Skip to main content
  • Meeting abstracts
  • Open access
  • Published:

French Intensive Care Society, International congress – Réanimation 2016


O1 Impact of tracheal cuff shape on microaspiration of gastric contents in intubated critically ill patients: a multicenter randomized controlled study (BEST CUFF)

Correspondence: Emmanuelle Jaillette -

Annals of Intensive Care 2016, 6(Suppl 1):O1

Introduction Ventilator-associated pneumonia (VAP) is the most common ICU-acquired infection in intubated critically ill patients. Microaspiration of gastric and oropharyngeal contaminated secretions represents the primary mechanism involved in the pathogenesis of VAP. Tracheal cuff plays an important role in stopping the progression of contaminated secretions into the lower respiratory tract. In vitro and animal studies suggested that leakage was significantly reduced with polyvinyl chloride (PVC) conical-cuffed tubes compared with barrel (standard) or cylindrical cuffs. Clinical studies found conflicting results. Therefore, the aim of this study is to determine the superiority of PVC conical- versus barrel (standard)-cuffed tracheal tube on abundant microaspiration of gastric contents in intubated critically ill patients.

Materials and methods BEST CUFF is a prospective multicenter (ten French ICUs) cluster randomized controlled crossover and open-label trial performed in patients with predicted duration of mechanical ventilation ≥48 h. Patients were allocated to be intubated using a PVC standard (barrel)-shaped or a PVC conical-shaped tracheal tube. Prevention measures of VAP were standardized in all ICUs. The main objective was to demonstrate the superiority of conical versus standard cuff shape in reducing abundant microaspiration of gastric contents (pepsin level >200 ng/ml in at least 30 % of tracheal aspirates). After inclusion, tracheal aspirates were collected for 48 h to measure pepsin and salivary amylase, and diagnose gastric and oropharyngeal microaspiration. To diagnose tracheobronchial colonization, quantitative aspirate was performed after intubation and two times a week until extubation. In patients with suspected VAP, quantitative tracheal aspirate or bronchoalveolar lavage was performed to confirm the diagnosis. We hypothesized that the use of conical-cuffed tracheal tubes would reduce the incidence of abundant microaspiration of gastric contents from 50 to 30 % of study patients. With a two-sided alpha risk of 5 %, a power of 80 %, and to account for an anticipated rate of 10 % of patients without any tracheal secretions, 312 patients had to be recruited.

Results A total of 326 patients were included during the study period (June 2014–September 2015). Pepsin and salivary amylase measurement is actually performed, and all results should be available in November 2015. Statistical analyses will be performed in mid-December, and study results will be presented at the 2016 Réanimation Congress.

Discussion Our study is sufficiently powered to detect a significant difference in microaspiration of gastric contents between patients intubated with conical-cuffed tracheal tubes and standard-cuffed tracheal tubes. One of the strengths of this study is the use of quantitative measurement of pepsin as a marker of microaspiration.

Conclusion BEST CUFF is the first randomized controlled study evaluating the impact of PVC tracheal cuff shape on microaspiration of gastric contents. The results will be presented at the 2016 Réanimation Congress.

Competing interests None.

O2 Bicarbonate versus saline for contrast-induced acute kidney injury prevention in critically ill patients

Correspondence: Xavier Valette -

Annals of Intensive Care 2016, 6(Suppl 1):O2

Introduction The administration of contrast media for imaging and interventional procedures is frequent in intensive care unit (ICU). Whether volume expansion with bicarbonate is better than isotonic saline to reduce the risk of contrast-induced acute kidney injury (CI-AKI) in critically ill patients is unknown.

Materials and methods The HYDRAREA study was a randomized, double-blind, multicenter trial including adults admitted in three ICUs in France who received intravascular contrast media. Patients with unstable renal function (increase in serum creatinine level of at least 0.3 mg/dL during the previous 48 h, anuria during the previous 12 h, renal replacement therapy) or contraindication to volume expansion or bicarbonate load (uncontrolled cardiogenic pulmonary edema, metabolic alkalosis with pH >7.50, hypokalemia <3 mmol/L, chronic hypercapnia) were not included. Patients were randomized in a 1:1 ratio to receive 0.9 % sodium chloride or 1.4 % sodium bicarbonate administered with the same infusion protocol: 3 mL/kg for 1 h started before contrast media administration followed by 1 mL/kg/h for 6 h. The primary outcome was the incidence of CI-AKI defined by the Acute Kidney Injury Network criteria within 72 h after contrast medium administration. Secondary outcomes included CI-AKI using alternative definitions (AKIN without urine output criteria, RIFLE classification, creatinine increase ≥0.5 mg/dL or 25 %), need for renal replacement therapy, ICU length of stay and mortality.

