compared concenring medical and or invasive treatment received (thoracentesis, chest drain insertion with or without instillation of fibrinolytic agents and surgical techniques).

Results: 43 children benefited from an invasive treatment strategy (IT), and 36 from antibiotics alone (AA). The epidemiological data of the 2 groups was comparable (comorbidities, vaccination status). Significant differences were the duration of oxygen dependence- 3.1 days in the AA group versus 7.6 days in the IT group (p < 0.001), the time elapsed to obtaining apyrexia- 4 days in the AA group versus 7.9 days in the IT group (p = 0.009), overall hospital stay and duration of intensive care. Initial C-reactive Protein levels were measured at 160.5 mg/L in the AA group versus 258 mg/L in the IT group (p < 0.001). We proposed a predictive score for an invasive procedure based on initial CRP and pleural ultrasound data.

Conclusion: Pneumonia in children with low to moderate volume effusion may show a favorable clinical course treated by antibiotics only. Prospective randomized controlled trials are needed to reassess the indication of invasive procedures in the management of children’s pleural infection.

F-30 Mechanical ventilation under pediatric V–V ECMO

Jerome Rambaud (speaker), Julien Jegard, Isabelle Guellec, Pierre Louis Leger, Sandrine Jean, Yohan Soreze, Jean Eude Piloquet, Julia Guilbert, Cecile Valentin, Alexandra Bower

Armand-Trousseau, pediatric intensive care, Paris, FRANCE
Correspondence: Jerome Rambaud -

Annals of Intensive Care 2019, 9(Suppl 1):F-30

Introduction: Protective mechanical ventilation and adjuvant therapies for severe ARDS are well-defined in adult and pediatric population. However, no clearly identified recommendations are available to perform a protective ventilation during pediatric V–V ECMO. The aim of this study was first to describe potential associations between ventilatory settings during ECMO and outcome in ARDS patients. The secondary goal was to compare three periods of interest to identified significant modification and their potential consequences on the survival rate.

Patients and methods: We performed an observational monocentric retrospective study, from January 2007 to December 2017. All patients treated by ECMO V–V for a refractory ARDS were included. We collected data’s at day 1, day 3 day 7 and day 14 of ECMO. Three periods of interest were defined (before 2010, from 2010 to 2014 and after 2014). We retrospectively collected the data’s from our local database, approved by the French Data Protection Authority.

Results: 83 patients treated by extracorporeal membrane oxygenation were included. We identified an increase of the number of ECMO V–V for pediatric refractory ARDS associated with a higher survival rate throughout the three periods. The OSI (oxygenation saturation index) was the only pre-ECMO parameter significantly associated with a higher mortality.

We identified a significant modification the adjuvant therapy illustrated by a sharpe increase for the use of neuromuscular blockers (from 14% to 52%) and the prone positioning before ECMO (from 5% to 85%). We also show evidence of a strong modification of the ventilatory parameters during ECMO. As example, the tidal volumes are significantly lower throughout the periods (5 cc/kg vs 3.5 cc/kg) such as the driving pressure (28 vs 14 cm of H2O). In contrary, the PEEP is higher in the most recent period. Finally, we identified an improvement of the survival rate all over the three period.

Conclusion: Recent modifications of ventilatory parameters during V–V ECMO for pediatric ARDS aimed at implementing a better lung protection. These modifications are associated with a better survival. However, the correlation between survival and ventilators settings remained unclear and a multicentric study should help physician to identify prognosis factors.

F-31 Septic shock and toxic shock syndrome- two infectious shocks with different immune response

Solenn Remy (speaker)1, Karine Kolev-Descamps2, Morgane Gossez3, Fleur Cour-Andlauer2, Fabienne Vene3, Tiphanie Ginhoux6, Guillaume Monneret3, Etienne Javouhey2

1Hospices Civils de Lyon, Lyon, FRANCE; 2Réanimation Pédiatrique - Hôpital Femme Mère Enfant, Lyon, FRANCE; 3Laboratoire d’Immunologie cellulaire - Hôpital Edouard Herriot, Lyon, FRANCE; 6Centre D’Epidémiologie Clinique - EPICIME, Lyon, FRANCE
Correspondence: Solenn Remy -

Annals of Intensive Care 2019, 9(Suppl 1):F-31

Introduction: Septic shock (SS) has recently been redefined in adults as life-threatening organ dysfunction caused by dysregulated host response to infection. Due to pediatric specificities, adult definition cannot be just transpose to children. Toxic shock syndrome (TSS) is a particular entity of infectious shock, with large pediatric prevalence. Some toxins specific to Streptococcus A and Staphylococcus aureus lead to superantigenic activation of T-lymphocytes, responsible for major cytokines storm with multi-organ failure. While immunosuppression induced by SS is now demonstrated in adults and children, we investigated whether similar immune disorders arise during TSS.

Patients and methods: Single-center prospective study included all children under 18 years-old, consecutively admitted into Pediatric Intensive Care Unit for SS (“Surviving Sepsis Campaign” Goldstein criteria), or TSS (Center for Disease Control), between September 2014 and July 2018. Controls were recruited from outpatients admitted for an elective benign surgery, without any criteria of infection. Immune monitoring realized by flow cytometry included HLA-DR expression on monocytes (mHLA-DR), total lymphocyte count, and lymphocyte sub-populations’ proportions (CD4* and CD8* T cells, regulatory T cells, NK cells, B cells). Samples were analyzed at Day 1, 3 and 7, after shock onset. Clinical data were collected prospectively, as well as severity scores and secondary nosocomial infection occurrence.

Results: Forty-six SS, 12 TSS and 30 controls were recruited. At each time points, mHLA-DR in SS and TSS groups were decreased, compared with controls (fig 1). Moreover, mHLA-DR was significantly higher in TSS at day 1 and 3, than SS. Lymphocytes’ time course also differs between SS and TSS- more profound lymphopenia occurred at day 1 in TSS than SS, but correction was faster in TSS between day 1 and 3, while between day 3 and 7 in SS. No difference was observed concerning regulatory T cells. Thirteen patients with SS presented secondary infections (28%), compared to only one in TSS group (8%).

Conclusion: Our study showed that despite similar initial shock, immune response is significantly different between SS and TSS. TSS didn’t induce persistent immune-suppression, as seen in SS- highlighted by different time course of mHLA-DR, lymphocytes and secondary infection occurrence. There is no only one type of septic shock but different infectious shocks with different immune responses and clinical outcomes. These results reinforce objective to better characterize immune state of patients in intensive care, in order to propose personalized medicine with adapted immune-modulatory therapies.

F-32 The Septic Shock Score in children in septic shock treated with extracorporeal assistance

Clémence Marais (speaker)

Correspondence: Clémence Marais -

Annals of Intensive Care 2019, 9(Suppl 1):F-32

Introduction: Septic shock is a common pathology in intensive care units responsible for a high mortality rate. Extracorporeal life support (ECLS) is used when patients no longer respond to standard treatments, including inotropes. Recently, in a multicenter study involving more than 500 children with septic shock, the Septic Shock Score (SSS) proved highly reliable in identifying patients at risk of death and was able to define refractory septic shock. The objective of our study is to evaluate two versions of the SSS, the bedside SSS (bSSS) and the computed (cSSS) in a group of patients hospitalized with septic shock who received ECLS support.

Patients and methods: This retrospective study includes patients aged 1 month to 18 years hospitalized in the intensive care units of the Necker Enfant Malade, Trousseau and Bicêtre hospitals for septic shock requiring ECLS assistance between January 2010 and March 2018. Five data collection times were chosen- sepsis time, ECLS decision time, ECLS starting time and end of hospitalization time. At the first 4 collections, clinical and biological criteria were collected to calculate the different predictive scores of septic shock. The group of deceased patients and the group of living patients were compared at these different times.

Results: 38 patients were included in our study, 24 of which died during the hospitalization. At all times studied, both the bSSS and the cSSS had poor reliability in identifying deceased children. The vasoactive-inotropic score is significantly higher at ECLS starting time in deceased patients and it decreases between ECLS decision time and ECLS starting time in living patients.

Conclusion: This is a pilot study that tested a database of children in septic shock who had received ECLS in APHP’s intensive care units. The continuation of the study will be carried out as part of a collaborative project of the European Society of Pediatric and Neonatal Intensive Care (ESPNIC), and will extend to 27 other pediatric intensive care units to clarify ECLS success criteria, characterize the evolution of organ failure and thus better define ECLS uses in children with septic shock.

F-33 Prediction of Complicated Outcomes in children with sickle cell anemia: a CARABDREPA Cohort

Amélie Rolle (speaker)1, Raphael Blanc2, Jérôme Pignol2, Frédéric Martino1, Pascale Piednoir1, Bertrand Pons1, Hossein Mehdaoui1, Michel Carles1

1CHUG, Pointe-à-Pitre, FRANCE; 2CHUM, Fort-De-France, FRANCE
Correspondence: Amélie Rolle -

Annals of Intensive Care 2019, 9(Suppl 1):F-33

Introduction: Sickle cell disease (SCD) is an increasing global health problem + approximately 300,000 infants born every year. SCD is associated with a decreased life expectancy, half of the deaths occurring in the ICU. Preexisting prediction model built by The Cooperative Study of SCD do not show relationship existed between early clinical predictors and complicated outcome (CO). The identification of risk factors could potentially have an immediate effect in preventing clinical complications and improving the quality of life for hundreds of thousands of children worldwide. We aimed to identify early predictors of a CO, defined as an ICU stay > 2 days, the need for vital support or death in children with SCD.

Patients and methods: Retrospective observational cohort study of SCD patients over a 5-year period were conduced in French territories in the Americas teaching hospital and SCD referral center.

Results: Of the 2559 infant’s admissions in the Carabdrepa cohort, 174 (6.8%) had a CO, of whom 6 (0.2%) death. Using multivariate analysis, we found significant predictors of CO- an episode of dactylitis (defined as pain and tenderness in the hands or feet) before the age of one year (OR 3, IC 95% 1.9–45.5), a hemoglobin level of less than 7 g per deciliter (OR 1.75, IC 95% IC 0.98–2.84), and leukocytosis (OR 1.21, IC 95% 0.8–1.83), a respiratory rate more than or equal to 32 cycles min (OR 1.01, IC 95% 0.84–1.18).), a Delay between first symptom and medical contact (OR 1.3, IC 95% 1.04–1.65), and an admission for sepsis (OR 1.32, IC 95% 0.44–3.85). Our model demonstrated good predictive performances in terms of discrimination (c-statistic- 0.813) and calibration.

Conclusion: Sickle-cell disease children are at high risk of life threatening complications. Episodes of dactylitis, with a sustained drop of hemoglobin + associated with a sepsis context and a delay in medical care are strong predictors of a complicated outcome.

F-34 Hemorrhagic shock in multiple trauma children- epidemiological aspects and application of the TRISS methodology in this population

Luis Ferreira (speaker)1, Gilles Orliaguet2, Caroline Duracher-Gout2, Stephane Blanot2, Estelle Vergnaud2, Thomas Baugnon2, Philippe Meyer2

1Necker enfant malade, Ivry Sur Seine, FRANCE; 2APHP- Hopital Necker, Paris, FRANCE
Correspondence: Luis Ferreira -

Annals of Intensive Care 2019, 9(Suppl 1):F-34

Introduction: Traumatology is the leading cause of death in young adults and children. Hemorrhagic shock is a major aggravating factor in trauma, however there are very few data available in the pediatric population. TRISS method [1] offers a standard approach for evaluating outcome of trauma care, enabling the determination of an individual probability of survival (Ps) for each patient according to Trauma and Injury Severity Score. The aims of our study were- 1) to identify multiple trauma children admitted with hemorrhagic shock in our pediatric Trauma Center, 2) to analyze factors that could influence the outcome using the TRISS method.

Patients and methods: We performed a monocentric, observational, descriptive, retrospective study on medical records. The included patients were children under 18 years of age admitted for multiple trauma and presenting with hemorrhagic shock upon arrival. The patients included were identified by means of a computerized database internal to the service. The primary outcome was death.

Results: From January 2014 to April 2018, 947 multiple trauma children were admitted in the service. Among them, 41 (4.3%) were in hemorrhagic shock upon admission and included in the study. The median interquartile age and weight were 3.0 [2.0–10.7] years and 16.0 [12.9–32.5] kg, respectively. Their trauma profile was very similar to other polytraumatized children, with 73% of head trauma. The global principles of Damage Control Resuscitation have been met, including the application of the massive transfusion protocol. However, only 10% of children actually correspond to the usual definition of “massive blood transfusion” (70 ml kg over 24 h) and only 37% required a surgical hemostasis procedure. The overall mortality was 41% with an average TRISS of 45%. We had 2 “unexpected” survivors and 1 “unexpected” death according to the TRISS method in our population.

Conclusion: In our study, 4.3% of the multiple trauma children were admitted in hemorrhagic shock in our center. These children were heavily traumatized with an overall mortality rate of 41%. However, the TRISS method revealed an observed mortality 3.9% lower than the predicted mortality, with +2.4% of “excess survivors”. When we checked the medical records of the 2 “unexpected” survivors, they seemed to have beneficiated from a more aggressive prehospital resuscitation care.

  1. 1.

    Boyd CR, Tolson MA, Copes WS (1987) Evaluating trauma care- the TRISS method. Trauma Score and the Injury Severity Score. (J Trauma 27- 370-378).


Entire population (n = 41)

«Unexpected» survivors (n = 2)

Predicted death probability by TRISS

37.4 [10.3-81.5]

73.2 ; 86.7

 Mean TRISS death probability (%)



Prehospital major events and treatment


 Cardiac arrest before emergency teams arrived

8 (19%)


 Cardiac arrest during transport

4 (10%)


 Received Intravenous fluids (ml kg−1)

35 (85%) / 20.0 [10.5-33.0]

2 (100%) / 33.0 ; 50.0

 Received Blood transfusion (ml kg−1)

3 (7%) / 13.0 [8.0-15.7]


 Blood transfusion and plasma expanders

1 (2%)


 Received Tranexamic Acid (ml kg−1)

4 (10%) / 12.0 [10.0-17.5]

1 (50%) / 10.0

 Patient intubated upon arrival

32 (78%)

2 (100%)

 Administration of catecholamine

21 (51%)

2 (100%)

Norepinephrine only (μg kg−1 min−1

9 (22%) / 0.4 [0.3-0.6]

2 (100%) / 0.3 ; 0.5

Title: Prehospital major events and treatment in the entire population and the « unexpected » survivors group found through the TRISS methodology. Results shown as number of patients (%) and mediane [interquartile].

F-35 Diagnosis performance of repeated ECG for prediction of a coronary cause after cardiac arrest

Pierre Dupland (speaker)1, Florence Dumas2, Wulfran Bougouin3, Julien Charpentier4, Olivier Varenne5, Lionel Lamhaut6, Marine Paul 4, Jean-Daniel Chiche4, Frédéric Pène4, Jean-Paul Mira4, Alain Cariou4

1APHP, Paris, FRANCE; 2Emergency Department, Cochin University Hospital (APHP) and Paris Descartes University, Paris, FRANCE; 3Paris Cardiovascular Research Center, INSERM U970 (team 4), Paris, FRANCE; 4Medical Intensive Care Unit, Cochin University Hospital (APHP) and Paris Descartes University, Paris, FRANCE; 5Cardiology, Cochin University Hospital (APHP) and Paris Descartes University, Paris, FRANCE; 6SAMU 75, Necker University Hospital (APHP) and Paris Descartes University, Paris, FRANCE
Correspondence: Pierre Dupland -

Annals of Intensive Care 2019, 9(Suppl 1):F-35

Introduction: Electrocardiogram (ECG) is an essential tool for the diagnosis of acute coronary syndromes (ACS). However, diagnostic performances of post-resuscitation ECG are usually poor in the setting of out-of-hospital cardiac arrest (OHCA). Our aim was to evaluate the performances of repeated ECG during the pre-hospital period in order to identify patients who require an emergency percutaneous coronary intervention (PCI) after OHCA.

Patients and methods: We included a consecutive series of OHCA patients with no obvious extra-cardiac cause, in whom an immediate coronary angiogram (CAG) was performed at admission. The first ECG (early ECG) after the return of spontaneous circulation (ROSC) and the last ECG performed prior to CAG (late ECG) were classified into 4 groups, blinded to the angiographic result- (1) ST segment elevation (ST +), (2) left bundle branch block, (3) anomaly other than ST + but suspected ischemia, (4) no sign of ischemia. Respective performances of early and late ECGs were assessed using the need for early PCI as the main endpoint.

Results: Between 2011 and 2016, 287 patients were included, of whom 34% had a PCI at admission. A change in classification between early and late ECGs was observed in 111 patients (39%). A pattern of ST-elevation (group 1) was present on 26% of late ECGs, which predicted the need for PCI with a good specificity (87.3%) but a poor sensitivity (53.1%) (Table). In multivariate analysis, ST-elevation pattern (group 1) on late ECG was a stronger predictor for the need of PCI (OR = 6.81 (3.58–12.93), p < 0.001) as compared with the same pattern observed on early ECG (OR = 4.5 (2.59- 7.82 p < 0.001). Presence of any pattern of ischemia on the late ECG was also an independent predictor for the need of PCI (OR = 3.63 (1.74–7.6) p < 0.001), with a good sensitivity (86.4%), but a low specificity (36.2%). (Table 1) Absence of any ischemic aspect on late ECG performed after 43 min from ROSC was an independent predictor of no PCI (OR = 15.94 (3.26–77.99) p < 0.001).

Conclusion: Although performing better than the immediate post-resuscitation ECG, we observed that the reliability of the late ECG is insufficient in order to distinguish patients who will require an early PCI and those in whom an early CAG can be avoided.

F-36 Major traumatic complications after out-of-hospital cardiac arrest- Insights from the Parisian registry

Pierre-Alexandre Haruel (speaker)

Ambroise Paré, Paris, FRANCE
Correspondence: Pierre-Alexandre Haruel -

Annals of Intensive Care 2019, 9(Suppl 1):F-36

Introduction: Due to collapse and cardiopulmonary resuscitation (CPR) maneuvers, major traumatic injuries may complicate the course of resuscitation for out-of-hospital cardiac arrest patients (OHCA). Our goals were to assess the prevalence of these injuries, to describe their characteristics and to identify predictive factors.

Patients and methods: We conducted an observational study over a 9-year period (2007–2015) in a French cardiac arrest (CA) center. All non-traumatic OHCA patients admitted alive in the ICU were studied. Major injuries identified were ranked using a functional two-level scale of severity (life-threatening or consequential) and were classified as CPR-related injuries or collapse-related injuries, depending of the predominant mechanism. Factors associated with occurrence of a CPR-related injury and ICU survival were identified using multivariable logistic regression.

Results: A major traumatic injury following OHCA was observed in 91 1310 patients (6.9%, 95%CI- 5.6, 8.3%), and was classified as a life-threatening injury in 36% of cases. The traumatic injury was considered as contributing to the death in 19 (21%) cases. Injuries were related to CPR maneuvers in 65 patients (5.0%, (95%CI- 3.8, 6.1%)). In multivariable analysis, age [OR 1.02 + 95%CI (1.00, 1.04); p = 0.01], male gender [OR 0.53 + 95%CI (0.31, 0.91); p = 0.02] and CA occurring at home [OR 0.54 + 95%CI (0.31, 0.92); p = 0.02] were significantly associated with the occurrence of a CPR-related injury. CPR-related injuries were not associated with the ICU survival [OR 0.69 + 95%CI (0.36, 1.33); p = 0.27].

Conclusion: Major traumatic injuries are common after cardiopulmonary resuscitation. Further studies are necessary to evaluate the interest of a systematic traumatic check-up in resuscitated OHCA patients in order to detect these injuries.

F-37 PROSEDA, effectiveness of Procedural Sedation and Analgesia in emergency department, a prospective multicentric observational study

Romain Bouygues (speaker), Pierre Deneau, Geoffroy Rousseau, Saïd Laribi

Emergency Department, CHRU Tours, Chambray Les Tours, FRANCE
Correspondence: Romain Bouygues -

Annals of Intensive Care 2019, 9(Suppl 1):F-37

Introduction: The objective of our study was to observe practices of procedural sedation and analgesia (PSA) in emergency departments (EDs) and to evaluate its effectiveness.

Patients and methods: From January to July 2018, we conducted a prospective and multicentric study, in both EDs and prehospital setting. We enrolled adult patients needing painful procedures that would require PSA. Procedures were divided into 3 groups (A- dislocations + B- displaced fractures + C- other procedures such as abscess drainage or foreign body removal…). PSA drugs were divided into 3 groups (1- sedative drugs such as Propofol, Ketamine, or Midazolam + 2- Morphine + 3- absence of drugs or 50% Oxygen Nitrous-Oxyd Premix). We used a composite primary outcome to define the success of the PSA- successful procedure, EP feeling of sufficient sedation, and the patient’s absence of painful memories from the procedure.

Results: 108 patients were enrolled. 61 PSA (56.5%) were successful according to our primary outcome criteria (3 points out of 3) and 35 (32.4%) reached only 2 points. Pain decrease was measured by numeric rating scale (NRS) and its median was 4.5 (IQR 3–7). Number of patient in each procedure groups were as follow- Group A- n = 48 (44.44%), Group B n = 32 (29.63%), group C n = 28 (25.93%). Drug’s group 1 was predominant (n = 59, 54.6%) and its success rate was higher (76.3%) compared to group 2 and group 3 (22.7% and 40.7% respectively, p < 0.0001). In multivariable analysis, PSA success was independently associated with only 2 factors: Ramsay sedation scale > 2 (OR = 4.584 {1.927–11.501}, p = 0.0008), as a protective factor, while morphine group was against PSA success (OR = 0.181 {0.051–0.555} p = 0.0044). Adverse event rate in our study (15.7%) was comparable with the rate in other international studies. All adverse events were easily treated and had no serious consequences.

Conclusion: This prospective and multicentric study of 108 patients showed that efficiency was perfect in only 56.5% of the cases, and had a satisfactory result in 88.9%. Global efficiency was positively linked to the use of sedative drugs and negatively to cutaneous procedures such as abscess drainage.

F-38 CT scan quantification of pelvic and retroperitoneal hematoma predicts transfusion requirements, pelvic hemostatic management and outcome of severe trauma patients with pelvic fracture

Séverin Ramin (speaker), Pierre Cavaille, Margaux Hermida, Ingrid Milliet, Pauline Deras, Xavier Capdevila, Jonathan Charbit

Hôpital Lapeyronie, CHRU Montpellier, Montpellier, FRANCE
Correspondence: Séverin Ramin -

Annals of Intensive Care 2019, 9(Suppl 1):F-38

Introduction: Retroperitoneal hematoma (RPH) is frequently observed in case of pelvic fracture. RPH is the sign of traumatic pelvic hemorrhage. No study has validated a quantification score of RPH in a traumatic context. The aim of this study was to test the performance of a HRP quantification score to predict the outcome of polytraumatized patients with pelvic fracture.

Patients and methods: A retrospective study was performed in our trauma center between 2010 and 2015. All severe trauma patients with pelvic fracture who underwent a CT scan on admission were included. The amount of retroperitoneal effusion was quantified using a semi-quantitative method by counting the number of compartments affected by the spread of blood. Ten compartments in the retroperitoneal cavity were considered- prevesical space, laterovesical space, laterorectal space, presacral space, right and left iliopsoas space, periaortic and psoas space, right and left parietocolic space, and perirenal fascia extension. RPH was categorized in each compartment as absent (0), moderate (1), and large or bilateral (2) for a total score on 20. Patients were classified into 3 groups according to their RPH score (mild or none [0–5], moderate [6–9] and abundant [10–20]) and compared in terms of transfusion requirements, pelvic hemostatic management and prognostic variables.

Results: A total of 311 severe trauma patients were included (mean age- 42 ± 20 years, mean ISS- 27 ± 19, average RPH score- 6.5 ± 4.6). Among these patients, 68 (22%) had abundant RPH, 115 (37%) had RPH moderate and 128 (41%) had mild or no RPH. Massive transfusions requirement was more important in abundant RPH group (53% vs. 9% vs. 10%) as well as pelvic embolization requirement (10% vs. 3% vs 0%), pelvic fixator (15% vs. 4% vs. 1%), and mortality (26% vs. 10% vs. 8%) + P < 0.001. Similarly, days of mechanical ventilation and length of stay were significantly associated with RPH abundance (P < 0.001). The predictive value of moderate and abundant HRP for determining transfusion and interventional needs, as well as the risk of death were presented in figure 1. Using ROC curves analysis, the global ability of RPH score was robust to predict transfusion > 5 RBCs [0.75 (0.69–0.82)], massive transfusion [0.74 (0.65–0.83)] and death [0.70 (0.62–0.79)].

Conclusion: CT scan semi-quantitative analysis of HRP at admission allows for reliable prediction of transfusion requirements and outcome for Trauma patients with pelvic fracture.

F-39 Early post traumatic pulmonary embolism in intensive care unit

Mariem Dlela (speaker), Abir Bouattour, Olfa Turki, Hela Kallel, Mabrouk Bahloul, Hédi Chelly, Mounir Bouaziz

Hbib bourguiba university hospital, Sfax, TUNISIA
Correspondence: Mariem Dlela -

Annals of Intensive Care 2019, 9(Suppl 1):F-39

Introduction: Venous thromboembolism (VTE) is a well-established complication of trauma. Recent studies suggest that pulmonary embolism (PE) may occur very early and even immediately, after injury. The aim of this study was to analyze the incidence, risk factors and prognosis of early PE among intensive care unit (ICU) trauma patients.

Patients and methods: We conducted a twenty month long prospective cohort, including all trauma patients with a confirmed PE diagnosis, who were admitted to our ICU between January 1st, 2017 and August 31st, 2018. Early post traumatic PE was defined as pulmonary embolism diagnosed within the first 72 h of injury. All patients, included, were screened for early PE at day 3. Factors associated with early PE were identified using both univariate and multivariate analysis.

Results: During the study period’s, 66 patients with positive diagnosis of PE, were included. According to our analysis, 45% (30 cases) of the patients presented with PE within 72 h of trauma events.The patients in early PE group were older than those in the late PE group (p = 0.038), had a body-mass-index (BMI) above thirty (p = 0.021) and high sequantial organ failure score (SOFA) on admission (p = .001). On the day of PE diagnosis, early group also presented with higher SOFA scoring (p < 0.001) and higher infection rate (p = 0.005). Biological assessment revealed lower platelet levels (p = 0.001) and lower P F ratio in the early group (p = 0.008). Our study showed that early PE was associated with more transfusions (p = 0.002) and surgical treatment measures (p = 0.023). The incidence of long bone fractures in lower extremities was higher in those with early PE compared with the other patients (p = 0.039). Using the multivariable logistic regression model, higher age (p = 0.028), SOFA score (p = 0.013), BMI over thirty (p = 0.002), and the use of surgical treatment measures (p = 0.046) were predictive of early timing of PE in trauma patients. Whereas, pulmonary infection was independently associated with late PE. Long bone fracture was not independently related to early occurrence of PE.

Conclusion: Our cohort demonstrated that many of the post-traumatic PEs occur early in the post-traumatic period. To the best of our knowledge, this is the first prospective study conducted in ICU. Further studies with larger patient populations are required to create more accurate predictive models.

F-40 Gastric inflation induced by bag mask ventilation during different strategies of chest compressions in a cadaver model for cardiac arrest

Dominique Savary (speaker)1, Emmanuel Charbonney2, Ian Drennan3, Bilal Badat4, Paul Ouellet5, Stephane Delisle6, Caroline Fritz 7, Alain Mercat8, Laurent Brochard9, Jean Christophe Richard1

1CH Annecy Genevois, Annecy, FRANCE; 2Centre de Recherche de l ‘Hôpital du Sacré-cœur, Montreal, CANADA; 3St Michael’s Hospital, Toronto, CANADA; 4Air Liquide Medical Systems, Antony, FRANCE; 5Vitalité Health Network, Edmunston, CANADA; 6Collège Ellis, Drummondville, CANADA; 7Inserm U1116, Nancy, Nancy; 8CHU, Angers, FRANCE; 9ID Crit Care Med University, Toronto, CANADA
Correspondence: Dominique Savary -

Annals of Intensive Care 2019, 9(Suppl 1):F-40

Introduction: Bag mask ventilation is the most spread technic for ventilation for cardio pulmonary resuscitation (CPR) despite several adverse effects. Gastric gas insufflation may favor lung regurgitation and as result occurrence of aspiration pneumonia (1). We hypothesized that continuous chest compression (CC) may limit the risk of high tidal volume and gastric inflation compared to a 30–2 interrupted CC strategy. The aim of this experimental study was assess the impact of different CC ventilation strategies on gastric inflation and ventilation during a 6 min prolonged simulated CPR.

Patients and methods: 5 Thiel Embalmed Cadavers (TEC) from a donation program of the Université du Quebec Trois-Rivieres (CER-14-201-08-03.17) were ventilated 30 min to recruit lungs. (2). Flow and Airway Pressure were measured at the airway opening (AcqKnowledge software Biopac©). A surgical gastrostomy was performed through a 5 cm midline laparotomy to introduce a cuffed tracheal tube (size 6) into the stomach cavity + this tube was connected to a Wright spirometer to measure cumulated gastric inflated volume. Experimental protocol- 3 strategies were randomly applied during 6 min on each cadaver. 1. 30-2 with Interrupted Chest Compressions (ICC 30-2). Two successive bag insufflations after interrupting CC every 30 CC + . 2. 30-2 with Continuous Chest Compressions (CCC 30-2). Same CC ventilation ratio than in the first strategy but without interrupting CC. 3. Continuous CC with 1 bag insufflation every 6 s (CCC 10 min). Before each strategy the stomach was completely emptied through the gastrostomy tube, and the order of each strategy experimented were randomized for each cadaver.

Results: 5 cadavers were analyzed (mean age 75 ± 8 years, 60% female, PBW 56 ± 10 kg). Expired bag tidal volume averaged during the 6 min long period was 319 ± 165 ml during (ICC 30-2), 341 ± 142 ml during (CCC 30-2) and 277 ± 103 ml (CCC 10 min). Cumulated gastric inflated volume was significantly higher during (ICC 30-2) compared to CCC strategies (fig)- 5.9L 6 min (ICC 30-2) + 2.1L 6 min (CCC 30-2) and 2.24L 6 min (CCC 10 min) (p < 0.005).

Conclusion: Compared to the recommended (ICC 30-2) strategy, continuous chest compression significantly reduced cumulated gastric inflation. Interestingly, CCC did not affect ventilation actually delivered during CPR. The optimal ratio between chest compression en ventilation during continuous chest compression remains open to discussion.

F-41 Heart-brain interaction in acute cerebral injury

Mohamed Anass Fehdi (speaker), Amine Raja, Mohammed Mouhaoui

CHU Ibn Rochd, Casablanca, MOROCCO
Correspondence: Mohamed Anass Fehdi -

Annals of Intensive Care 2019, 9(Suppl 1):F-41

Introduction: Cardiac events are often seen in acute cerebral palsy, and they would be an indicator of poor prognosis. The purpose of this study was to clarify the heart-brain interaction in terms of incidents and repercussions on the morbidity and mortality of cerebral palsy patients.

Patients and methods: It was a prospective study, over 6 months, including consecutively all the acute cerebral palsy patients, initially admitted to the vital emergency room, and having benefited from a systematic cardiac assessment, namely an ECG, an echocardiogram and a troponin assay during the 24 h. Patients transferred from another hospital and or having a cardiac check-up after 24 h were excluded. The epidemiological, clinical, paraclinical, therapeutic and evolutionary parameters were studied. A univariate statistical study was carried out to deduce the prognostic factors of early mortality (< 48 h) among cardiac assessment components (p < 0.05).

Results: 76 patients were included, with an average age of 56.16 ± 13.56 years, and sex ratio 1.23 (42H 34F). The diagnoses related to the cerebro-lesion were- ischemic stroke (28 cases 36%), severe head trauma (24 cases 32%), hemorrhagic stroke (20 cases 26%), severe meningitis (4 cases 6%). The cardiac events observed were- an electrical anomaly in 19 patients (25%), an increase in troponin in 13 patients (17%) and an echocardiographic anomaly in 7 patients (9%). The distribution of cardiac events by type of brain injury is shown in Table 1. The early mortality rate (< 48 h) was 16% (12 deaths)- 6 hemorrhagic stroke, 3 ischemic stroke, 3 severe head trauma, and no severe meningitis. The prognostic value of various cardiac events in cerebral palsy patients by type of aggression is shown in Table 1.

Discussion: Our study did not show a statistically significant difference in terms of early mortality according to the presence or absence of one or more cardiac incidents. On the other hand, very important NPVs have been noted, all brain lesions combined.

Conclusion: The absence of cardiac events in cerebral palsy patients would favor a better early evolution. This finding should be confirmed by a broader study. We recommend performing a cardiac checkup in any cerebral palsy patient during the first 24 h.

F-42 Renal Doppler in predicting acute kidney injury (AKI) within 3 days in critically ill patients without AKI- Results of a multicenter cohort study

David Schnell (speaker)1, Aurélie Bourmaud2, Marie Reynaud3, Stéphane Rouleau1, Ferhat Meziani4, Alexandra Boivin4, Mourad Benyamina5, François Vincent6, Alexandre Lautrette7, Christophe Leroy7, Yves Cohen8, Matthieu Legrand5, Jérôme Morel3, Jeremy Terreaux3, Michael Darmon9

1Réanimation, CH, Angoulême, FRANCE; 2Lucien Neuwirth Cancerology Institute, Saint-Priest-En-Jarez, FRANCE; 3Réanimation chirurgicale, Hôpital nord, Saint-Priest-En-Jarez, FRANCE; 4Médicale, Nouvel Hôpital Civil, Strasbourg, FRANCE; 5Réanimation chirurgicale et unité des brûlés, Hôpital Saint Louis, Paris, FRANCE; 6Réanimation polyvalente, CH, Montfermeil, FRANCE; 7Réanimation médicale, hôpital Gabriel Montpied, Clermont-Ferrand, FRANCE; 8Réanimation médico-chirugicale, CHU Avicenne, Bobigny, FRANCE; 9Réanimation médicale, Hôpital Saint Louis, Paris, FRANCE
Correspondence: David Schnell -

Annals of Intensive Care 2019, 9(Suppl 1):F-42

Introduction: Doppler-based resistive index (RI) and semi-quantitative evaluation of renal perfusion using colour-Doppler (SQP) have been suggested as potential predictors of AKI occurrence in ICU patients without renal dysfunction (1). These results are however limited to preliminary data (1). This study aimed at evaluating the performance of RI and SQP to predict AKI within 3 days in critically-ill patients without overt AKI.

Patients and methods: Post-hoc analysis of a multicentre prospectively collected dataset. Adult patients without cardiac arrhythmia and requiring mechanical ventilation were included. Patients with severe chronic renal dysfunction or known renal artery stenosis were excluded. AKI was defined according both urinary output and serum creatinine criteria of the KDIGO definition. Renal Doppler was performed at study inclusion. Results are reported in n (%) or median (IQR). Adjusted factors associated with AKI development were assessed using mixed logistic regression model taking centre as random effect.

Results: Overall, of the 351 patients included in this study, 118 had no AKI at study inclusion and were ultimately included in the post hoc analysis. Half of the patients were of male gender (55.6%; n = 66) and median age was 58 [IQR 44–67]. In addition to mechanical ventilation, 47 patients (39.8%) required vasopressors, and 46 (39.0%) had a sepsis at inclusion. Median LOD score was 7 [IQR 5–9]. Overall, 34 patients developed AKI during the first 3 days of ICU stay (28.8%). Semi-quantitative perfusion score (ranging from 3—full perfusion to 0 no perfusion) was 2 [2–3] and 2 [1–3] respectively in patients with and without AKI at day 3 (P = 0.06). Doppler-based resistive index was 0.64 [0.57–0.70] and 0.67 [0.62–0.70] respectively in patients with and without AKI at day 3 (P = 0.18; figure). Area under ROC curve in predicting AKI within 3 days was 0.60 (0.49–0.71) for SQP and 0.58 (0.47–0.60) for RI. After adjustment for confounders, neither SQP (OR 0.58; 95%CI 0.31–1.10) nor Doppler-based RI (OR 29.15; 95%CI 0.12–703) were associated with AKI at day 3.

Conclusion: Our results suggest that neither Doppler-based resistive index nor semi-quantitative renal perfusion is accurate in predicting occurrence of AKI in ICU patients requiring mechanical ventilation.

F-43 Performance of urinary TIMP-2 and IGFBP7 and Doppler-based Resistive Index to predict reversibility of acute kidney injury in critically ill patients

Fanny Garnier (speaker)1, Delphine Daubin1, Romaric Larcher1, Anne-Sophie Bargnoux2, Laura Platon1, Vincent Brunot1, Yassir Aarab1, Noémie Besnard1, Anne Marie Dupuy2, Boris Jung1, Jean Paul Cristol2, Kada Klouche1

1Department of Intensive Care Medicine, Lapeyronie University Hospital, Montpellier, FRANCE; 2Department of Biochemistry, Lapeyronie University Hospital, Montpellier, FRANCE
Correspondence: Fanny Garnier -

Annals of Intensive Care 2019, 9(Suppl 1):F-43

Introduction: The performance of urinary tissue inhibitor of metalloproteinase-2 and insulin-like growth factor binding protein7 (TIMP-2*IGFBP7) to predict renal recovery has been poorly studied. In preliminary studies, Doppler-based renal resistive index (RI) might help in differentiating transient from persistent acute kidney injury (AKI). The aim of this study was to compare the performance of TIMP-2*IGFBP7 and RI in predicting short-term reversibility of AKI in critically ill patients.

Patients and methods: This prospective and monocentric study included consecutive critically ill patients with AKI. RI was measured within 12 h after admission and urinary TIMP-2*IGFBP7 was measured at H0, H6, H12 and H24. Renal recovery was evaluated at day 3. Receiver-operating characteristic curves (ROCs) were plotted to evaluate diagnostic performance of RI and TIMP-2*IGFBP7 to predict a persistent AKI.

Results: Of the 100 patients included, 50 had transient AKI and 50 had persistent AKI. The RI was 0.61 ± 0.05 in the transient AKI group and 0.72 ± 0.05 in the persistent AKI group (p < 0.001). TIMP-2*IGFBP7 was not significantly different at each time between both groups. The performance of TIMP-2*IGFBP7 was poor with respectively an area under ROC curves of 0.57(95%CI 0.45–0.68), 0.58(95%CI 0.47–0.69), 0.61(95%CI 0.50–0.72), 0.57(95%CI 0.46–0.68) at H0, H6, H12 and H24. The area under the ROC curve for RI was 0.93 (95%CI 0.89–0.98). A RI > 0.685 predicting persistent AKI with 78% (95% CI 64–88) sensitivity and 90% (95%CI 78–97) specificity. The RI was neither correlated with age (rho = 0.12, p = 0.23), nor with mean arterial pressure (rho = -0.14, p = 0.16) nor with quantity of fluid (rho = 0.07, p = 0.49). Logistic regression found that RI (Odds ratio [OR] = 83.29 0.1-unit step, CI95% 14.91–465.14, p < 0.0001) and Sepsis-related Organ Failure Assessment score (OR = 1.51, CI95% 1.12–2.03, p = 0.001) predicted persistent AKI.

Conclusion: Doppler-based renal resistive index had the better performance for predicting the reversibility of AKI in critically ill patients. Urinary TIMP-2*IGFBP7 was unable to differentiate transient from persistent AKI. Further studies are needed to precise adequately the factors influencing RI.