Results Among the 307 patients randomized, the mean (SD) age was 56 ± 15 years and 208 (68 %) patients were male. Patients were admitted in ICU for medical reason in 132 (43 %) patients, for emergency surgery in 118 (38 %) patients, following trauma in 45 (15 %) patients and after a scheduled surgery in 12 (4 %) patients. The mean (SD) SAPS II and SOFA scores at ICU admission were 42 ± 17 and 7 ± 4, respectively. Contrast media were used for computed tomography in 193 (63 %) patients and for arteriography in 114 (37 %) patients. Low-osmolar contrast media were used in 276 (90 %) patients. At randomization, 243 (79 %) patients were under mechanical ventilation, 100 (33 %) patients needed vasopressor, 104 (34 %) patients had sepsis, and 40 (13 %) patients had a RIFLE stage equal or higher than the “Risk of Kidney Injury” class. Patient characteristics were well balanced between groups excepted for the use of nephrotoxic medications which was more common in saline group (40 (26 %) patients vs 59 (38 %) patients; p < 0.05).

As expected, patients receiving bicarbonate (n = 151) had higher urinary pH at the end of the infusion than patients receiving saline (n = 156) (6.7 ± 2.1 vs 6.2 ± 1.8, respectively; p < 0.001). The incidence of CI-AKI was similar in both groups: 52 (33.3 %) patients in saline group and 53 (35 %) patients in bicarbonate group (RR 0.96; 95 % CI [0.70–1.30]; p = 0.81). There was no difference between groups whatever the definition used to define CI-AKI. The need for renal replacement therapy (five (3.2 %) patients in saline group vs six (4 %) patients in bicarbonate group; RR 0.81; 95 % CI [0.25–2.59]; p = 0.77), ICU length of stay (24.7 ± 22.9 days in saline group vs 23 ± 23.8 days in bicarbonate group, p = 0.52) and mortality (25 (16 %) patients in saline group vs 24 (15.9 %) patients in bicarbonate group; RR 1.00; 95 % CI [0.60–1.68]; p = 0.97) were also similar between groups.

Conclusion Hydration with sodium bicarbonate for CI-AKI prevention in critically ill patients does not provide supplementary beneficial effect in comparison with hydration by isotonic sodium chloride.

Competing interests None.

O3 Neurally adjusted ventilatory assist in the early phase of weaning from mechanical ventilation: a multicenter randomized study

Correspondence: Alexandre Demoule -

Annals of Intensive Care 2016, 6(Suppl 1):O3

Introduction Neurally adjusted ventilatory assist (NAVA) is a ventilatory mode that tailors the level of assistance delivered by the ventilator to the electromyographic activity of the diaphragm. The objective of the present study was to compare the impact of mechanical ventilation with NAVA or pressure support ventilation (PSV) on the early phase of weaning from mechanical ventilation.

Patients and methods A multicenter randomized controlled trial of 128 intubated adults recovering from an acute respiratory failure was conducted in 11 intensive care units (ICUs) from April 2010 to June 2012. Patients were randomly assigned to NAVA or PSV. The primary endpoint was the probability to remain in an assisted mode during the entire first 48 h. Secondary endpoints included asynchrony index, duration of mechanical ventilation and hospital mortality at 28 days.

Results The proportion of patients remaining in an assisted mode during the entire first 48 h was 67.7 % in the NAVA group versus 69.4 % in the PSV group (p = 0.47). The time spend in an assisted mode similar in the two groups (47 [43–48] vs. 47 [40–48], p = 0.55). The asynchrony index was lower in the NAVA group (19.7 vs. 32.6 %). Duration of mechanical ventilation was similar in the two groups (12.0 [7.0–17.0] in the NAVA group vs. 13.5 [10.0–21.0] in the PSV group, p = 0.121) as was the ICU mortality rate (12.9 vs. 21.2 %, p = 0.21). More patients required post-extubation noninvasive mechanical ventilation in the PSV group (66.6 vs. 43.5 %, p = 0.008).

Conclusion NAVA can be applied efficiently in a clinical setting and improves patient ventilator interaction. However, NAVA does not increase the probability to remain in an assisted mode during the first 48 h.

Competing interests A. Demoule: reserche contract with Maquet, Drager, Philips, Covidien. Consulting for Covidien. Speaker for Covidien.