F-44 Renin–angiotensin–aldosterone system blockers after severe acute kidney injury- use and impact on 2-year mortality

Mathilde Scarton (speaker)1, Anne Oppenheimer2, Khalil Chaibi1, Didier Dreyfuss1, Stéphane Gaudry3

1Hôpital Louis Mourier, Colombes, FRANCE; 2Hôpital Antoine-Béclère, Clamart, FRANCE; 3Hôpital Avicenne, Bobigny, FRANCE
Correspondence: Mathilde Scarton -

Annals of Intensive Care 2019, 9(Suppl 1):F-44

Introduction: Acute kidney injury (AKI) in intensive care units (ICU) carries high mortality and morbidity. Potential activation of the renin–angiotensin–aldosterone system during AKI may play a role through pro-fibrotic pathways. Renin–angiotensin–aldosterone blockers (ACEi ARB) have well known benefits for chronic kidney diseases but may be potentially nephrotoxic during AKI. Nevertheless, an ancillary study from the FROG-ICU cohort (ICM May 2018) has recently shown a lower mortality after 1 year of follow-up for patients receiving an ACEi ARB after an episode of AKI (KDIGO 1 to 3) at ICU discharge (20 109 (18%) vs 153 502 (31%), p = 0.001). The present study analyzes the use of ACEi ARB after KDIGO 3 AKI and their potential effect on long-term mortality.

Patients and methods: Ancillary of the AKIKI study (NEJM, 2016 + 375-122-133). All patients discharged alive from ICU were included and their long-term prognosis (2-year mortality) was assessed according to treatment with ACEi ARB at ICU discharge using both univariate and multivariate analyses after adjustment for potential confounding factors.

Results: Among 348 patients discharged alive, 45 (12.9%) received an ACEi ARB at ICU discharge. Table 1 details patient characteristics. Patients without ACEi ARB were more severe as attested by a higher SAPS 3 (p = 0.02) and a higher rate of catecholamine infusion (p = 0.008) during AKI. However, 2-year mortality did not significantly differ between the two groups (12 45 (27%) with ACEi ARB vs 55 303 (18%), p = 0.18). Mortality risk was not associated to non-prescription of ACEi ARB after adjustment for prognostic variables (p = 0.16).

Discussion: A substantial proportion of patients received an ACEi ARB at ICU discharge after an episode of severe AKI. We did not find a difference in mortality in favor of the renin–angiotensin–aldosterone system blockers, such as observed in the ancillary study of FROG-ICU (including 109 patients with ACEi ARB whereas the present study included 45). This could be explained by a different population (less severe AKI in FROG-ICU) and or a lack of power of our study.

Conclusion: This study does not confirm a positive effect on long-term mortality. A randomized controlled trial of ACEi ARB at ICU discharge after an episode of severe AKI is warranted.

F-45 Kinetic of uremic toxins’ concentrations during acute kidney injury and their role during endotoxemia

Pauline Caillard (speaker)1, Eleonore Ourouda-Mbaya1, Youssef Bennis1, Said Kamel1, Gabriel Choukroun2, Ziad Massy3, Julien Maizel1

1University of Picardie, Amiens, FRANCE; 2EA7517, University of Picardie, Amiens, FRANCE; 3Hopital Ambroise Paré, Paris-Ile-de-France-Ouest University, Amiens, FRANCE
Correspondence: Pauline Caillard -

Annals of Intensive Care 2019, 9(Suppl 1):F-45

Introduction: Chronic kidney disease (CKD) is associated with high mortality related to sepsis. Some studies have already shown uremic toxins’ action on vascular and immune disorders during CKD but their potential role during acute kidney injury (AKI) is undetermined. The kinetic of uremic toxin’s concentrations and their consequences during AKI disserve to be addressed. The aim of our study was to evaluate the kinetic of three uremic toxin’s concentrations (Indoxyl sulfate (IS), Para-crésyl sulfate (PCS) and FGF-23) during the first weeks of kidney failure in a uremic mice model and their consequences during endotoxemia.

Patients and methods: In this study we explored in vivo the kinetic of the three uremic toxins’ concentrations between the 7th and 45th days in controls (sham) and after kidney injury induction (KI). KI was obtained by electrocauterization followed by contralateral nephrectomy two weeks later. Uremic toxin’s concentrations were determined in sera after sacrifice at 7th, 15th and 45th days after KI induction. LPS challenge was performed (5 mg kg IP) in sham and KI groups at 15th and 45th days for survival follow up. Two other groups of KI mice fed with arabinoxylan-oligosaccharides (a chelator of PCS -group AXOS-) and sevelamer (a chelator of phosphate known to decrease FGF 23 -group sevelamer-) were exposed to LPS at 45th days of KI and followed for survival.

Results: PCS, IS and FGF 23 concentrations increased rapidly after kidney injury at day 7 and last until 45th in the KI group compared to the group sham (figure 1). This was associated with the elevation of cytokines concentration in serum (TNFα, IL-1β, and IL-6). At 15th and 45th days, all KI mice exposed to LPS died whereas all sham LPS animals survived (Fig 2). The treatment with Axos and sevelamer during the 45th days preceding the LPS challenge decreased respectively the PCS concentration (AXOS group), the FGF 23 concentration (Sevelamer group) and improved the survival of KI animals.

Conclusion: In this experimental study, the accumulation of uremic toxins appeared early in the course of kidney failure and was associated with local expression of pro-inflammatory molecules. This accumulation of toxins was associated with a higher mortality to endotoxemia. Decrease of PCS and FGF 23 concentrations was associated with the improvement of survival to LPS challenge. The early accumulation of uremic toxins and their potential role in sepsis mortality need to be confirmed during AKI in humans.

F-46 Incidence and patterns of renal recovery in critically-ill patients with hematological malignancies

Arthur Orieux (speaker)1, Jean-Baptiste Lascarrou2, Cyril Touzeau3, Amélie Seguin2, Arnaud-Felix Miailhe2, Maëlle Martin2, Jean Reignier2, Emmanuel Canet2

1CHU de Bordeaux, L’Herbergement, FRANCE; 2Médecine Intensive Réanimation - CHU, Nantes, FRANCE; 3Service d’hématologie - CHU, Nantes, FRANCE
Correspondence: Arthur Orieux -

Annals of Intensive Care 2019, 9(Suppl 1):F-46

Introduction: AKI is a dreaded complication in patients with hematological malignancies, associated with increased morbidity and mortality. Moreover, AKI may preclude the ability to receive further chemotherapy. However, current knowledge on the kinetics of renal recovery or persistent renal dysfunction is limited.

Patients and methods: We conducted a retrospective single-center study which included all patients with hematological malignancies admitted to the intensive care unit (ICU) of Nantes University Hospital from January to December 2017. Our purpose was to study the epidemiology of AKI and the occurrence of Major Adverse Kidney Events (MAKE).

Results: One hundred and three patients were included, among which 92 (89%) had AKI. Median age was 62.5 (49.75–68) years old and 59 (64%) were male. Most common malignancies were lymphoma (26, 28%), acute myeloblastic leukemia (20, 22%) and myeloma (19, 21%). 17 (18%) patients were newly diagnosed for the malignancy and 25 (27%) had relapsing diseases. Chemotherapy was administered in 64 (70%), 14 (2–31) days before ICU admission. Simplified Acute Physiology Score II (SAPS) at day 1 was 48 (39–62). During ICU stay, 37 (40%) patients received vasopressors and 28 (30%) required invasive mechanical ventilation. ICU, hospital, day-90, and day-180 mortalities were 22, 33, 38, and 43%, respectively. According to the KDIGO criteria, 47 (51%) had AKI stage 1, 16 (17%) AKI stage 2, and 29 (32%) AKI stage 3, of whom 14 (15%) required renal replacement therapy (RRT). ICU mortality of RRT patients was 78%. Overall, median duration of AKI was 5 (3–10) days, and increased from 4 (3–7) days in KDIGO stage 1 patients, to 3.5 (2.8–7.5) days in KDIGO stage 2 patients, and 12 (5–25) days in KDIGO stage 3 patients. Among survivors, renal recovery occurred in 48 (69%) patients at ICU discharge, 35 (61%) patients at day-90, and 32 (60%) patients at day-180. MAKE were reported in 32 (35%) patients at ICU discharge, 35 (38%) patients at hospital discharge, 41 (45%) 90 days after discharge, and 44 (48%) at day-180.

Conclusion: Critically-ill patients with hematological malignancies had a high incidence of AKI. Although almost 80% of the patients were discharged alive from the ICU, MAKE at day-90 accounted for roughly 1 in every 2 patients. Further research is needed to identify predictors of MAKE, so as to develop new therapeutic strategies which might translate into better long-term outcomes.

F-47 Medico-economic impact of renal replacement therapy initiation strategies in the ICU

Abirami Thiagarajah (speaker)1, Pierre-Antoine Billiet1, Anne Oppenheimer2, Julien Maizel3, Fouad Fadel4, Eric Boulet5, Guillaume Chevrel6, Laurent Martin-Lefevre7, Saad Nseir8, Didier Dreyfuss9, Stéphane Gaudry10

1APHP Louis Mourier, Colombes, FRANCE; 2APHP Antoine Béclère, Clamart, FRANCE; 3CHU Amiens-Picardie, Amiens, FRANCE; 4Centre Hospitalier René Dubos, Pontoise, FRANCE; 5Groupe Hospitalier Carnelles Portes de l’Oise, Beaumont Sur Oise, FRANCE; 6Centre Hospitalier Sud Francilien, Corbeil Essonnes, FRANCE; 7Centre Hospitalier Départemental Vendée, La Roche Sur Yon, FRANCE; 8Centre Hospitalier régional, Lille, FRANCE; 9APHP Louis Mourier, Colombes, FRANCE; 10 APHP Avicenne, Bobigny, FRANCE
Correspondence: Abirami Thiagarajah -

Annals of Intensive Care 2019, 9(Suppl 1):F-47

Introduction: Indications and modalities of renal replacement therapy (RRT) in intensive care unit (ICU) patients with acute kidney injury (AKI) are still debated. The AKIKI trial (1) showed that a delayed RRT strategy (in the absence of life-threatening condition) did not affect mortality but allowed nearly 50% patients to escape RRT compared to an early RRT strategy in patients with KDIGO3 AKI. This has obvious economic counterparts which are evaluated in this study.

Patients and methods: Financial costs of RRT (both continuous and intermittent) were assessed in 45 patients in seven French ICUs from September 2017 to August 2018 taking into account catheters, circuits, dialyzer membranes and dialysate replacement fluid prices. Medical and nursing working times including time for venous catheterization and RRT duration (including circuit preparation and restitution time only) were also recorded and their cost was computed. We then extrapolated these figures to the AKIKI population in order to estimate the cost difference between the two strategies (early and delayed). Analysis was restricted to the first 72 h after inclusion.

Results: The mean working time was 121 (+ -60) minutes for intermittent RRT and 119 (+ -91) minutes for continuous RRT. The mean financial costs for the first 72 h after RRT initiation were 111.53 (+ -44.91) euros for the intermittent RRT and 636.21 (+ - 254.05) euros for continuous RRT (p < 0.001). Extrapolating these figures to the 619 patients included in AKIKI, revealed that the early strategy was associated with a cost of 104,619 euros and the delayed one with a cost of 54,702 euros during the first 72 h after randomization.

Conclusion: This study highlights the considerable savings that can safely be obtained with a delayed RRT strategy in ICU patients with severe AKI.


(1) Gaudry S, Hajage D, Schortgen F, Martin-Lefevre L, Pons B, Boulet E, et al. Initiation Strategies for Renal-Replacement Therapy in the Intensive Care Unit. N Engl J Med. 2016 Jul 14 + 375(2)-122–33.

F-48 At-risk drinking is independently associated with acute kidney injury in non-trauma critically ill patients

Arnaud Gacouin (speaker), Mathieu Lesouhaitier, Aurélien Frerou, Benoit Painvin, Florian Reizine, Sonia Rafi, Adel Maamar , Yves Le Tulzo, Jean Marc Tadié

Service des maladies infectieuses et réanimation médicale, Rennes, FRANCE
Correspondence: Arnaud Gacouin -

Annals of Intensive Care 2019, 9(Suppl 1):F-48

Introduction: Unhealthy use of alcohol and acute kidney (AKI) injury both are major public health problems. Chronic alcohol exposure may be directly or indirectly associated with kidney damage but little is known about the impact of current excessive alcohol consumption on kidney function in non-trauma critically ill patients. We aimed to determine whether unhealthy use of alcohol is independently associated with AKI in the intensive care unit (ICU) and worst kidney function at hospital discharge.

Patients and methods: Prospective cohort study on non-cirrhotic, non-liver transplant recipients, and non-chronically dialyzed patients admitted in a 21- bed polyvalent ICU in a university hospital. The study was designed to have a 90% power to detect a 15% difference in the incidence of AKI between not at-risk drinkers and at-risk drinkers at a two-sided alpha error of 5%. At-risk dinking was defined according to the National Institute on Alcohol Abuse and Alcoholism criteria and AKI according to Kidney Disease Improving Global Outcomes (KDIGO) criteria. Patients were followed until hospital discharge or day 60.

Results: Over a 30-months period we calculated the cumulative incidence of stage 2–3 AKI in the 320 at-risk drinkers (29%) and 787 in not at-risk drinkers (71%). Stage 2–3 AKI was significantly more frequent at admission to the ICU in at-risk drinkers than in not at-risk drinkers (42.5% versus 18%, p < 0.0001). The cumulative incidence of stage 2–3 AKI was significantly higher in at-risk than in not at-risk drinkers (Figure 1) (p < 0.0001, log-rank test). After adjustment on susceptible and predisposing factors for AKI, at-risk drinking was significantly associated with AKI (Hazard ratio (HR) = 2.53 (2.08–3.08), p < 0.0001). The proportion of patients with stage 2–3 AKI at hospital discharge among survivors was significantly higher in at-risk than in not at-risk drinkers (10% versus 5% respectively, p = 0.01). At-risk drinking remained independently associated with stage 2–3 AKI in the subgroup of 832 patients without stage 2–3 AKI at admission to the ICU (HR = 2.34 (1.68–3.72), p < 0.0001).

Conclusion: Our results suggest that kidney dysfunction is significantly more frequent in at-risk than in not at-risk drinkers. We believe that systematic and accurate identification of patients with prior alcohol misuse may allow for prevention of AKI.

F-49 Incidence, Epidemiology and Prognosis of Viral Respiratory Infections in Severe Acute Respiratory Failure in Human immunodeficiency virus (HIV) infected Adults

Alexandre Elabbadi (speaker)1, Jérémy Pichon2, Benoit Visseaux3, Quentin Philippot2, Aurélie Schnuriger4, Muriel Fartoukh2, Stéphane Ruckly5, Jean-François Timsit8, Guillaume Voiriot2

1CHU Tenon, Paris, FRANCE; 2Service de Réanimation médico-chirurgicale, Hôpital Tenon, Paris, FRANCE; 3Service de Virologie, Hôpital Bichat, Paris, FRANCE; 4Service de Virologie, Hôpital Trousseau, Paris, FRANCE; 5UMR 1137-IAME Team 5-DeSCID, Inserm Université Paris Diderot, Sorbonne Paris Cité, Paris, Paris; 8Service de Réanimation médicale et infectieuse, Hôpital Bichat, Paris, FRANCE
Correspondence: Alexandre Elabbadi -

Annals of Intensive Care 2019, 9(Suppl 1):F-49

Introduction: Acute respiratory failure remains the main reason for admission to intensive care in HIV-infected Adults. There is little data on viral epidemiology in lower respiratory tract infections in this population.

Patients and methods: Cases of acute respiratory failure in HIV-infected adults admitted to two intensive care units between 2011 and 2017, who underwent screening for respiratory virus by multiplex polymerase chain reaction, were retrospectively selected.

Results: A total of 123 cases were included. An HIV infection was newly diagnosed in 9% of cases and 72% of the population was taking antiretroviral therapy with treatment compliance in almost 76% of cases. A documented viral respiratory infection was found in 33 patients (27%). Rhinovirus was the main virus, found in 33% of cases (n = 15) followed by Parainfluenza (n = 5) and Influenza A (n = 5). A co-infection was found in 22 patients (67% of cases) with only one virus–virus co-infection. Overall, neither the level of HIV-related immunodeficiency nor the use of ARVs at admission seems to be associated with an increased risk of respiratory viral infection. Nevertheless the subpopulation of Rhinovirus infection was associated with a low CD4 count. Outcome in the ICU was similar regardless of whether or not a respiratory viral infection was present.

Conclusion: Respiratory viruses are frequently found during the acute respiratory failure of the HIV subject. This proportion appears to be similar to other studies that have looked at the proportion of respiratory viruses in the general population.

F-50 Is Human Metapneumovirus identified in the Respiratory Tract of Immunocompromised Patients with Acute Respiratory Failure Clinically Relevant?

Natacha Kapandji (speaker), Jerome Le Goff, Maud Salmona, Michael Darmon, Elie Azoulay, Virginie Lemiale

Hôpital Saint Louis, Paris, FRANCE
Correspondence: Natacha Kapandji -

Annals of Intensive Care 2019, 9(Suppl 1):F-50

Introduction: Human metapneumovirus (hMPV) may be responsible for pulmonary infections in immunocompromised patients. We sought to assess clinical and radiological characteristics associated with the presence of hMPV in the respiratory tract of critically ill immunocompromised patients.

Patients and methods: This single center retrospective cohort included adult immunocompromised patients admitted to intensive care unit (ICU) in whom hMPV was detected in respiratory tract from September 2010 to June 2018. Results are reported as n (%) and median (IQR). Comparison were performed by Fisher exact test or Wilcoxon test as appropriate.

Results: Of the 1185 immunocompromised patients admitted to ICU during this period, 26 (2.1%) patients presented hMPV- 15 (58%) males, aged 66 (IQR 56–74), 21 (81%) with hematological malignancies including 11 (42%) allogeneic stem cell transplant recipients. In addition to respiratory failure, 5 (19%) presented with shock. Clinical picture included fever in 23 patients (89%), cough in 19 (73%), whereas extrapulmonary symptoms were less common (11 to 31%). CT-scan patterns included alveolar consolidations in 14 (54%) patients, ground glass opacities in 10 (38%), and septa thickening in 3 (12%). All but one patient presented another pathogen- bacterial infection in 15 patients (70%), viral infection in 8 (31%) and fungal infection in one (invasive aspergillosis). At ICU admission, SAPS2 score was 41 (IQR 37–58), mechanical ventilation being required in 11 patients (42%), vasopressors in 11 (42%) and renal replacement therapy in 4 (15%). Hospital mortality rates was 31% and was found to be associated with hemodynamic failure, renal replacement therapy and neutropenia.

Conclusion: hMPV is found in the respiratory tract of 2.1% of immunocompromised patients with acute respiratory failure, and in all but 8 cases, another pathogen is also identified. CT patterns have nothing unique and could be ascribable to the second pathogen. Whether positive hMPV is a relevant result is unclear and remains unaddressed by this set of data. Comparing clinical and radiographic pictures and outcomes of patients with positive hMPV to matched patients with documented influenza infections is ongoing and will be displayed at the conference.

F-51 Relative lymphopenia in patients with severe influenza A in ICU- Incidence and relation with severity illness

Helmi Amri (speaker)1, Amira Jamoussi1, Samia Ayed1, Takoua Merhebene1, Dhouha Lakhdher1, Amine Slim2, Jalila Ben Khelil1, Mohamed Besbes1

1Medical Intensive Care Unit, Abderrahmen Mami pneumology hospital, Ariana, TUNISIA; 2Virology Unit, Microbiology Laboratory, National Influenza Centre, Charles Nicolle’s Hospital, Tunis, TUNISIA
Correspondence: Helmi Amri -

Annals of Intensive Care 2019, 9(Suppl 1):F-51

Introduction: Relative lymphopenia (RL) is considered when lymphocytes count is ≤ 21% of all white blood cells. It has been reported as a possible marker of influenza A and especially pandemic influenza A H1N1 2009 infection. The aim of this study was to assess RL incidence in patients hospitalized in intensive care unit (ICU) for influenza A (H1N1 or H3N2) infection and to find whether it was related to severity illness.

Patients and methods: This was a retrospective monocentric study conducted in the medical ICU of Abderrahmen Mami hospital in Tunisia, between September 2009 and September 2018. Inclusion criteria were ICU patients admitted for influenza A infection. We collected clinical, biological and outcome data. Then we compared initial severity and outcome between RL patients group and normal lymphocyte count patients group.

Results: During the ten-year study period, we collected 72 patients with influenza A infection. The strain detected with PCR technique was H1N1 (n = 65) and H3N2 (n = 7). The mean age was of 48.2 ± 15.2 years and the sex-ratio M F was of 1.4 (42 30). Seven women were pregnant. White blood cell (WBC) value was found to be normal in 47.2% and high in 38.9% of patients with influenza A infection. Leucopenia was seen in only 10 patients (13.9%). Thrombocytopenia (≤ 150 × 103 platelets mm3) was present in 24 patients (30.6%). RL was present in 56 patients (77.8%). Ratio Lymphocyte Monocyte < 2 was noticed in 7 patients (9.7%). At admission, severity was assessed by SAPS II score (31.2 ± 16.5) and APACHE II score (12.5 ± 8.12). The main reason for ICU admission was acute respiratory failure (97.2%) from whom 35 patients (48.6%) had ARDS. Non-invasive ventilation was needed in 68% patients and 45.8% were intubated. The mean duration of mechanical ventilation was 7.96 days [0–47].The mean length of stay in ICU was 10.56 days [1–65] and the ICU mortality rate was 33.3%. Comparison of severity illness components according to RL presence are represented in table 1.

Conclusion: RL is frequent in ICU patients with influenza A infection. It is significantly more frequent in older patients. RL is associated to higher severity scores, but not to respiratory impairment degree or death.

Table 1. Univariate analysis: Severity illness components according to RL


RL (n = 56)

NL (n = 16)


Age mean ± SD; years

50.09 ± 15.02

41.69 ± 14.36



13.95 ± 8.34

7.38 ± 4.54


SAPS II mean ± SD

33.98 ± 17.02

21.44 ± 10.01


PaO2/FiO2 mean ± SD; mm Hg

165.39 ± 104.28

157 ± 78.41


ARDS n (%)

26 (46.4%)

9 (56.2%)


IMV n (%)

29 (51.7%)

4 (25%)


ICU mortality n (%)

21 (37.5%)

3 (18.7%)


RL: relative lymphopenia; NL: normal lymphocyte; ARDS: acute respiratory distress syndrome; IMV: invasive mechanical ventilation

F-52 Critically ill influenza patients- co-infections and cardiovascular events

Erwan Begot (speaker)1, Renaud Prevel2, Fabrice Camou3, Walter Picard4, Chloé Gisbert-Mora5, Alexandre Boyer3, Didier Gruson3

Correspondence: Erwan Begot -

Annals of Intensive Care 2019, 9(Suppl 1):F-52

Introduction: Critically ill influenza patients have a mortality rate about 20%. Co-infections, especially invasive aspergillosis, have recently been documented amongst these patients and are thought to play a major role in the prognosis of these patients. Nevertheless, the need of systematic microbiological documentation is still not widely accepted by ICU health workers. Moreover, a high incidence of cardiovascular events (acute coronary syndrome or stroke) within the first 15 days after the influenza illness has been described for a few years. The aims of our study were to assess the daily practice regarding microbiological investigation of these coinfections and to assess the cardio-vascular events rates.

Patients and methods: A retrospective, observational study from November 2017 to April 2018 in four French ICUs including all critically ill influenza patients was conducted. Influenza diagnosis was confirmed by polymerase chain reaction or rapid test. Co-infections were confirmed using clinical and standard microbiological criteria. Stroke was diagnosed by sudden neurological symptom with MRI or contrast tomodensitometry confirmation. Acute coronary syndrome was diagnosed according to ESC guidelines.

Results: 111 patients were included. 98/106 patients (92%) were classified as suffering from Acute Respiratory Syndrome. 69/111 patients (62%) received non invasive ventilation during 5.7 (± 2.8) days. 57/111 patients (51%) received invasive mechanical ventilation for 10.5 (± 8.9) days. 76/111 patients (69%) have been treated by oseltamivir for 5.1 (± 1.9) days. 81/111 (73%) were still alive at Day 28. 90/111 (81%) had a bacterial sputum culture with 38/90 (42%) suffering from a bacterial co-infection, mainly S. pneumoniae and Methicillin-Susceptible Staphylococcus aureus. Only 22/111 (20%) patients had a fungal sputum culture and 27/111 (24%) a blood galactomannan dosage. Amongst those 22 patients, 5 exhibited fungi in their sputum culture, 2 of which corresponding to Aspergillus spp. and 3 to Candida spp. Viral co-infection occurred among 14/73 patients (19%), especially by Coronavirus. Regarding cardiovascular events, acute coronary syndrome occurred in 11/111 patients (10%) and stroke in 2/111 patients (1.8%) during the month after the diagnosis of influenza.

Conclusion: Critically ill influenza patients are frequently co-infected, by bacteria but also by virus and fungi. Fungal and viral coinfections seem to be still under investigated by ICU workers. Further studies investigating the prophylaxis and management of these coinfections are needed. Cardiovascular events prevention should also be studied among critically ill influenza patients.

F-53 Human metapneumovirus and risk of lower respiratory tract infection, ICU admission and hospital mortality- results of a systematic review and meta-analysis

Natacha Kapandji (speaker), Virginie Lemiale, Lara Zafrani, Elie Azoulay, Michael Darmon

Hôpital Saint Louis, Paris, FRANCE
Correspondence: Natacha Kapandji -

Annals of Intensive Care 2019, 9(Suppl 1):F-53

Introduction: Impact of human metapneumovirus (hMPV) on lower respiratory tract infections is unclear in both children and adult patients. The aim of this review was to evaluate the prevalence of lower respiratory tract infections (LRTI), need for intensive-care-unit (ICU) admission and hospital mortality in immunocompromised adults with previous detection of hMPV.

Patients and methods: This systematic review was performed according to PRISMA statements and registered in the PROSPERO database (CRD42018106617). Studies reporting rate of LRTI, ICU admission and mortality were searched on PubMed (2008–2018) for immunocompromised patients. Prevalence and its confidence interval (95%CI) were plotted. Publication bias was assessed by visually inspecting the funnel plot and summary estimates of relative risk and their 95% confidence interval were calculated using random-effects model.

Results: Overall, 42 citations were identified and 29 studies, reporting 1407 patients, were ultimately included (21 cohort studies and 8 case series). Median sample size was 13 patients (IQR 7–48) and median admission year was 2009 (IQR 2008–2012). Diagnosis of hMPV infection was performed using RT-PCR on nasopharyngeal swab in 22 studies (76%). Proportion of patients with hematological malignancies was 97% (IQR 0–100), and proportion of allogeneic stem cell transplantation was 11% (IQR 0–47). LRTI prevalence was 60% (95%CI 48–73 + I2 = 94%). ICU admission was required in 20% of patients (95%CI 14–27 + I2 = 90%). Last, hospital mortality was 6% (4–9 + I2 = 79%). Factors associated with heterogeneity were assessed using meta-regression. Heterogeneity of the results were partly explained by study design, proportion of patients with hematological malignancy and proportion of patients with co-infection.

Conclusion: In this systematic review, two third of immunocompromised patients in whom hMPV was detected had LRTI, and as many as 20% required ICU admission. A high heterogeneity was noted that may be explained by study design, underlying immune status and underlying disease and rate of co-infections. Most of the included studies were however at high risk of bias justifying need for additional studies in this field.

F-54 Impact and characteristics of respiratory viruses infection during severe acute respiratory failure in adults

Jérémie Pichon (speaker)1, Lila Bouadma2, Benoit Visseaux3, Aurélie Schnuriger4, Jean-François Timsit2, Muriel Fartoukh5, Guillaume Voiriot6

1CHU Tenon, Paris, FRANCE; 2Service de Réanimation médicale et infectieuse, Hôpital Bichat, Hôpitaux Universitaires Paris Nord Val de Seine, APHP, FRANCE; 3Université Paris Diderot, Sorbonne Paris Cité, Service de Virologie, Hôpital Bichat, Hôpitaux Universitaires Paris Nord Val de Seine, APHP, FRANCE; 4Service de Virologie, Hôpital Trousseau, Hôpitaux Universitaires de l’Est Parisien, APHP, Sorbonne Université, Paris, FRANCE; 5Service de Réanimation médico-chirurgicale, Hôpital Tenon, Hôpitaux Universitaires de l’Est Parisien, APHP, Sorbonne, Paris, FRANCE
Correspondence: Jérémie Pichon -

Annals of Intensive Care 2019, 9(Suppl 1):F-54

Introduction: The mPCR (multiplex Polymerase Chain Reaction) tests are widely used in clinical routine. However, clinical and paraclinical features of virus-associated respiratory tract infections in adults are poorly described, especially in acute respiratory failure requiring critical care.

Patients and methods: This was a prospective non-interventional 2-center clinical study. All adult patients admitted to intensive care unit for an acute respiratory failure and subjected to a multiplex PCR (16 virus, nasopharyngeal swab and or distal sample) during the first 72 h of ICU stay were included. Prior to get the mPCR results, clinicians were invited to subject patients to a questionnaire asking for 37 symptoms (respiratory and extra-respiratory) coded from 0 (none) to 4 (maximum). Other sources may be examined, such as patient relatives, general practitioner or medical reports.

Results: From December 2015 to April 2017, 339 patients were included (209 men, age 63 years [50–73]), of whom 82 were intubated. The final diagnosis was pneumonia in 121 patients, exacerbation of chronic bronchial disease in 117 subjects and another diagnosis in 101 subjects. At least one virus was identified in 121 patients, including 40 Rhinovirus, 34 Influenza, 13 Respiratory syncytial virus, 14 Metapneumovirus and 15 Coronavirus. In univariate analysis, an active smoking, a close contact with somebody ill, a rhinorrhea, a bronchorrhea, an earache, a diarrhea, a lymphopenia, a thrombocytopenia and elevated CPK were associated with the documentation of respiratory viruses within airways.

Conclusion: A phenotype suggesting a virus-associated respiratory tract infection was described. It may help to rationalize the use of mPCR tests.

F-55 Transfusion of packed red blood cells is associated with an increased risk of ICU-acquired infections and mortality in septic shock patients

Edwige Peju (speaker)1, Jean-François Llitjos1, Julien Charpentier1, Alain Cariou1, Jean-Daniel Chiche1, Jean-Paul Mira1, Matthieu Jamme2, Frédéric Pène1

1Médecine-Intensive Réanimation, Paris, FRANCE; 2Urgences néphrologiques et transplantation rénale, Paris, FRANCE
Correspondence: Edwige Peju -

Annals of Intensive Care 2019, 9(Suppl 1):F-55

Introduction: Transfusion of packed red blood cells (RBC) is commonly indicated in septic patients to improve tissue oxygen delivery. Besides uncertain benefits, RBC transfusions carry immunomodulatory properties likely to increase the susceptibility to further ICU-acquired infections or the mortality rate. The aim of this study is to address the impact of RBC transfusion on ICU-acquired infections and mortality in septic shock patients.

Patients and methods: This was a 10-year (2008–2017) monocenter retrospective study. All consecutive adult patients diagnosed for septic shock within the first 48 h were included. Septic shock was defined as a microbiologically proven or clinically suspected infection, associated with acute circulatory failure requiring vasopressors. The number of packed RBC and the transfusion day were recorded. The diagnosis of nosocomial infections was based on current international guidelines. Patients alive at day 3 were evaluated for the risk of ICU-acquired infections. The determinants of ICU-acquired infections and 30-day mortality were addressed in a multivariate time-dependent Cox regression analysis.

Results: Among 1152 patients admitted for septic shock, 63% were males, the median age was 69 (57–79) years old and the crude 30-day mortality rate was 31.5%. Overall, 512 patients (44%) received RBC transfusions, with a median of 3 (2–6) units. In multivariate analysis, transfusion was independently associated with prior immunosuppression (OR = 1.57, 95%IC [1.21–2.04], p = 0.001), with chronic kidney disease (OR = 1.89, 95%IC [1.30–2.74], p = 0.001) and a higher admission SOFA score (OR = 1.03, 95%IC [1.01–1.06], p = 0.007). 1038 patients were alive at day 3, of whom 253 (24%) developed ICU-acquired infections. Among them, 197 (78%) patients had received RBC prior to the episode of ICU-acquired infection. In multivariate analysis, RBC transfusion was independently associated with the development of ICU-acquired infections (OR = 1.88, 95%IC [1.21–2.93], p = 0.005). 30-day mortality was significantly higher in patients receiving RBC (37.6% vs. 27.8%, p < 0.001). In multivariate analysis, RBC transfusion was independently associated with 30-day mortality (OR = 1.50, IC95% [1.07–2.09], p = 0.02).

Conclusion: Transfusion of RBC in septic shock patients is associated with an increased risk of ICU-acquired infections and 30-day mortality. These results support a restrictive RBC transfusion policy in septic shock patients.

F-56 Effect of age on mortality in patients with haematological malignancy in intensive care units

Jean-Edouard Martin (speaker)1, Michael Darmon1, Virginie Lemiale1, Djamel Mokart2, Frédéric Pène3, Achille Kouatchet4, Julien Mayaux 5, Francois Vincent6, Martine Nyunga7, Fabrice Bruneel8, Christine Lebert9, Pierre Perez10, Anne-Pascale Meert11, Dominique Benoit12, Rebecca Hamidfar13, Mercé Jourdain14, Lionel Kerhuel1, Laure Calvet1, Etienne Ghrenassia1, Samir Jaber15, Elie Azoulay1, Audrey De Jong15

1Saint-Louis Hospital, Paris, FRANCE; 2Institut Paoli Calmette, Marseille, FRANCE; 3Cochin Hospital, Paris, FRANCE; 4Centre Hospitalier Universitaire, Angers, FRANCE; 5Pitié-Salpétrière Hospital, Paris, Paris; 6Avicenne Hospital, Bobigny, FRANCE; 7Victor Provo Hospital, Roubaix, FRANCE; 8Mignot Hospital, Versailles, FRANCE; 9Montaigu Hospital, La Roche-Sur-Yon, FRANCE; 10Brabois Hospital, Nancy, FRANCE; 11Institut Jules Bordet, Bruxelles, BELGIUM; 12Ghent University Hospital, Ghent, BELGIUM; 13Albert Michallon Hospital, Grenoble, FRANCE; 14Salengro Hospital, Lille, FRANCE; 15CHU, Montpellier, FRANCE
Correspondence: Jean-Edouard Martin -

Annals of Intensive Care 2019, 9(Suppl 1):F-56

Introduction: Respective influence of age and performance status (PS) of critically ill patients with malignancy has been poorly studied. The main objective of this study was to analyse the impact of age on day-90 mortality in this population.

Patients and methods: We performed a post hoc analysis of prospective multicentre data from France and Belgium to identify the relation between age and day-90 mortality. Five classes of age were computed according to quintile of ages. The best threshold of age was determined using a Youden index analysis. Univariate and multivariate Cox analysis of day-90 mortality were performed. Kaplan–Meier curves of day-90 mortality according to age and main risk factors were computed.

Results: 1011 patients were included. Age categories according to quintile repartition were the following- 18 to 45 years old (y) (n = 206), 46 to 56 y (n = 194), 57 to 63 y (n = 213), 64 to 71 y (n = 189), and 72 to 87 y (n = 209). Next, two age groups was separated- younger group, age < 64 y and older group age ≥ 64 y with a median age of 52 y (25–75% IQR, 41–59) and 72 y (67–77) respectively. Older age (≥ 64) was significantly associated with higher mortality rate in univariate analysis (HR = 1.54 (1.26–1.88), p < 0.0001). After multivariate cox analysis, main risk factor for mortality were age ≥ 64 (HR = 1.55 (1.25–1.92), p < 0.0001), PS ≥ 3 (HR = 1.47 (1.40–1.87), p = 0.0024) and severity at inclusion assessed by sequential organ failure assessment (SOFA) score at admission in ICU per one unit increase (HR = 1.18 (1.15–1.21), p < 0.0001). However, haematological malignancy, Charlson comorbidity score index without age and reason for ICU admission were not significantly associated with day-90 mortality. Figure 1 shows the Kaplan–Meier survival analysis of the relation between age, PS and day-90 mortality (p < 0.0001).

Conclusion: Older age is significantly associated with higher day-90 mortality. A threshold of 64 y was found to be the most accurate to discriminate dead from alive patients. Despite its prognostic impact, survival was meaningful in subgroups of oldest patients with moderately limited autonomy. This study may allow a better selection of oldest patients likely to benefit of ICU admission according to three simple variables- age, PS and SOFA score.

F-57 Hemodynamic failure in critically ill patients with hemophagocytic syndrome

Thomas Frapard (speaker)1, Sandrine Valade1, Eric Mariotte1, Jehane Fadlallah2, Lionel Galicier2, Michael Darmon1, Elie Azoulay1

1Médecine intensive et réanimation Saint Louis, APHP, Paris, FRANCE; 2Immunopathologie clinique Saint Louis, APHP, Paris, FRANCE
Correspondence: Thomas Frapard -

Annals of Intensive Care 2019, 9(Suppl 1):F-57

Introduction: Hemophagocytic syndrome (HS) is a rare life-threatening condition that can lead to multi organ failure, including shock. In severe HS, symptomatic treatment relies on Etoposide (VP16) infusion. Hemodynamic instability during HS has been poorly studied. Objectives of this study were to describe the characteristics of HS patients with shock, prognostic factors and the impact of etoposide injection on the hemodynamic parameters.

Patients and methods: Adult critically ill patients with HS and managed in a multidisciplinary national reference center between 2007 and 2017 were retrospectively included. Patients without vasopressors or not requiring Etoposide infusion were excluded.

Results: Forty patients were included. Two-third (n = 28) were of male gender and median age was 48y [IQR 37–62]. Shock (n = 15, 37%), acute respiratory failure (n = 10, 25%) and monitoring (n = 8, 20%) were the main reasons for ICU admission.

The most common HS-triggers were underlying hematologic disease (malignancies HHV8-related disease) in 31 patients (77%), infectious diseases in 4 (10%), and systemic rheumatic diseases in 3 (8%). Median SOFA score was 11 [9–13], 85% of the patients required mechanical ventilation (n = 34) and median lactate level was 4 mmol/L [2.7–6.9]. Hospital mortality was 47% (n = 19) and was associated with severity as assessed by need for mechanical ventilation (100% vs. 73% + P = 0.04) and male gender (90% vs. 47% in survivors + P = 0.01). Etoposide infusion (H0) was followed by increased norepinephrine doses (P = 0.03) and a trend toward higher lactate levels (P = 0.07 + figure 1). No statistically significant change was observed as regard to mean arterial pressure, heart rate and renal function, assessed by serum creatinine and oliguria.

Conclusion: Our results suggest a high severity of HS patients with acute circulatory failure, a high hospital mortality and a hemodynamic worsening in the 24 h following etoposide infusion. Change in hemodynamic through ICU stay and comparison to patients with cytopenias and shock in the absence of HS are currently ongoing and will be presented at the congress.

F-58 Long-term prognosis of high-grade glioma admitted in the intensive care unit

Maxens Decavèle (speaker)1, Nicolas Gatulle2, Nicolas Weiss3, Léa Lemasle2, Ahmed Idbaih5, Julien Mayaux2, Thomas Similowski2, Alexandre Demoule2

1Hôpital La Pitié-Salpêtrière, Paris, FRANCE; 2AP-HP, Groupe Hospitalier Pitié-Salpêtrière Charles Foix, Service de Pneumologie et Réanimation Médicale (Département R3S), Paris, FRANCE; 3Unité de Réanimation Neurologique, Département de Neurologie, Pôle des Maladies du Système Nerveux et Institut de Neurosciences Translationnelles, IHU, Paris, FRANCE; 4Inserm U 1127, CNRS UMR 7225, Sorbonne Universités, UPMC Univ Paris 06, Institut du Cerveau et de la Moelle épinière, ICM. AP-HP, Paris, FRANCE
Correspondence: Maxens Decavèle -

Annals of Intensive Care 2019, 9(Suppl 1):F-58

Introduction: Only limited data are available concerning prognosis of primary malignant brain tumors in the intensive care unit (ICU). Among them, high-grade gliomas (HGG) are the most frequent and those associated with the poorest survival. Whereas long-term prognosis after ICU admission of patients with other malignancies is now better known, no such data exist regarding patients with HGG. The aims of our study were 1) to analyze factors associated with 1-year mortality in patients with HGG admitted to the ICU and 2) to assess the functional status and anti-cancer therapy course in ICU survivors.