O4 Very high volume hemofiltration with the Cascade system in septic shock patients

Correspondence: Jean-Pierre Quenot -

Annals of Intensive Care 2016, 6(Suppl 1):O4

Introduction We compared hemodynamic and biological effects of the Cascade system, which uses very high volume hemofiltration (HVHF) (120 ml/kg/h), with those of usual care in patients with septic shock.

Patients and methods Multicenter, prospective, randomized, open-label trial was conducted in three intensive care units (ICU). Adults with septic shock with administration of epinephrine/norepinephrine were eligible. Patients were randomized to usual care plus HVHF (Cascade group) or usual care alone (Control group). Primary endpoint was the number of catecholamine-free days up to 28 days after randomization. Secondary endpoints were number of days free of mechanical ventilation, renal replacement therapy (RRT) or ICU up to 90 days, and 7-, 28- and 90-day mortality.

Results We included 60 patients (29 Cascade and 31 usual care). Baseline characteristics were comparable. Median number of catecholamine-free days was 22 [IQR 11–23] versus 20 [0–25] for Cascade versus Control; there was no significant difference even after adjustment. There was no significant difference in number of mechanical ventilation-free days or ICU requirement. Median number of RRT-free days was 85 [46–90] versus 74 [0–90] for Cascade versus Control groups, p = 0.42. By multivariate analysis, the number of RRT-free days was significantly higher in the Cascade group (up to 25 days higher after adjustment). There was no difference in mortality at 7, 28 or 90 days.

Conclusion Very HVHF using the Cascade system can safely be used in patients presenting septic shock, but was not associated with a reduction in the need for catecholamines during the first 28 days.

Competing interests None.

O5 Effect of rapid response systems on hospital mortality, a prospective interventional study and systematic review

Correspondence: Boris Jung -

Annals of Intensive Care 2016, 6(Suppl 1):O5

Introduction Although rapid response systems (RRSs) are known to reduce in hospital cardiac arrest rate, their effect on mortality remains in question. The present study aimed at evaluating the effect of a medical emergency team (MET) implementation on mortality in hospitalized patients.

Patients and methods A prospective study was conducted in the four hospitals of the regional healthcare center of Montpellier, France. An intensivist-led MET was implemented on a 24/7 basis in only one of the four hospitals from January 2012 to June 2012. Patients >18 years admitted for more than 24 h in the medical-surgical wards from July 2010 to December 2011 (pre-MET period) and from July 2012 to December 2013 (MET period) were included. The main outcome was unexpected mortality in hospitalized patients. An updated systematic review and meta-analysis were also performed.

Results A total of 137,251 patients were admitted for 24 h or more in the medical-surgical wards during the pre-MET and MET periods. MET implementation was associated with a decrease in unexpected mortality rate in the hospital that implemented MET (from 21.9 to 17.4 per 1000 admissions; P = 0.002). Reduction in unexpected mortality associated with MET implementation could be estimated at 1.5 lives saved per week in the MET hospital. In the three other hospitals, mortality rate was not significantly modified (from 19.5 to 19.9 per 1000 admissions; P = 0.69). Patients in the MET hospital were more frequently admitted to the intensive care unit (ICU) during the MET period (45.8 vs 52.9 per 1000; P = 0.002), and their sequential organ failure assessment (SOFA) score upon ICU admission significantly decreased from 7 [4–10] to 5 [2–9]; P < 0.001. The updated meta-analysis including the present results showed that RRS was associated with a significant decrease in both unexpected (OR 0.51; 95 % CI 0.35–0.76) and overall mortality (OR 0.89; 95 % CI 0.85–0.93).

Conclusion In the present prospective study, implementation of a MET was associated with a decrease in unexpected and overall mortality. Updated meta-analysis confirms the benefit of RRS on unexpected and overall mortality (Fig. 1).

Fig. 1
figure 1

Evolution of unexpected death rate per 1000 admissions by month in the MET pavilion (left side) and in the three other pavilions (right side). Dotted lines represent the mean rate per month. Grey rectangles represent the standard deviations

Competing interests None.

O6 Beta-lactams serum concentrations in critically ill cirrhotic patients: a matched control study

Correspondence: Olivier Lheureux -

Annals of Intensive Care 2016, 6(Suppl 1):O6

Introduction Few data are available on the pharmacokinetics of β-lactams in critically ill cirrhotic patients. The objective of this study was to evaluate whether β-lactam concentrations were altered in patients with cirrhosis compared with other critically ill patients and to identify the principal risk factors for any differences.