Patients and methods: Eight-year, bicentric, retrospective cohort study. All consecutive patients with HGG, admitted to the ICUs were included. Functional status was assessed with the Karnofsky Performance Status (KPS). Mutation in isocitrate dehydrogenase (IDH) 1 and 2 was also collected. The anti-cancer therapy course after ICU discharge was classified in 1) continued without change, 2) changed (modified or stopped after ICU discharge), and 3) initiated after ICU discharge for a HGG diagnosed during the ICU stay).

Results: Seventy-eight patients (age 58 [45–67] year-old, SAPSII 32 [21–52]) were included, of which 62 (79%) were glioblastoma. Main reasons for admission were coma and acute respiratory failure (47 (60%) and 17 (22%), respectively). Mechanical ventilation and vasopressors were required in 40 (51%) and 17 (22%) of cases. ICU and 1-year mortality was 12 (15%) and 62 (79%). Among ICU survivors, anti-cancer therapy course was continued, changed and initiated in 26 (33%), 41 (53%) and 11 (14%) patients, respectively. One-year survival was significantly higher in patients in whom anti-cancer therapy was continued, as compared to others (16 (62%) vs. 46 (92%), p = 0.002). Among ICU survivors, the KPS did not vary between 1 month and 1 year after ICU discharge (55 [48–78] vs. 60 [50–58], p = 0.212). In multivariate analysis, factor associated with 1-year survival were the KPS at admission (OR 0.894 95%CI [0.807–0.953], p = 0.005), and anti-cancer therapy course continued (OR 0.009 [0.001–0.102], p = 0.002). The IDH status did not impact on 1-year mortality.

Conclusion: Despite high long-term mortality rate, 85% of patients survived to the ICU, near a half continued their planned anti-cancer therapy course and more than 20% were alive 1 year after ICU discharge, with good functional status.

F-59 Early Identification of Sickle-Cell Disease Patients at Risk for Complicated Outcome in Intensive Care Unit. Aggregation of multiple sickle cell prediction model, updating and validation on CARADBDREPA cohort

Amélie Rolle (speaker)1, N’Guyen Tri Long2, Thomas P. A. Debray3, Zakaria Mahi1, Bertrand Pons1, Ruddy Valentino4, Hossein Mehdaoui 4, Michel Carles1

1CHUG, Pointe-à-Pitre, FRANCE; 2CHU Montpellier_Nimes, Montpellier, FRANCE; 3Julius Center for Health Sciences and Primary Care, Utrecht, THE NETHERLANDS; 4CHUM, Fort-De-France, FRANCE
Correspondence: Amélie Rolle -

Annals of Intensive Care 2019, 9(Suppl 1):F-59

Introduction: Sickle cell disease (SCD) is an increasing global health problem. Better prediction of the severity of sickle cell disease could lead to more precise a treatment and management. However, though several prediction models for SCD have been published, their external validity remains unclear. Objectives- The objective of our study was to validate existing prediction models SCD patients at risk for complicated outcome, and to combine and simultaneously update these models into a new so-called ‘meta-model’. Primary endpoint- Composite, binary outcome (‘complicated outcome’) defined by intensive care unit stays > 2 days, need for vital support or death.

Patients and methods: From January 2012 to December 2017, a retrospective cohort study was conducted at the University Hospital of Guadeloupe (French territories in the Americas). It included all patients (children and adults) admitted to this institution. To combine existing prediction models into a ‘meta-model’, we performed a systematic review searching in PubMed, EMBASE and bibliographies of articles retrieved. We screened relevant studies that enrolled sickle-cell disease children and adult patients for whom a prediction model was presented.

Results: Of the 3829 patient admissions in our cohort, 210 (23%) experienced a complicated outcome, defined as death (8 patients, 3.8%), acute respiratory failure (119 patients, 56%) hemodynamic failure (24 patients, (11%), or renal failure (12 patients, 6%). Four prediction models were combine to predict patient risk of complicated outcomes. The resulting meta-model demonstrated good predictive performances in terms of discrimination (c-statistic- 0.8) and calibration.

Conclusion: Combining existing prediction models might help clinicians obtain valid predictions of complicated outcomes and rapidly improve the quality of care of SCD patients admitted to emergency department.

F-60 Performances of HLH criteria and H-Score in ICU patients with severe hemophagocytic syndrome

Sandrine Valade (speaker), Grégoire Monseau, Laure Calvet, Eric Mariotte, Virginie Lemiale, Zafrani Lara, Elie Azoulay, Michaël Darmon

Réanimation Médicale, Hôpital Saint-Louis, AP-HP, Paris, FRANCE
Correspondence: Sandrine Valade -

Annals of Intensive Care 2019, 9(Suppl 1):F-60

Introduction: Hemophagocytic syndrome (HS) is a serious condition that can lead patients to intensive care unit (ICU) admission. Diagnosis may be difficult in these patients who may have multiple organ failures. HLH criteria are the most commonly used, but a new diagnostic score has recently been established (the H-Score). The main objective of this study is to analyze diagnostic performance of these diagnostic scores in ICU patients.

Patients and methods: Two convenient samples were analyzed including a sample of 150 patients with confirmed HS (HS +). A second sample of 1011 patients without HS (HS-) was obtained from a multicenter cohort of onco-hematological patients. Results are presented as median (interquartile range) and numbers (%).Area under ROC curves were established to assess discriminancy of both scores in diagnosing HS. A sensitivity analysis was performed after propensity score (PS) matching according to temperature and cytopenia.

Results: Overall, 1161 patients were included in this study. HS + patients were younger (median age 48.5 years [38–59] vs 60 [49–70], p < 0.001), had more severe cytopenia (hemoglobin 8.3 g/dL [7.23–9.17] and platelets 44000 mm3 [21000–79000] vs 62000 [29000–140000]), had more often organomegaly (hepatomegaly in 68.7% vs 8%, splenomegaly in 61.3% vs 9%). Mortality rate was 45.8% in hemophagocytic patients and 38.8% in control patients. Median H-Score was 235 [205–262] in SH + patients and 42 [18–62] in SH- patients. Number of HLH criteria was 4 [4–5] and 1 [0–1] respectively. Diagnostic performance of both score was excellent with area under ROC curve of 0.99 (95%CI according to DeLong Method of 0.99–0.99) and 0.99 (95%CI 0.99–0.99) for HLH and H-score respectively (figure). After propensity score matching (n = 144*2), the median H-Score was of 234 [205–262] in SH + patients versus 49 [18–71] in SH- patients. Median number of HLH criteria were 4 [4–5] in SH + and 1 [0–1] in SH- patients. Area under ROC curve was of 0.98 (CI95% 0.96–0.99) for HLH criteria and 0.99 (CI95% 0.99–1) for H-Score.

Conclusion: H-Score and HLH criteria are highly sensitive and specific in ICU patients. Further studies in unselected cohort of consecutive ICU patients with suspected HS are warranted in order to confirm our results and optimal cut-off for these scores.

F-61 Post-transfusion platelet increments in critically ill cancer patients with hypoproliferative thrombocytopenia

Elodie Baron (speaker)1, Anne François2, Julien Charpentier1, Habib Ben Hadj Amor3, Bassem Habr1, Alain Cariou1, Jean-Daniel Chiche1, Jean-Paul Mira1, Matthieu Jamme4, Frédéric Pène1

1Réanimation, CHU Cochin, Paris, FRANCE; 2EFS, CHU HEGP, Paris, FRANCE; 3CHU Cochin, EFS, Paris, FRANCE; 4UNTR, CHU Tenon, Paris, FRANCE
Correspondence: Elodie Baron -

Annals of Intensive Care 2019, 9(Suppl 1):F-61

Introduction: Thrombocytopenia is a common disorder in intensive care unit (ICU) and is associated with an increased risk of bleeding. Most data about platelet transfusions in the ICU have been obtained from general cohorts with peripheral thrombocytopenia and ongoing active bleeding or subjected to invasive procedures. In patients with hypoproliferative thrombocytopenia, the management of platelet transfusions remains somewhat empirical, derived from studies performed in hematology patients under stable clinical conditions. We herein described and analysed the determinants of post-transfusion platelet increments in cancer patients with hypoproliferative thrombocytopenia in the ICU.

Patients and methods: This was a single-center retrospective observational study over a 9-year period (2009–2017). Patients with malignancies and hypoproliferative thrombocytopenia who had received at least one platelet transfusion in the ICU were included. For each transfusion episode, a poor platelet yield was defined as a body surface area-adjusted corrected count increment (CCI) < 7, or alternatively as a weight-adjusted platelet transfusion yield (RTP) < 0.2. Patients were considered refractory to platelet transfusions when they experienced poor platelet increments (CCI < 7 or RTP < 0.2) following two consecutive ABO-compatible transfusions containing at least 0.5x1011 platelets per 10 kg bodyweight.

Results: 326 patients who received a total of 1470 platelet transfusions were analyzed. Indications for platelet transfusions were distributed into prophylactic (44.5%), securing an invasive procedure (18.1%) and therapeutic for active bleeding (37.4%). Transfusion thresholds were lower for prophylactic indications than for securing an invasive procedure or for therapeutic indications (13 [8–22] G L vs. 20 [13–31] G L vs. 21 [11–36] G L, respectively). Regardless of indications, 54.6% and 55.4% of transfusion episodes were associated with a CCI < 7 or a RTP < 0.2. Compared to prophylactic indications, the transfusion yields were better when securing an invasive procedure. In multivariate analysis, the factors associated with poor post-transfusion increments were lower body mass index (BMI), severity on the day of transfusion, depth of pre-transfusion thrombocytopenia, time between platelet transfusion and post-transfusion platelet count, fever ≥ 39 °C, antibiotic therapy, and storage duration of platelet concentrates. 48 patients developed refractoriness to platelet transfusion, associated with lower BMI, stem cell transplantation and spleen enlargement.

Conclusion: Platelet transfusions are often associated with poor increments in critically ill cancer patients with hypoproliferative thrombocytopenia. Our data suggest ways to improve the efficiency platelet transfusion in this setting.

F-62 Differential Clinical Characteristics, Management, and Outcome Of Delirium in Ward and ICU Patients

Emmanuel Canet (speaker)1, Sobia Amjad2, Raymond Robbins2, Jane Lewis2, Michelle Matalanis2, Daryl Jones2, Rinaldo Bellomo2

1Nantes, FRANCE; 2Austin Health Hospital, Melbourne, AUSTRALIA
Correspondence: Emmanuel Canet -

Annals of Intensive Care 2019, 9(Suppl 1):F-62

Introduction: To study patient demographics, clinical phenotype, management, and outcomes of patient with delirium in hospital wards compared to the ICU.

Patients and methods: Cohort of patients admitted to an Australian university-affiliated hospital between March 2013 and April 2017 and coded for delirium using the ICD-10 criteria.

Results: Among 61,032 hospitalized patients, 2,864 (4.7%) were coded for delirium. From these, we selected a random sample of 100 ward patients and 100 ICU patients for detailed analysis. Ward patients were older (median age- 84 vs. 65 years + P < 0.0001), more likely to have pre-existing neurological disease (53% vs. 13% for ICU patients + P < 0.0001) and less likely to have had surgery (24 vs. 62% + P < 0.0001). Of ward patients, 74% had hypoactive delirium, while 64% of ICU patients had agitated delirium (P < 0.0001). Persistent delirium at hospital discharge was more common among ward patients (66% vs 17%, p < 0.0001). On multivariate analysis, age and pre-existing neurological disease predicted persistent delirium, while surgery predicted recovery.

Conclusion: Delirium in ward patients is profoundly different from delirium in ICU patients. It has a dominant hypoactive clinical phenotype, is preceded by chronic neurological conditions, is managed with fewer drugs and is less likely to recover at hospital discharge.

F-63 Traumatic quadriplegia- diagnostic and therapeutic strategy. (About 75 patients)

Amine Benhamed (speaker), Amel Zerhouni, Medjahed Medjahed, Lahcen Senhadji, Radouane Rachi

Faculté de medecine Oran, Oran, ALGERIA
Correspondence: Amine Benhamed -

Annals of Intensive Care 2019, 9(Suppl 1):F-63

Introduction: trauma to the spine is a common pathology that is constantly increasing in Algeria, mainly to road accidents, they are potentially serious, and associated with a spinal cord injury, they are life-threatening. We treated 75 traumatized cervical and thoracolumbar spine patients in our spine unit at CHU Oran.

Patients and methods: - Department of orthopedic and traumatological surgery- spine unit of CHU Oran. - Period- 24 months especially during the summer season. - Secondary support after the UAS. - Middle age- 38 years (14–81 years).

Results: 75 quadriplegic patients (58 men and 17 women) aged between 14 and 81 years, 60% of patients were between 14 and 60 years old, 27% between 40 and 60 years old and 13% over 60 years old, the main The causes of these traumatic quadriplegia are mainly due to AVP (67%), falls (27%) and sports accidents (judo, gymnastics) represent 6%.

Histopathological lesions are at C1-C2 level in 6 patients (8%) and C3-C7 in 69 patients (92%). clinically 36 patients (48%) already had complete quadriplegia at admission and 39 patients (52%) had incomplete tetraplegia. The neurological involvement was classified according to the FRANKEL classification. Our practical conduct was a decompression, arthrodesis, graft with a screwed plate. 30% of the patients benefited from a conservative treatment in a reduction by cranial traction. the course was marked by recovery in 8 patients (10.6%) who had incomplete tetraplegia and 13 patients (17.3%) died as a result of their complications.

Discussion: Traumatic quadriplegia is a major public health problem, few injuries are as devastating as those affecting the spinal cord; adult, youthful and adolescent men have the highest prevalence and suffer most of the time from a permanent deficit; quickly, the quadriplegic or paraplegic becomes aware of its deficit and its consequences. Hospitalization and rehabilitation, through their costs, represent a huge investment. The emotional damage the patient and his family are not measurable. After several weeks of treatment (surgery, resuscitation, rehabilitation ...) we found adverse results and treatment failure, which further complicated their insertion and care for their families.

Conclusion: Traumatic quadriplegia is especially aggravated during transport, the goal of surgery is to decompress the marrow and stabilize the spine.

F-64 Prognostic significance of standard electroencephalography findings in adult patients with delayed awakening in the intensive care unit

Camille Legouy (speaker), Laura Girard-Stein, Lila Bouadma, Claire Dupuis, Sonia Abid, Camille Vinclair, Stéphane Ruckly, Ruben Wanono, Anny Rouvel-Tallec, Marie-Pia D’Ortho, Jean-François Timsit, Romain Sonneville

Hôpital Bichat-Claude Bernard, Paris, FRANCE
Correspondence: Camille Legouy -

Annals of Intensive Care 2019, 9(Suppl 1):F-64

Introduction: Despite daily interruption of sedative infusions, delayed awakening is frequently observed in critically ill patients requiring invasive mechanical ventilation. We aimed to identify the prognostic significance of standard electroencephalography findings in adult patients with delayed awakening in the intensive care unit.

Patients and methods: Our retrospective study included consecutive patients under invasive mechanical ventilation in the intensive care unit who underwent standard EEG because of delayed awakening. Delayed awakening was classified in 3 groups- coma, hypo-active delirium or hyperactive delirium according to RASS at inclusion. The primary endpoint was a good neurological outcome, defined as the proportion of patients alive and awake (i.e. responding to simple commands on 2 consecutive days) 7 days after EEG. Secondary endpoints included the prevalence of the different etiologies of delayed awakening, defined in 6 categories (hypoxic + metabolic + septic + antibiotic + sedation + acute brain injury) and the proportion of patients alive and awake at ICU discharge and at 90 days. Data are presented as median (interquartile range) or numbers (percentages). Cause-specific prevalence models were used to identify independent parameters associated with awakening and death, respectively.

Results: 121 patients (age 64 years [54 + 71], SAPS2 score of 61 [45 + 76]) with a RASS of -4 [-4 + -3] at inclusion were studied. At 7 days, 58 (48%) patients were awake, 40 (33%) were alive but not awake, and 23 (19%) were dead. In univariate analysis, the only parameter associated with awakening was RASS ≥ - 3, whereas parameters associated with mortality were a slow EEG background, a discontinuous EEG background and unreactive EEG background. Multivariate analysis revealed that discontinuous EEG background was associated with mortality (Table). By contrast, background frequency > 4 Hz associated with a preserved reactivity were protective. The etiologies of delayed awakening were- 65 (54%) sepsis, 49 (41%) hypoxia, 32 (27%) sedations, 25 (21%) neurotoxic antibiotics, 14 (12%) metabolic causes, with for some patients a multifactorial origin. Hypoxic encephalopathy was associated with short-term mortality. At the end of ICU stay, 60 (50%) patients were awake, 7 (6%) were alive but not awake and 53 (44%) were dead. At 90 days, 55 (45%) were awake, 6 (5%) were not awake and 60 (50%) were dead.

Conclusion: Delayed awakening in ICU is likely of multifactorial origin and characterized by a favorable outcome in about 50% of cases. Background EEG abnormalities (frequency, continuity) and reactivity provide major prognostic information on short-term mortality in this population.

F-65 Cardiac Arrest in Patients Managed for Convulsive Status Epilepticus- Characteristics, Predictors and Outcome

Stephane Legriel (speaker)1, Edouard Bresson1, Nicolas Deye2, David Grimaldi3, Bertrand Sauneuf4, Olivier Lesieur5, Jean-Baptiste Lascarrou6, Laurent Argaud7, Jonathan Chelly8, Pascal Beuret9, David Schnell10, Anne-Laure Chateauneuf1, Mathilde Holleville1, François Perier1, Virginie Lemiale11, Cédric Bruel12, Pierrick Cronier13, Nicolas Pichon14, Nicolas Mongardon15, Nicolas De Prost15, Florence Dumas16, Alain Cariou16

1Centre Hospitalier de Versailles, Le Chesnay, FRANCE; 2Centre Hospitalier Universitaire Lariboisiere, Paris, FRANCE; 3Université Libre de Bruxelles (ULB), Erasme Hospital, Bruxelles, BELGIUM; 4Cotentin Public Hospital Center, Cherbourg-En-Cotentin, FRANCE; 5Centre Hospitalier Saint-Louis de la Rochelle, La Rochelle, FRANCE; 6Centre Hospitalier Départemental Vendée, La Roche-Sur-Yon, La Roche-Sur-Yon; 7Hospices civils de Lyon, Edouard Herriot Teaching Hospital, Lyon, FRANCE; 8Centre Hospitalier Marc Jacquet, Melun, FRANCE; 9Centre Hospitalier de Roanne, Roanne, FRANCE; 10Centre Hospitalier d’Angouleme, Angoulême, FRANCE; 11Centre Hospitalier Universitaire Saint Louis, Paris, FRANCE; 12Groupe Hospitalier Paris Saint-Joseph, Paris, FRANCE; 13Centre Hospitalier Paris Sud Francilien, Corbeil-Essonnes, FRANCE; 14Centre Hospitalier Universitaire de Limoges, Limoges, FRANCE; 15Centre Hospitalier Universitaire Henri Mondor, Créteil, FRANCE; 16Centre Hospitalier Universitaire Cochin, Paris, FRANCE
Correspondence: Stephane Legriel -

Annals of Intensive Care 2019, 9(Suppl 1):F-65

Introduction: Cardiac arrest (CA) is among the most catastrophic early complication seen during convulsive status epilepticus (CSE) management. Factors that may contribute to CSE-related CA (CSE-CA) include comorbidities, severe systemic complications (particularly in the event of uncontrolled seizure activity, injuries caused by the loss of consciousness and seizure, treatment complications, and cause of CSE. Although CSE-CA is an event of considerable concern, few studies have assessed its characteristics and long-term survival and functional outcomes. The objective of this retrospective study was to identify early factors associated with CA in adults managed for CSE and admitted to the intensive care unit (ICU). Knowledge of such factors might help to identify areas for improvement in the management of CSE.

Patients and methods: Retrospective multicenter study including consecutive patients admitted to 17 university or university-affiliated ICUs in France and Belgium for management of successfully resuscitated out-of-hospital cardiac arrest complicating the initial management of CSE between 2000 and 2015. Patients were compared with controls without CA identified in a single-center registry of CSE patients, regarding characteristics, management, and outcome.

Results: We included 49 cases with CSE-CA and 235 controls. In the cases, median time from medical team arrival to CA was 25 min [IQR, 5–85]. First recorded rhythm was asystole in 25 (51%) and pulseless electrical activity in 13 (27%) patients. A significantly larger proportion of patients had a favorable 1-year outcome (Glasgow Outcome Scale score of 5) among controls (90 235, 38%) than among cases (10 49, 21%, P = 0.02). By multivariate analysis, independent predictors of CA were pulse oximetry < 97% on scene (OR, 2.66 + 95%CI, 1.03–7.26, P = 0.04), drug poisoning as the cause of CSE (OR, 4.13 + 95%CI, 1.27–13.53, P = 0.02), and complications during early management (OR, 11.98 + 95%CI, 4.67–34.69, P < 0.0001). Having at least one comorbidity among cardiac, respiratory, and neurological (other than epilepsy) conditions predicted absence of CA (OR, 0.28 + 95%CI, 0.10–0.80, P = 0.02).

Conclusion: In patients managed for CSE, relative hypoxemia, on-scene management complications, and drug poisoning as the cause of CSE were strong early predictors of CA, suggesting areas for improvement.

F-66 Feasibility and reliability of somatosensory evoked potentials performed by intensivists in the prognosis of post-cardiac arrest coma

Damien Bouvier (speaker), Quentin Levrat, Virginie Verrier, Olivier Lesieur

Hôpital Saint-Louis La Rochelle, Saint-Brice Sous Forêt, FRANCE
Correspondence: Damien Bouvier -

Annals of Intensive Care 2019, 9(Suppl 1):F-66

Introduction: Post-cardiac arrest coma is a common cause of brain injury in the ICU. Predicting neurological outcome is of crucial importance to provide the most objective information to loved ones and opt for the best therapeutic options (including withholding or withdrawal of treatments deemed hopeless). Among the prognostic tools available, somatosensory evoked potentials have proven efficiency- under certain conditions, the absence of N20 cortical wave is associated with an unfavorable neurological prognosis with a specificity closed to 100% (1). Traditionally, this test requires the expertise and availability of a neurophysiologist. We assume that this technique can also be performed by trained intensivists.

Patients and methods: Two physicians from our ICU received two days of specific training in a university neurophysiology laboratory by a specialist in the interpretation of evoked potentials. Patients concerned had prolonged post-cardiac arrest coma after cessation of sedation. The records were interpreted and sent to the neurophysiology referral center for review by a specialist within 24 h. The feasibility and reliability of tests were evaluated retrospectively.

Results: From September 2011 to June 2018, somatosensory evoked potentials were recorded in 59 patients (64 [55–74] year old + M F ratio 3.9). Circulatory arrests were of cardiac origin in 42% of cases and respiratory in 46%. All patients underwent electroencephalography and had no reactivity to stimuli. Most of the recordings were made and interpreted easily by intensivists, with a similar conclusion by the neurophysiologist. Only one was doubtful and could only be assessed by the neurophysiologist. N20 wave was present in 26 patients and absent in 31 patients. Only two tests were uninterpretable.

Conclusion: Recording somatosensory evoked potentials in the ICU is simple, reliable, reproducible and can be performed by trained critical care physicians. However, its interpretation must be validated by a neurophysiologist given the implications in terms of therapeutic decisions.

F-67 Sodium disturbances in the neuro-intensive care unit

Mariem Dlela (speaker), Manel Zekri, Rania Ammar, Aziza Talbi, Chokri Ben Hamida, Mounir Bouaziz

Hbib bourguiba university hospital, Sfax, TUNISIA
Correspondence: Mariem Dlela -

Annals of Intensive Care 2019, 9(Suppl 1):F-67

Introduction: Sodium disturbances are the most common and probably the most poorly understood electrolyte disorders in neurological diseases. Complications can be minimized by better recognition, diagnosis, and treatment of sodium disorders. In this study, we aim to analyze the incidence, etiologies and impact of dysnatremia on brain damaged population, and we hypothesize that changes in sodium levels could be indicative of recent neurological deterioration.

Patients and methods: We conducted a six month long prospective cohort, including all brain damaged patients, who were admitted to our ICU between March 1st, 2018 and August 31st, 2018 and with a minimum length of stay (LOS) of 14 days. All patients, included, were screened for sodium disorders in the first 2 weeks of ICU stay. Outcome was measured by incidence of death, Glasgow outcome scale (GOS) on discharge and LOS. Patients were also monitored for neurological deterioration, including cognitive decline, convulsive seizures, increase in cerebral edema and brain herniation that were contemporary to sodium disorders. Both univariate and multivariate analysis were used to determine level of significance.

Results: During the study period, one hundred patients were admitted to our ICU for neuro-intensive care, among which 77 were included in this study. Patients were admitted for traumatic brain injury (TBI) in 75.3% of cases. According to our analysis, 35 (45.45%) patients presented with hyponatremia, 26 (74.3%) among them, were diagnosed with the syndrome of inappropriate antidiuretic hormone secretion (SIADH), 8 (22.9%) with corticosteroid deficiency and in one case with cerebral salt wasting syndrome. SIADH was attributed to convulsive seizures in 7(26.9%) cases, meningitis in 3(11.5%) cases and TBI in 11 (42.3%) cases. Hyponatremia was found to be a predictive factor of mortality in ICU (p = 0.022), of LOS (p = 0.032) and a sign of neurological deterioration (p = 0.03) on the day of diagnosis. Our study results’ showed an incidence of hypernatremia of 26% (20 cases), among which 55% (11 cases) were attributed to central diabetes insipidus. Hypernatremia was found to be a predictive factor of mortality in ICU (p < 0.00), of GOS (p < 0.00) and a sign of neurological deterioration (p < 0.00) on the day of diagnosis.

Conclusion: In summary, this study demonstrates that sodium disturbances are common in neuro-intensive care units and associated with increased ICU mortality. Besides it indicates that changes in sodium levels could be revealing of serious neurological complications.

F-68 Characterisation of cardiovascular phenotypes in septic shock. Focus on LV systolic dysfunction and its impact on prognosis

Guillaume Geri (speaker)1, Philippe Vignon2, Alix Aubry1, Anne-Laure Fedou2, Cyril Charron1, Stein Silva3, Xavier Repesse1, Antoine Vieillard-Baron1

1University Hospital Ambroise Paré, Boulogne-Billancourt, FRANCE; 2University Hospital, Limoges, FRANCE; 3University hospital, Toulouse, FRANCE
Correspondence: Guillaume Geri -

Annals of Intensive Care 2019, 9(Suppl 1):F-68

Introduction: Left ventricular (LV) systolic dysfunction is frequent in septic shock patients, but its prognostic impact remains unknown.

Patients and methods: Two published databases from 12 different ICUs including echocardiographic monitoring performed at the initial phase of septic shock were merged. Patients with a history of chronic heart failure or atrial fibrillation or dobutamine infusion at the time of echocardiography were excluded from the analysis. Hierarchical clustering in a principal components approach was used to define five cardiovascular phenotypes using haemodynamic, clinical and echocardiographic parameters. Missing data were imputed. The relationship between cluster and mortality (day-7 and ICU) was evaluated using a multivariable logistic regression.

Results: 324 patients (median age 64 [55, 74]) were included in the analysis. Five different clusters were individualised- patients well resuscitated (cluster 1, n = 76) without LV systolic function, right ventricular (RV) failure or fluid responsiveness, patients with LV systolic dysfunction (cluster 2, n = 41), patients with hyperkinetic profile (cluster 3, n = 70), patients with RV failure (cluster 4, n = 76), and patients with persistent hypovolemia (cluster 5, n = 61). Day-7 mortality was higher in cluster 2 than in the others (37 vs. 12, 13, 24 and 20%, in clusters 1, 3, 4, and 5, respectively, p = 0.04), while ICU mortality did not differ across clusters. In multivariable logistic regression, LV systolic dysfunction was independently associated with increased day-7 mortality (odds ratio 2.80 [95% confidence interval 1.05, 7.76]).

Conclusion: Among the five cardiovascular phenotypes individualised in this large cohort of septic shock patients, LV systolic dysfunction was independently associated with day-7 mortality.

F-69 Early predictive factors of 30-days mortality in cardiogenic shock- An analysis of the FRENSHOCK multicenter prospective registry

Clément Delmas (speaker)1, Bruno Levy2, Nicolas Lamblin3, Nadia Aissaoui4, Etienne Puymirat4, Guillaume Leurent5, Vincent Labbe6, Sébastien Champion7, Stéphane Manzo-Silberman8, Meyer Elbaz1, Laurent Bonello9, Edouard Gerbaud10, Francois Roubille11, Eric Bonnefoy12, Patrick Henry8

1Rangueil University Hospital, Toulouse, FRANCE; 2Intensive Care Unit, Vandoeuvre Les Nancy, FRANCE; 3Lille University Hospital, Lille, FRANCE; 4Hôpital Européen Georges Pompidou, Paris, FRANCE; 5Rennes University Hospital, Rennes, FRANCE; 6Tenon Hospital AP-HP, Paris, Paris; 7Parly II Clinic, Le Chesnay, FRANCE; 8Lariboisière University Hospital AP-HP, Paris, FRANCE; 9Aix-Marseille University Hospital, Marseille, FRANCE; 10Bordeaux University Hospital, Pessac, FRANCE; 11Montpellier University Hospital, Montpellier, FRANCE; 12Lyon University Hospital, Lyon, FRANCE
Correspondence: Clément Delmas -

Annals of Intensive Care 2019, 9(Suppl 1):F-69

Introduction: Cardiogenic shock (CS) remains a severe but poorly understood pathology. Many predictive death scores have been previously described but have focused in ischemic CS and took into account data related to the management of these patients. So, there is an urgent need for simple and objective criteria to assess the short-term CS mortality regardless of the initial etiology.

Patients and methods: FRENSHOCK registry (NCT02703038) was a large prospective multicenter registry of CS patients admitted in intensive cardiac and general critical care units between April and October 2016 in France. Patients were prospectively included regardless of the CS etiology if they met at least one criterion of (1) low cardiac output (systolic blood pressure (SBP) < 90 mmHg and or the need of amines, or a low cardiac index < 2.2L min m2 on echocardiography or right heart catheterization + and (2) clinical, radiological, biological (NtproBNP or BNP), echocardiographical, or invasive hemodynamics overload signs + and (3) a clinical (oliguria, marbling, confusion) and or biological hypoperfusion (lactates > 2 mmol/L, hepatic and or renal failure). We studied factors related to 30d mortality using Kaplan–Meier analyses and Cox proportional hazards modeling.

Results: 772 patients were included (male 72%, median age 66yo). Non-ischemic CS were predominant (n = 491, 64%) although type 1 infarction was infrequent (n = 134, 17%). Mortality at 30-days was 26% (n = 201). Non survivors were older, had more previous renal failure, marbles, and atrial fibrillation at admission. They had lower SBP and DBP. Diagnostic tests revealed higher arterial lactate – CRP – natriuretic peptids – kaliemia + and lower pH - prothrombin time – hemoglobin – eGFR but also LVEF. Multivariate analysis retained age (especially > 75y), low systolic blood pressure (especially < 90 mmHg), high arterial lactate (especially > 4 mmol/L), low eGFR (especially < 30 ml/min/m2), low LVEF (especially < 30%) as significant predictors of 30-days mortality. Ischemic etiology or type 1 infarction were not predictive.

Conclusion: Our multicentric and prospective design confirmed the heterogeneity of CS in terms of presentation and prognosis. Five simple, practical and easy to find signs were found significant predictors of short term mortality and could be useful in providing a more accurate and stratified definition of CS’s patients in order to tailor additional therapies.

F-70 Incidence, predisposing factors and prognosis of acute postoperative Right ventricular failure in cardiac surgery- a prospective cohort study

Ahlem Trifi (speaker)1, Imen Ben Naoui2, Sami Abdellatif3, Adel Ammous2, Raouf Denguir4, Mohamed Sami Mourali5, Salah Ben Lakhal6

1Faculty of Medicine of Tunis, Tunis, TUNISIA; 2Anaesthesia and Surgical Intensive Care department, La Rabta Hospital, Tunis, TUNISIA; 3Medical intensive care unit. La Rabta Hospital, Tunis, TUNISIA; 4Cardiovascular surgery service. La Rabta Hospital, Tunis, TUNISIA; 5Department of Functional Investigations and Cardiac intensive care unit. la Rabta hopital, Tunis, TUNISIA; 6Medical intensive care unit. La Rabta Hospital, Tunis, TUNISIA
Correspondence: Ahlem Trifi -

Annals of Intensive Care 2019, 9(Suppl 1):F-70

Introduction: Acute postoperative cardiac surgery (POCS) right ventricular failure (RVF) is uncommon and worsened the patient’s prognosis. We aimed to study the incidence, risk factors and outcome of acute RVF in cardiac surgery under extracorporeal circulation (ECC) patients.

Patients and methods: a prospective cohort study over one year (December 2016-December 2017). Were included, patients candidates for cardiac surgery (CS) with extra corporeal circulation and having a normal RF systolic function. Transthoracic-echocardiography (TTE) Doppler was performed on day 1, day 3, day 7 and 1 postoperative month. TAPSE < 13 mm and an S-wave velocity < 10 cm s during the first postoperative week defined the POCS-RVF. Thus, patients were divided into two groups (POCS-RVF group versus non POCS-RVF group) and compared. Outcomes were- catecholamine support, septic events, length of stay (LOS), ventilator days and 30-day mortality.

Results: 128 among 131 patients were included (POCS-RVF group, n = 49 versus non POCS-RVF, n = 79). The incidence of acute POCS-RVF was 38.2%. Acute RVF occurred at the 1st post operative day and remained during 30 days (attached fig). Mitral valve replacement, aortic clamping time above than 90 min, preoperative arrhythmia and bleeding were significantly related to acute POCS-RVF with respectively (OR = 11.75 + IC [2.18–13.16]), (OR = 4.36 + IC [1.01–18.68]), (OR = 6.55 + IC [2.38–17.96]), (OR = 3.4 + IC [2.38–17.96]). Acute POCS-RVF increased mortality [21(43%) vs 16 (20%), p = 0.006] and reduced survival time by 5 days but no significant link was showed between POCS-RVF and death. It depended to the left ventricular (LV) systolic function. LV dysfunction in POCS-RVF patients increased the death risk by 3 and its absence improved survival. Other factors were significantly associated to mortality Bentall and coronary tube procedures and ECC time > 120 min.

Conclusion: the incidence of acute POCS-RVF is not negligible and several preoperative factors predispose to this phenomenon. LV failure worsened the outcome. These findings should sustain preventive measures to limit myocardial damage during cardiac surgery.

F-71 High risk of Chronic Kidney Disease after VA-ECMO - Results of one year follow-up of a monocentric cohort of 132 patients

Camille Vinclair (speaker)1, Romain Sonneville1, Jean Reuter1, Radj Cally1, Mathilde Neuville1, Jordane Lebut1, Claire Dupuis1, Stéphane Ruckly2, Jean-François Timsit1, Lila Bouadma1

1HUPNVS - Hôpital Bichat - Claude-Bernard, Paris, FRANCE; 2UMR 1137 - IAME Inserm. Paris-Diderot University, Paris, FRANCE
Correspondence: Camille Vinclair -

Annals of Intensive Care 2019, 9(Suppl 1):F-71

Introduction: Veno-Arterial Extra-Corporeal Membrane Oxygenation (VA-ECMO) is a life support technique associated with a major incidence of acute kidney injury (AKI). Risk of chronic kidney disease (CKD) following AKI is high. The objectives of the study were to describe renal natural history within one year following VA-ECMO and to identify early predictors of long-term renal impairment.

Patients and methods: We retrospectively analyzed consecutive adult patients without preexisting end-stage renal disease (ESRD) who received VA-ECMO for more than 48 h in the 20-bed medical ICU of a university hospital, in Paris, France, between January 2014 and December 2016. AKI severity during ICU stay was defined according to KDIGO classification. Renal function at 1 year was assessed with estimated glomerular filtration rate (eGFR) using the MDRD equation. The primary endpoint was a composite of poor renal outcome (eGFR ≤ 60 mL/min 1.73m2) or death at one year, defining a bad outcome. Factors associated with bad outcome were identified by multivariate logistic regression analysis. Quantitative variables are reported as median (interquartile range, IQR) and qualitative variables as numbers (percentage). Results of multivariate analysis are reported as odds-ratio (OR) and 95% confidence interval.

Results: 132 patients with available 1-year follow-up (male sex 75.5%, age 58 [46 + 66] years, SAPS II 55[38 + 67], SOFA 9 [6 + 12], time on ECMO 7.5 [4 + 12] days) were included in the study. 72 (54%) patients died in ICU and 80 (60.6%) within a year. 121 (92%) patients developed AKI during ICU stay, 73 (55%) required renal replacement therapy. 38 (74%) of survivors had an abnormal decline of eGFR at one year- median decline rate was 30 [18 + 55] mL/min 1.73m2. Four (3%) patients had ESRD at one year, three of them did not required RRT during ICU but experienced recurrent AKI afterwards. In multivariate analysis, a best baseline renal function was protective of a bad renal outcome at one year (OR = 0.981 for a higher eGFR, [0.966 + 0.996], p = 0.01). Severity of AKI in ICU was significantly associated with one year renal outcome. (OR = 67.190 [6.48 + 697] p < 0.01 for KDIGO stage 3).

Conclusion: Among survivors of VA-ECMO therapy, long-term renal impairment is major particularly in those with previous CKD and severe AKI during ICU stay. ESRD is rare and occurs in patient with recurrent AKI.



95% CI


Baseline eGFR





AKI KDIGO stage at canulation time


 No AKI (ref)




 Stage 1





 Stage 2





 Stage 3





Worst AKI KDIGO stage in ICU


 No AKI (ref)




 Stage 1





 Stage 2





 Stage 3





SOFA without hemodynamic and renal score





Indication of ECMO


 Medical cause





 Post-operative with CBP < 140 min





 Post-operative with CBP ≥ 140




Number of RBC transfusion





Activated Partial Thromboplastin Time





F-72 Impact of levosimendan on peripheral veno-arterial extracorporeal membrane oxygenation weaning in intensive care unit

Shamir Vally, Cyril Ferdynus, Romain Persichini, Eric Braunberger, Hugo Lo Pinto, Bruno Bouchet, Olivier Martinet , Thomas Aujoulat, Jérôme Allyn, Nicolas Allou

CHU Félix Guyon, Saint-Denis, FRANCE
Correspondence: Shamir Vally-

Annals of Intensive Care 2019, 9(Suppl 1):F-72

Introduction: Few data are available concerning the impact of levosimendan in patients with refractory cardiogenic shock supported by peripheral veno-arterial extracorporeal membrane oxygenation (VA-ECMO). The aim of this study was to evaluate impact of levosimendan on VA-ECMO weaning in patients hospitalized in intensive care unit (ICU).

Patients and methods: This retrospective cohort study was conducted in a French university hospital, in one ICU from 2010 to 2017. All patients hospitalized in ICU who underwent VA-ECMO were consecutively evaluated.

Results: A total of 150 patients with VA-ECMO were eligible for the study. A propensity score matched 38 patients in the levosimendan group and 65 patients in the non-levosimendan group. In patients treated with levosimendan, 24 h after infusion of medication left ventricular ejection fraction increased from 21.5 ± 9.1% to 30.7 ± 13.5% (P < 0.0001) while aortic velocity–time integral increased from 8.9 ± 4 cm to 12.5 ± 3.8 cm, (P = 0.002). After matching on propensity score, levosimendan was the only remaining factor associated with a significant reduction of VA-ECMO weaning failure (Hazard Ratio = 0.16, 95%confidence interval- 0.04–0.7, P = 0.01). Kaplan–Meier survival curves showed that the survival rates at 30 days were 78.4% in the levosimendan group and 49.5% in the non-levosimendan group (P = 0.0.37). However, no significant difference was found between the levosimendan and non-levosimendan groups regarding 30 days mortality after propensity score analysis (Hazard Ratio = 0.55, 95% confidence interval- 0.27–1.10, P = 0.09).

Conclusion: This study suggests that levosimendan could be associated with a beneficial effect on VA-ECMO weaning in ICU patients. However, the use of levosimendan tended to decrease 30 days mortality after propensity matched analysis (P = 0.09).

F-73 Serial neuron specific enolase (NSE) serum levels and neurologic outcome of cardiogenic shock patients treated by venoarterial extracorporeal membrane oxygenation (ECMO)

Jean Reuter (speaker)1, Katell Peoc’H2, Lila Bouadma1, Dorothée Faille1, Marie-Charlotte Bourrienne1, Claire Dupuis1, Eric Magalhaes 3, Sébastien Tanaka1, Camille Vinclair1, Etienne De Montmollin1,Nadine Ajzenberg1, Jean-François Timsit1, Romain Sonneville1

1HUPNVS - Hôpital Bichat - Claude-Bernard, Paris, FRANCE; 2HUPNVS - Hôpital Beaujon, Clichy, FRANCE; 3Centre Hospitalier Sud Francilien, Corbeil-Essonnes, FRANCE
Correspondence: Jean Reuter -

Annals of Intensive Care 2019, 9(Suppl 1):F-73

Introduction: Cardiogenic shock patients treated with venoarterial ECMO (VA-ECMO) may develop brain injury during ECMO support. We aimed to assess the predictive value of serial measurements of neuron specific enolase (NSE) to identify poor neurologic outcome and CT-defined acute brain injury in this setting.

Patients and methods: We conducted a prospective cohort study in consecutive adult patients cannulated with VA-ECMO for refractory cardiogenic shock in the medical ICU of a university hospital in Paris, France. Plasma was sampled at predefined time points, 1 day, 3 days and 7 days after VA-ECMO cannulation, until ECMO removal or death. Plasma samples were collected and stored at − 80 °C. The primary endpoint was poor outcome, a composite endpoint of CT-defined brain injury or death 28 days after VA-ECMO cannulation. The secondary endpoint was CT-defined brain injury. Plasma NSE levels were measured at the end of study. Data are presented as median (interquartile range) or number (percentages). The association between NSE levels and outcome was explored by multivariate logistic regression analysis, with NSE levels being dichotomized according to median values at day 1 and day 3.

Results: A total of 104 patients (males (n = 67, 64%)) with a SOFA score at admission of 11 (8–14) were included, of whom 26 (25%) underwent cardiopulmonary resuscitation before VA-ECMO cannulation. At VA-ECMO cannulation, all patients were mechanically ventilated, 83 (80%) were sedated, and 81 (78%) were receiving vasopressors. Plasma NSE levels were 36 (26–50) µg L at day 1, 25 (19–38) µg L at day 3 and 22 (16–31) µg L at day 7. A poor outcome occurred in 56 (53%) patients and CT-defined brain injury was observed in 16 45 (36%) patients.

Plasma NSE levels at day 1 and day 3 were associated with poor outcome in crude analyses. In multivariate analysis, only NSE levels at day 3 remained independently associated with a poor outcome (table). In patients who underwent brain CT during VA-ECMO support, both NSE levels at day 1 and day 3 were associated with CT-defined brain injury.

Conclusion: In cardiogenic shock patients treated by VA-ECMO, plasma NSE levels measured 3 days after VA-ECMO initiation are independently associated with short term acute brain injury or death, irrespective of pre-ECMO characteristics. Patients with persistent elevated NSE levels 3 days after VA-ECMO initiation may benefit from advanced neuromonitoring while on ECMO support.


Odds ratio

95% CI

Odds ratio

95% CI

(univariate analysis)

(multivariate analysis)






CPR before ECMO cannulation





SOFA at ICU admission





SOFA at ECMO initiation





Day-3 NSE > 25 mg/L





F-74 Impact of hyperoxia on patients hospitalized in intensive care unit for pulmonary congestion due to acute heart failure

Julien NaËl (speaker)1, Mathilde Ruggiu1, Clotilde Bailleul1, Sofia Ortuno1, Jean-Luc Diehl1, Damien Vimpere1, Aymeric Lancelot 1, Amélie Couteau1, Emmanuel Guerot1, Nicolas Danchin2,Etienne Puymirat2, Nadia Aissaoui3

1HEGP Critical care, Paris, FRANCE; 2HEGP Department of cardiology, Paris, FRANCE; 3HEGP- AP-HP-Université Paris Descartes, Paris, FRANCE
Correspondence: Julien NaËl -

Annals of Intensive Care 2019, 9(Suppl 1):F-74

Introduction: Oxygen therapy (OT) remains a cornerstone of acute heart failure (AHF) therapy in patients with pulmonary congestion (PC). While avoiding hypoxemia has long been a goal of critical care practitioners, less attention has been paid to the potential hazard related to excessive oxygenation and or hyperoxia. Recent studies highlighted the uselessness or the potential hazard of hyperoxia in patients admitted for acute medical emergencies. Our main objective was to evaluate the impact of an early hyperoxia exposure among critically ill patients hospitalized for AHF.

Patients and methods: In this observational, retrospective study led in a Parisian medical intensive care unit (ICU), we assessed AHF patients admitted for PC from 01 01 2015 to 12 31 2016. Patients with cardiac arrest, severe chronic obstructive pulmonary disease, and long-term OT were not included. Hyperoxia was defined as a PaO2 > 100 mmHg on blood gaz analysis. The hyperoxia group was defined by having at least one PaO2 > 100 mmHg the first day following the ICU admission. The principal endpoint was a 30-day composite one combining all-cause mortality and unplanned hospital admission. The secondary endpoints were occurrence of a pneumonia bacteriemia, ICU length of stay (LOS) and hospital LOS. Multivariate analysis was performed to determine if hyperoxia was independent risk factor of 30-day mortality.

Results: Among the 1541 patients admitted in ICU during the period study, 75 patients with PC due to AHF were included. Forty-one patients (54.7%) required mechanical ventilation. During the first 24 h, 43 patients (57.3%) presented at least one hyperoxia on ABG [the hyperoxia group (H)] whereas 32 patients (42.7%) did not [the control group (C)]. The baseline characteristics according to the two groups did not differ [Table 1]. The composite primary endpoint did not differ between the two groups (27.9% vs 21.8%, P = 0.85). 30-day mortality was 14% in H versus 12.5% in C, P = 0.85. 30-day unplanned hospital admission was increased in H (16.3%) compared to C but it did not reach the significance (P = 0.21). The secondary endpoints were not significantly different between the two groups [Table 1]. In multivariate analysis, hyperoxia was not associated with 30-day mortality [OR = 0.44(95%CI- 0.14–1.40), P = 0.44].

Conclusion: Hyperoxia is not useful in critically ill patients with AHF but its benefit on the short-term outcome remains to demonstrate.

Table 1. Baselines characteristics and Outcomes


Total patients

(n = 75)

Hyperoxia group

n = 43

Control group

n = 32


Age, (years)

76 [68–83]

74 [64–85]

76.5 [65–88]


Male sex

38 (51)


13 (41)


Supraventricular arythmia

42 (56)

20 (46.5)

22 (69)


Chronic lung disease

14 (19)

3 (7.0)

11 (34)

< 0.001

Chronic kidney failure

23 (31)

11 (26)

12 (37.5)


Ischemic cardiopathy

33 (44)

23 (53.5)

10 (31)


Ejection Fraction, (%)

50 [47–54]

49 [44–53]

49 [44–53]


SOFA Score

5 [4–6]

4 [3–7]

5 [3–7]


Mechanic ventilation


- Mechanical Ventilation

40 (53.3)

23 (53)

17 (53)


- Non-invasive ventilation

28 (37.3)


15 (47)


- Invasive ventilation

12 (16)

10 (23)

2 (6)


Duration of mechanical ventilation, (days)

2 [1–3]

2 [1–3]

2 [0.6–3.4]




- Inotropic agents

11 (15)

9 (21)

2 (6)

< 0.01

- Diuretics

59 (79)

35 (81)

24 (75)


- Vasodilators

26 (35)

16 (37)

10 (31)



7 (9)

3 (7)

4 (12.5)


Duration of inotrope infusion, (days)


0.6 [0.2–1]

0.6 [0.2–1.4]


Duration of dialysis, (days)





Infection occurence

12 (16)

6 (14)

6 (19)


LOS in ICU, (days)

4.0 [2.4–7.3]

3.3 [2.0–6.9]

4.5 [3.4–7.8]


LOS in hospital, (days)

12.2 [7.5–17.5]

12.2 [7.5–16.3]

12.1 [8.3–17.6]


30-day mortality

10 (13.3)

6 (14)

4 (12.5)


30-day unplanned hospital admission

10 (13.3)

7 (16.3)

3 (9.3)


30-day death and/or unplanned hospital admission

20 (26.7)

13 (27.9)

7 (21.8)


Abbreviations: AF = Atrial fibrillation, SOFA = Sequential Organ Failure Assessment , LOS = length of stay

Values are median [Q1–Q3] or n (%)

F-75 Does having multiple causes of hypoxemia increase hypoxemia severity ? An ancillary study of the SPECTRUM study

David Grimaldi (speaker)1, Sami Hraiech2, Florence Boissier3, Philippe Michel4, Jean-Claude Lacherade5, Tai Pham6, Jean-Christophe Richard 7, Arnaud Thille3, Stephan Ehrmann8, Nadia Aissaoui9,Gregoire Muller10

1CHUB Hôpital Erasme, Bruxelles, BELGIUM; 2Hôpital Nord AP-HM, Marseille, FRANCE; 3CHU, Poitiers, FRANCE; 4CH, Pontoise, FRANCE; 5CH, La Roche Sur Yon, FRANCE; 6St Michael’s hospital, Toronto, CANADA; 7Hôpital de la Croix Rousse, Lyon, FRANCE; 8CHU, Tours, FRANCE; 9HEGP-APHP, Paris, FRANCE; 10CH, Orléans, FRANCE
Correspondence: David Grimaldi -

Annals of Intensive Care 2019, 9(Suppl 1):F-75

Introduction: In ICU patients several causes mechanisms of hypoxemia can be present. However the cumulative effect of multiple causes of hypoxemia is unknown. Using the data from a point-prevalence-day study on hypoxemia, we analyzed the relationship between the number of hypoxemia causes and hypoxemia severity.

Patients and methods: The SPECTRUM study was conducted in 117 ICUs in 7 countries during spring 2016. We collected the presence of 23 causes mechanisms of hypoxemia based on investigators judgment. We compared hypoxemia causes according to the class of hypoxemia. We classified the patients in 4 groups according to the number of causes (0, 1, 2 and > 2) and analyzed if the number of conditions was associated with hypoxemia severity and ICU-mortality using univariate followed by multivariate analyses.

Results: Among the 859 hypoxemic patients included the day of the study, 853 could be analyzed. The median number of hypoxemia causes was 2 (1–3). 107 (12%) patients had no cause of hypoxemia recorded, 230 (27%) had 1 cause, 221 (26%) had 2 causes and 295 (35%) had 3 or more causes. Main causes were pneumonia (53%), fluid overload (33%), pleural effusion (23%), atelectasis (21%), acute on chronic respiratory failure (19%). Repartition of causes differ across the 3 class of hypoxemia severity.

Patients with more hypoxemia causes had more often chronic respiratory, heart and renal failure, had a higher SAPS-2 and were more often under vasopressors, (p < 0.05 for all). Patients with more causes of hypoxemia were more invasively ventilated (p < 0.001) and the number of causes was associated with hypoxemia severity (p < 0.001). After multivariate linear regression, number of causes of hypoxemia was negatively associated with P/F (coeff -3 (95%CI -6 + 0), p = 0.05). ICU-mortality across the 4 groups was 16, 21, 27 and 35% (p < 0.001). Multivariate logistic regression showed that in addition to age, chronic heart failure, SAPS-2, admission diagnosis, P/F < 100, and ARDS, having > 2 causes of hypoxemia was associated with ICU-mortality (OR 2 (95%CI- 1.03–3.8)), whereas obesity was protective.

Conclusion: Over 60% of hypoxemic patients had 2 or more causes of hypoxemia. The number of causes of hypoxemia was independently associated with a lower P/F ratio. Having 3 or more causes of hypoxemia was independently associated with ICU-mortality. Preventing patients to acquire supplementar conditions that induce hypoxemia may improve their prognosis.

F-76 Recruitment maneuver use in ARDS patients and mortality, a systematical review and meta-analysis

Joris Pensier (speaker), Audrey De Jong, Gérald Chanques, Nicolas Molinari, Samir Jaber

Département d’Anesthésie-Réanimation B - CHRU St Éloi, Montpellier, France
Correspondence: Joris Pensier -

Annals of Intensive Care 2019, 9(Suppl 1):F-76

Introduction: Acute respiratory distress syndrome (ARDS) is a common injury in intensive care units (ICU) patients, which is worsened by atelectasis. Lung recruitment maneuver (LRM) aims to open collapsed alveoli, increasing airway pressure for a short duration. Single studies of LRM’s use for ARDS patients in ICU, along with recent meta-analyses, have shown discrepancies regarding prognosis of ARDS patients. Two recent RCTs have been published, entailing around 1300 new patients. The aim of this study was to critically review the literature to investigate whether LRM reduces mortality in ARDS patients compared to a no-LRM strategy.

Patients and methods: We performed a systematic review and meta-analysis of randomized controlled trials (RCTs), by searching PubMed, CENTRAL, Web of Science and bibliographies of articles retrieved, among data from 1998 to 2018. We screened for relevant studies that enrolled ARDS adult patients. We included studies reporting mortality to perform a meta-analysis. We generated pooled relative risks (RR) across studies. The primary outcome was 28-day mortality. The secondary outcomes were ICU mortality and 60-day mortality.

Results: Eight RCTs with a total of 2,735 participants (1297 in LRM group and 1438 in No-LRM group) were included in the analysis (table 1). Compared to a No-LRM strategy, LRM did not reduce the 28-day mortality (RR = 0.90 (95% Confidence Interval (CI) 0.75–1.08, p = 0.26)). A significant heterogeneity was observed (p < 0.03, I2 = 52%). A sensitivity analysis excluding a study which did not use protective ventilation in control group showed similar results (RR = 0.96 (CI95% 0.83–1.12, p = 0.62)). No significant difference was found on mortality in the ICU (6 studies, 2412 patients, RR = 0.87 (CI95% 0.71–1.06, p = 0.16)). A significant heterogeneity was observed (p = 0.02, I2 = 63%). A sensitivity analysis excluding a study which did not use protective ventilation in control group showed similar results (5 studies, 2359 patients, RR = 0.93 CI95% 0.78–1.10, p = 0.40). No significant difference was found on mortality at day 60 (3 studies, 559 patients, RR = 0.92 (CI95% 0.71–1.19, p = 0.51)). No heterogeneity was found for this outcome (p = 0.84, I2 = 0%).

Conclusion: These results suggest that systematical LRM does not provide a significant benefit regarding mortality in ARDS. However, heterogeneity between studies was high and remains to be further explored.

F-77 Assessment of electrical impedance tomography to set optimal positive end-expiratory pressure for veno-venous ECMO-treated severe ARDS patients

Floriane Puel (speaker), Christelle Soulé, Fanny Bounes Vardon, Thierry Seguin, Vincent Minville, Bernard Georges, Jean-Marie Conil, Laure Crognier

CHU Rangueil, Toulouse, FRANCE
Correspondence: Floriane Puel -

Annals of Intensive Care 2019, 9(Suppl 1):F-77

Introduction: In patients with severe acute respiratory distress syndrome (ARDS) on extracorporeal membrane oxygenation (ECMO) and ventilated with very low tidal volume, the optimal positive end-expiratory pressure (PEEP) should combine the best alveolar recruitment with minimal overdistension. Electrical impedance tomography (EIT) provides a non-invasive, real-time, bedside imaging of the lungs. The aim of our study was to assess EIT’s ability to choose the best PEEP in patients on ECMO receiving ultra-protective mechanical ventilation.

Patients and methods: ARDS patients on veno-venous ecmo were included. A written information was given to all patients’ relatives before inclusion. A recruitment maneuver and a decremental PEEP trial from 20 to 5 cmH20 (in 5 cmH20 steps) were monitored by EIT, with lung images divided into four ventral-to-dorsal horizontal regions of interest. Driving pressure was maintained constant at 10cmH2O. For each patient, four EIT-based PEEP were defined- PEEP ODCLmin (pressure with the lowest EIT-assessed collapse lung [CL] and overdistension [OD]), PEEP ODCL15 (the lowest pressure able to limit EIT-assessed collapse below 15% with the least overdistension), PEEP Comp (PEEP with the EIT-based compliance maximum), and PEEP GI (PEEP with EIT-based global inhomogeneity (GI) index minimum). The concordance between these four EIT-based PEEP and the reference pulmonary PEEP P (defined according the respiratory clinical and ultrasound usual parameters) was evaluated by the Cohen’s kappa coefficient.

Results: In fourteen patients, the decremental PEEP trial induced a decrease in tidal impedance variation (TIV) of dependent regions, and an increase in TIV of non-dependent lung regions (Figure 1). High PEEP levels were significantly associated with more overdistension (Rho = 0.908 [0.848–0.945]) and fewer collapsed zones, while decreasing PEEP led to more collapsed zones (Rho = − 0.909 [− 0.946 to − 0.849]). The PEEP ODCL15 and the PEEP Comp were in successful agreement with the reference PEEP P (respectively 0.714 [0.363–1] and 0.714 [0.348–1]). The PEEP ODCLmin was in medium agreement with PEEP P (0.571 [0.146–0.997]), and PEEP GI was in disagreement (− 0.0833 [− 0.218–0.0517]).

Discussion: EIT allows to individually study the effects of PEEP levels during ultra-protective ventilation on ECMO with EIT-based overdistension, collapse, and compliance.

Conclusion: EIT may be an interesting non-invasive bedside tool to provide real-time monitoring of PEEP impact in severe ARDS patients under ECMO.

No conflict of interest.

F-78 Prone position is feasible with an extracorporeal decarboxylation during adult’s acute respiratory distress syndrome. Analysis of 58 sessions

Nicolas Boquillon (speaker), Jean François Georger

CHI Villeneuve Saint Georges, Villeneuve-Saint-Georges, FRANCE
Correspondence: Nicolas Boquillon -

Annals of Intensive Care 2019, 9(Suppl 1):F-78

Introduction: Toxicity of mechanical ventilation during acute respiratory distress syndrome (ARDS) can possibly be prevented by using “ultra-protective” ventilation which favors hypercapnia. An extracorporeal removal of CO2 (ECCO2R) can correct it. The issue of hypoxemia remains for which prone positioning (PP) is known to be effective. We therefore studied PP feasibility with ECCO2R.

Patients and methods: We conducted a retrospective analysis from August 2014 to December 2017 in our center. Indication for ECCO2R was a pH < 7.20 with an arterial partial pressure of CO2 > 50 mmHg with “protective ventilation”. Indication for PP was a PaO2 FiO2 ratio (arterial partial pressure of O2 fraction of inspired O2) < 150. Primary endpoint was discontinuation before 15 h of PP. Patients with ARDS (Berlin’s criteria) and ECCO2R were compared to patients without ECCO2R with Fischer’s exact test.

Results: 23 patients with ECCO2R underwent 58 PP procedures whose 56 lasted more than 15 h. 38 patients without ECCO2R underwent 53 PP procedures whose 47 lasted more than 15 h. There was no difference for early discontinuation of PP (p = 0.15).

Discussion: Our results were in accordance with scientific evidence and did not advocate for a change of practice.

Conclusion: It was possible to safely conduct 58 PP sessions with ECCO2R and “ultra-protective” ventilation during ARDS.

F-79 Ultra-protective ventilation reduces biotrauma in patients on veno-venous extracorporeal membrane oxygenation for severe acute respiratory distress syndrome

Sacha Rozencwajg (speaker)1, Amélie Guihot2, Guillaume Franchineau1, Mickael Lescroat1, Nicolas Bréchot1, Guillaume Hekimian1, Guillaume Lebreton1, Pascal Leprince1, Brigitte Autran2, Charles-Edouard Luyt1,Alain Combes1, Matthieu Schmidt1

1Sorbonne Universités, UPMC Univ Paris 06, INSERM UMRS_1166-iCAN, Institute of Cardiometabolism and Nutrition, Paris, FRANCE; 2Assistance Publique–Hôpitaux de Paris, Pitié–Salpêtrière Hospital, Immunology department, 75651 Paris Cedex 13, France, Paris, FRANCE
Correspondence: Sacha Rozencwajg -

Annals of Intensive Care 2019, 9(Suppl 1):F-79

Introduction: Ventilator settings in patients with severe Acute Respiratory Distress Syndrome (ARDS) supported by Veno-Venous Extra-Corporeal Membrane Oxygenation (VV-ECMO) are currently set arbitrarily. Our study aimed to evaluate the impact on serum and pulmonary markers of biotrauma of a) the transition to ultra-protective ventilation settings following the initiation of VV-ECMO, and b) different mechanical ventilation strategies while on VV-ECMO.

Patients and methods: Monocentric randomized clinical trial conducted during a 6-month period in patients receiving VV-ECMO for refractory severe ARDS. Once VV-ECMO started, patients were switched to the APRV mode with one second of 24 cmH2O high pressure (Phigh) and 2 s of 12 cmH2O low pressure (Plow) for 24 h. Patients were then randomized and assigned to each of the following three experimental steps according to a computer-generated allocation sequence- Phigh 24 cmH2O and Plow 20 cmH2O (“very high PEEP—very low driving pressure [VHPEEP-VLDP] step”) + Phigh 24 cmH2O and Plow 5 cmH2O (“low PEEP—high driving pressure [LPEEP-HDP] step”) + and Phigh 17 cmH2O and Plow 5 cmH2O (“low PEEP –low driving pressure [LPEEP-LDP] step”). Plasma and bronchoalveolar soluble receptor for advanced glycation end-products (sRAGE), and serum cytokines were dosed before VV-ECMO and after 12 h at each mechanical ventilation strategies.

Results: Sixteen patients on VV-ECMO after 7 (1–11) days on mechanical ventilation were included. “Ultra-protective” mechanical ventilation settings following ECMO initation was associated with a significant reduction in both plasma sRAGE and plasma cytokines. However, plasma sRAGE and cytokines were similar within each experimental steps during ECMO. Nevertheless, broncho-alveolar levels of sRAGE were at their lowest when driving pressure was minimal.

Conclusion: VV-ECMO allows an “ultra-protective ventilation”, which combines a significant reduction of tidal volume and driving pressure. This ventilation strategy significantly reduced pulmonary biotrauma, which could therefore decrease ventilator induced lung injury. However, the optimal “ultra-protective ventilation” strategy once VV-ECMO is initiated remains undetermined and warrants further investigations.

F-80 Effect of Prone Position during Acute Respiratory Failure Apart From ARDS- A Prospective Study

Sabrine Nakaa (speaker), Nejla Tilouche, Oussama Jaoued, Habiba Ben Sik Ali, Rim Gharbi, Mohamed Fekih Hassen, Souheil Elatrous

Hopital taher sfar mahdia, Mahdia, TUNISIA
Correspondence: Sabrine Nakaa -

Annals of Intensive Care 2019, 9(Suppl 1):F-80

Introduction: Prone position (PP) has been shown to be a useful adjunct in the treatment of patients with ARDS. But there is a lack of conclusive studies concerning the efficacy of PP in acute respiratory failure (ARF) without ARDS criteria. Objective- to evaluate the effect of prone positioning on oxygenation and survival among patients receiving invasive mechanical ventilation (MV) and having severe persistent hypoxemia due to another reason than ARDS.

Patients and methods: This prospective study was conducted in a 10- bed ICU between February 2015 and October 2017. Inclusion criteria were- patients requiring MV and who have a refractory hypoxemia. Exclusion criteria were- ARDS according to Berlin criteria and or ventilator associated pneumonia, contraindication to PP and a decision to withdraw or withhold therapy. Patients were placed in PP for at least 8 h. An improvement in oxygenation was defined as an increment of 20 mmHg or more in the PaO2 FiO2 after prone positioning. Outcome measures were improvement in oxygenation, ICU mortality, ventilator free days and ICU length of stay.

Results: During the study MV was used in 500 patients. We excluded 470 patients (70 patients didn’t met ARDS criteria and 400 patients were not hypoxemic). Thirty patients with a mean age of 50 ± 17 years were included. ARF was the most frequent reason of admission (36.7%). Twenty five patients (83.4%) were responders to PP. Indeed, PaO2 FiO2 increased significantly after PP (from 107 ± 41 to 185 ± 66 mmHg, p < 10-3) and the PaCO2 has decreased significantly (from 56 ± 17 to 48 ± 12, p < 10−3) compared to the supine position. Twelve patients (40%) presented one or more complications related to the PP. The most common complication was decubitus ulcers in 27% of cases. The comparaison between responders and non responders to PP shows no differences in demographic characteristics, co morbidities and PaO2 FIO2. The ICU mortality rate in PP responders was 32% compared to 20% in non responders (p = 1.00). The post hoc analysis in the sub groups of patients according to the degree of hypoxemia before PP (PaO2 FIO2 before PP <=100 and PaO2 FiO2 before PP > 100) has shown that the ICU mortality, ICU LOS and Ventilator-free days were similar in both groups.

Conclusion: In the present study prone positioning applied to severely hypoxemic patients without ARDS criteria was associated with an improvement in oxygenation and in carbon dioxide washing.

F-81 Incidence of and risk factors for thrombotic complications following venovenous extracorporeal membrane oxygenation- a CT-scan study

Gabriel Parzy (speaker)

APHM, Marseille, FRANCE
Correspondence: Gabriel Parzy -

Annals of Intensive Care 2019, 9(Suppl 1):F-81

Introduction: The aims of this study were to 1) analyze Cannula-associated deep vein thrombosis (CaDVT) incidence after veno-venous extracoporeal membrane oxygenation (VV-ECMO) using a CT-scan and 2) identify the associated risk factors of CaDVT.

Patients and methods: This retrospective observational analysis was conducted in a tertiary referral university teaching hospital. Patients under VV-ECMO with a femoro-femoral or femoro-jugular cannulation admitted for ARDS or primary dysfunction graft after pulmonary transplantation were included. Diagnosis of CaDVT was performed with a Iodinized CT-scan within 4 days after decannulation.

Results: 228 patients were screened, 105 were included. Bacterial pneumonia was the main indication of VV-ECMO (46.7%). A CaDVT was found in 75 patients (71.4%) despite a mean APTT at 1.60 ± 0.31. Specifically for femoral CaDVT, femoro-femoral cannulation induced more CaDVT than femoro-jugular cannulation (69.2% vs 63.1% respectively, p = 0.04). The mean number of patients requiring ECMO circuit replacement was significantly higher in CaDVT group (38.7% vs 16.7%, p = 0.04). Multivariate logistic regression analysis showed that thrombocytopenia < 100 G L was significant for a decreased risk to develop a CaDVT (HR 0.98 + CI95% [0.98–1.00], p = 0.02).

Conclusion: Cannula associated deep vein thrombosis after VV-ECMO is a frequent event (71.4%). This suggest that a systematic vascular axes imaging is necessary after VV-ECMO. Thrombocytopenia is associated with reduced thrombotic events.

F-82 Impact of the length of stay in emergency department on the mortality in intensive care unit of septic shock patients- a cohort study

Noémie Teixera (speaker)1, Geoffroy Rousseau2, Emeline Laurent3, Saïd Laribi4

1CHRU Tours, Cinq Mars La Pile, FRANCE; 2CHRU Tours - Hopital Trousseau, Tours, FRANCE; 3CHRU Tours-Hopital Bretonneau, Tours, FRANCE; 4CHRU Tours, Tours, FRANCE
Correspondence: Noémie Teixera -

Annals of Intensive Care 2019, 9(Suppl 1):F-82

Introduction: The overcrowding of emergency departments (EDs) increases. That affects the management of our patients, especially critically ill patients. Septic shock is a life-threatening disease with a mortality of 40%. The management of these patients and their transfer to intensive care units (ICUs) must be a priority. Did the length-of-stay (LoS) of septic shock patients in ED increase their mortality in ICU?

Patients and methods: We performed a monocentric retrospective cohort study. Patients hospitalized in ICU for septic shock between 2012 and 2017 and admitted via our ED were included. We excluded “Do not resuscitate” patients and patients admitted in ICU from another ED. We compared the LoS in ED between “alive group” and “dead group” in ICU, as IGS II score, age, delay of antibiotics, need of mechanical ventilation, use of vasopressive drugs, fluid management and qSOFA. Then, we realized a multivariate analysis.

Results: 115 patients were included in ICU for septic shock. The median LoS in ED was 317 [245–496] minutes. The mortality rate in ICU was 25.2%. In “alive group”, the LoS in ED was 343 [244–565] minutes vs 295 [240–367] minutes in “dead group”, there was no significant difference. In the univariate analysis, qSOFA, lactates, IGS II score and the use of mechanical ventilation in ED were significant. In multivariate analysis, IGS II score and the use of mechanical ventilation in ED were significantly associated with the mortality in ICU.

Discussion: The LoS in ED was not associated with the mortality in ICU in our study. “Dead group” was more often associated with high comorbidities and altered qSOFA at admission in ED. The use of mechanical ventilation in ED was associated with poor prognosis, whatever the origin of septic shock. It is important to confirm these results with a large prospective study.

Conclusion: Despite the overcrowding of EDs, emergency physicians and paramedics seemed to remain vigilant to detect, manage and transfer quickly septic shock patients.

F-83 Fluid balance impact in sepsis and septic shock

Khaoula Ben Ismail (speaker), Ines Fathallah, Ghada Sboui, Sahar Hbecha, Ameni Sghier, Asma Mehdi, Haifa Fazeni, Nadia Kouraichi

Hôpital Ben Arous Yassminet, Hammam Plage | Ben Arous, TUNISIA
Correspondence: Khaoula Ben Ismail -

Annals of Intensive Care 2019, 9(Suppl 1):F-83

Introduction: Aggressive fluid resuscitation is the initial approach for cardiovascular instability. Consequently, large volumes of fluid are given to septic patients during their management. Our study aimed to investigate the impact of cumulative fluid balance on critically ill patients with sepsis or septic shock admitted in intensive care unit (ICU).

Patients and methods: Retrospective monocentric study conducted in a medical ICU from September 2017 to September 2018. Patients with septic shock who required dialysis prior to hospitalization were not included.We included patients who presented sepsis or septic shock during their hospitalization and we studied the relation between fluid balance and prognosis.

Results: We enrolled 57 patients with an average age of 58 ± 16 years. The main reasons of hospitalization were respiratory failure and neurological disturbances (70% of cases). Median APACHE II and SOFA score at admission were respectively 15 [10 + 21] and 5 [3 + 8]. Sepsis and septic shock were related to pulmonary infections in 65% of cases, followed by urinary tract infections (14%) and catheter related infections (7%). The most frequent infectious agents were Gram negative bacilli (30%). Mechanical ventilation was required in 39 patients. Vasoactive drugs were used in 34 patients. Median duration of mechanical ventilation and hospital stay were respectively 12 [6 + 37] and 14 [5 + 24] days. The overall mortality was 40%. Multivariate analysis reveled that mortality was associated with positive cumulated fluid balance before and on the sepsis onset with relative risks respectively at 1.55(IC 95% [1.11 + 2.18], p = 0.013) and 3.32 (IC 95% [1.74 + 6.32], p < 0.001).

Conclusion: Positive fluid balance is associated to poor prognosis in septic patients.

F-84 Incidence and risk factors of venous thromboembolism in patients with septic shock

Sabrine Bradai (speaker), Manel Zekri, Abir Bouattour, Amal Triki, OlfaTurki, Mabrouk Bahloul, Mounir Bouaziz

Department of Intensive Care, Habib Bourguiba University Hospital, Sfax, TUNISIA
Correspondence: Sabrine Bradai -

Annals of Intensive Care 2019, 9(Suppl 1):F-84

Introduction: Venous thromboembolism (VTE) is a common and preventable complication among hospitalized patients in intensive care unit (ICU). However, little is known about it incidence and its particularities in patients with septic shock.

Patients and methods: It is a prospective analytical study, conducted at the ICU of Habib Bourguiba university hospital, Sfax, Tunisia, between January 01, 2017, and December 31, 2017. All Patients who developed septic shock due to bacterial infection during the study period were enrolled. Thrombo-prophylaxis was recorded for all patients. The diagnosis of TVE is confirmed by spiral computed tomography scan and compression venous ultrasound.

Results: Sixty patients with septic shock were included in the study. Among them, 24 developed VTE during their hospitalisation. The incidence of VTE was 40%. VTE was associated with increased length of stay, longer mechanical ventilation and more use of tracheotomy. The mortality rate was not higher in patients with acute VTE. (Table I).

Conclusion: Patients who develop septic shock are considered at higher risk for developing VTE. It is the result of multiple factors including immobility, activation of thrombo-inflammatory pathways, disseminated intravascular coagulation, venous stasis and central venous catheter insertion. Thus, more effective VTE prevention strategies are necessary in patients with sepsis.

Table I: Comparison between the two groups (with and without VTE)


VTE GROUP (N = 24)



Age (years)

51.8 ± 16.4

45.1 ± 20.3


Sex ratio (M/F)





35.3 ± 14.5

44.3 ± 16.2



5.5 ± 1.3

8 ± 4.1



10.6 ± 4

9.5 ± 4.3


Type of admission:



 - Traumatism




 - Medical




 - Surgical




Duration of mechanical ventilation (days)

32.7 ± 11.4

12.9 ± 9.3










Acute kidney failure








Length of stay (days)

37.3 ± 11.6

16.3 ± 11.7


Mortality rate





7.39 ± 0.09

7.37 ± 0.10


PACO2 (mmHg)

38.13 ± 8.38

37.45 ± 9.66


PAO2/FiO2 ratio

270 ± 112.8

267.4 ± 109.8


HCO3 (mmol/l)

22.47 ± 4.41

21.88 ± 6.10


Troponin (ngl/l)

0.079 ± 0.086

0.57 ± 1.03



71.4 ± 59.3

91.5 ± 142.6



57.6 ± 65.8

54.3 ± 55.5


Bilirubin (μmol/l)

45.9 ± 51.1

56.4 ± 118.6


Blood urea (mmol/l)

12.4 ± 7.5

17 ± 16


Blood creatinine (μmol/l)

109 ± 72

157 ± 126


CRP (mg/L)

231.9 ± 103

230 ± 138


Procalcitonin (ng/mL)

14.6 ± 20.9

11.3 ± 15.5



3391 ± 1430

3423 ± 1151


VTE: Venous thromboembolism; GCS Glasgow coma scale score; SAPSII: Simplified acute physiology score; DIC: Disseminated intra-vascular coagulation; SOFA score: Sepsis-related Organ Failure Assessment score; SGOT: Sérum Glutamooxaloacétate Transférase; SGPT: Sérum Glutamopyruvate Transférase; SChEA: Serum Cholinesterase Activity

F-85 Muscle Lactate and lactate to pyruvate ratio clearance as useful biomarkers for the prediction of mortality in septic shock patients

Zied Hajjej (speaker), Mayssa Daiki, Walid Sellami, Hedi Gharsallah, Iheb Labbene, Mustapha Fejani

Military Hospital, Tunis, TUNISIA
Correspondence: Zied Hajjej -

Annals of Intensive Care 2019, 9(Suppl 1):F-85

Introduction: Microcirculatory alterations are frequently observed in patients with sepsis. In vivo microdialysis (MD) is a bedside technique that can monitor tissue metabolic changes. We conducted this study aiming to assess the performance of muscle Lactate and lactate to pyruvate (L P) ratio clearance in predicting mortality in septic shock patients by using microdialysis.

Patients and methods: The study was designed as a prospective, controlled, clinical trial and performed in a multidisciplinary intensive care unit. 56 septic shock patients were enrolled. Interstitial tissue concentrations of lactate, pyruvate, glucose and glycerol were obtained at baseline and every 6 h for 3 days by using muscle microdialysis. Clearances of muscle lactate, and L P ratio were defined as the percentage change in muscle lactate level or L P ratio compared to baseline (H0) values. A positive value of clearance means a decrease in the rate of the parameter under study.

Results: We found an association between muscle lactate clearance and hospital mortality with a statistically significant difference at H54 (p = 0.037), H60 (p = 0.033) and H72 in the study (p = 0.012). We also found an association between clearance of muscle lactate to pyruvate ratio and hospital mortality with a statistically significant difference at H54 (p = 0.015), H60 (p = 0.001), and H72 of the study (p = 0, 04).

Conclusion: Among patients with septic shock, improvement of muscle Lactate and lactate to pyruvate ratio clearance from the 54th hour may indicate a resolution of global tissue hypoxia and is associated with decreased mortality rate.

F-86 Performance of the Sepsis- definitions in a Tunisian intensive care unit

Zied Hajjej (speaker), Ben Mahmoud Khalthoum, Walid Samoud, Olfa Yengui, Mayssa Daiki, Mustapha Ferjani

Military Hospital, Tunis, TUNISIA
Correspondence: Zied Hajjej -

Annals of Intensive Care 2019, 9(Suppl 1):F-86

Introduction: Since the first publication, in 2016, Sepsis-3 definitions are not universally accepted and are becoming a matter of controversy. Because clinical and laboratory parameters used for the development of these definitions were derived mainly from patients hospitalized in United States Intensive Care Units (ICU). The aim of this study was to evaluate the performance of the Sepsis 3 definitions for prediction of ICU-mortality in a Tunisian ICU population as compared to 1992 Consensus Definitions (Sepsis-2 definitions).

Patients and methods: It was a retrospective descriptive study performed in an 18-bed medical surgical intensive care unit at Tunis military hospital (Tunisia). From January 2012 to January 2016, all patients admitted to the ICU were eligible for this study. Inclusion criteria were as follows- age > 18 years, and an admission diagnosis of sepsis, severe sepsis or septic shock as defined according to the Surviving Sepsis Campaign guidelines (Sepsis-2 consensus). The new Sepsis-3 definition was secondary used. The primary outcome of interest was ICU mortality defined as death before ICU discharge.

Results: Of 3246 participants enrolled between 2012 and 2016, we included 1080 individuals with follow-up information available. When the Sepsis-2 definitions were used there was a difference in mortality only between septic shock and sepsis patients. While Sepsis-3 definitions show that mortality increased from 16% in no-dysfunction-infected patients to 30% in patients with qSOFA ≥ 2 and 44% or 46% for sepsis or septic shock patients, respectively.

Conclusion: sepsis-3 was better than sepsis-2 definitions at stratifying mortality among septic patients admitted to an ICU of an emerging country (Tunisia).

F-87 Prognostic factors and impact of hyperbaric oxygen therapy for the management of necrotizing soft tissue infections in critically ill patients

Cécile Bouges (