Materials and methods We reviewed data from critically ill cirrhotic patients and matched controls in which routine therapeutic drug monitoring of broad-spectrum β-lactam antibiotics (piperacillin/tazobactam; meropenem) was performed. Serum drug concentrations were measured twice during the elimination phase by high-performance liquid chromatography. Antibiotic pharmacokinetics was calculated using a one-compartment model. We considered therapy was adequate when serum drug concentrations were between 4 and 8 times the minimal inhibitory concentration of Pseudomonas aeruginosa during optimal periods of time for each drug ≥50 % for piperacillin/tazobactam; ≥40 % for meropenem).

Results We studied 38 cirrhotic patients (16 for piperacillin/tazobactam and 22 for meropenem) and 38 matched controls. Drug dosing was similar in the two groups. The pharmacokinetic analysis showed a lower volume of distribution of meropenem (p = 0.05) and a lower antibiotic clearance of piperacillin/tazobactam (p = 0.009) in patients with cirrhosis when compared to non-cirrhotic patients. Patients with cirrhosis more often had excessive (23/38) than insufficient (5/38) or adequate (10/38) serum β-lactam concentrations; they also had excessive drug concentrations more frequently than did control patients (23/38 vs. 13/38—p = 0.02), particularly for piperacillin/tazobactam (9/16 vs. 2/16—p = 0.01). The only variable significantly associated with insufficient drug concentrations was the creatinine clearance on the day of the therapeutic drug monitoring. No specific variable was associated with excessive drug concentrations.

Conclusion This case–control study shows that standard doses of β-lactams result in excessive serum concentrations in two-thirds of critically ill patients with cirrhosis, especially in those treated with piperacillin/tazobactam. Our results support routine β-lactam therapeutic drug monitoring in this patient population.

Competing interests None.


1. Roberts JA, Paul SK, Akova M, Bassetti M, De Waele JJ, Dimopoulos G et al. DALI: defining antibiotic levels in intensive care unit patients: are current β-lactam antibiotic doses sufficient for critically ill patients? Clin Infect Dis. 2014;58(8):1072–1083.

2. Gustot T, Felleiter P, Pickkers P, Sakr Y, Rello J, Velissaris D et al. Impact of infection on the prognosis of critically ill cirrhotic patients: results from a large worldwide study. Liver Int. 2014;34(10):1496–1503.

O7 Systematic overdosing of oxa- and cloxacillin in severe infections treated in ICU: Risk factors and side effects

Correspondence: Mathilde Neuville -

Annals of Intensive Care 2016, 6(Suppl 1):O7

Introduction Oxacillin and cloxacillin are the most frequently used penicillins for the treatment of severe methicillin-susceptible Staphylococcus aureus (MSSA) infections in intensive care units (ICU), especially endocarditis. No adaptation of the dosage is required in case of renal failure. We wanted to assess the risk factors for overdosing in ICU, and the related observed side effects.

Patients and methods All patients with a therapeutic drug monitoring (TDM) of oxa- or cloxa-cillin between 2008 and 2014 were included. The optimal trough concentration was considered to be 20 mg/L, i.e., 10 times the minimum inhibitory concentration of cloxaxillin for MSSA. Data concerning the medical history, the characteristics of the infection, the given treatment, the renal function and the attributed side effects of overdosing were collected. A negative binomial model was used to compute the measured trough concentrations.

Results Sixty-two patients were included in this study. We found a median trough plasma concentration of 134.3 mg/L [IQR 65.3–201 mg/L]. Only one patient reached the target concentration; all other patients were overdosed. Eleven patients (17.7 %) experienced neurological side effects attributed to a high antibiotic concentration, i.e., persistent coma and delirium. When adjusted on the dosage used, the first trough concentration was significantly increased by a creatinine clearance below 10 ml/min (with or without hemodialysis) and by endocarditis as the source of infection (Fig. 2).

Fig. 2
figure 2

Univariate and multivariate statistical analysis. Day 0 = day of ICU admission, T-day = day of treatment initiation, D-day = day of TDM

Conclusion With the recommended dose of 12 g/day for cloxacillin treatment in case of endocarditis and severe infections occurring in ICU, almost all patients are largely overdosed. Considering the high frequency of observed side effects, doses should be accurately monitored and reduced, particularly when renal replacement therapy is needed.

Competing interests None.

O8 Amikacin peak concentrations in patients receiving extracorporeal membrane oxygenation (ECMO) support: a case–control study

Correspondence: Elodie Gélisse -

Annals of Intensive Care 2016, 6(Suppl 1):O8

Introduction More than half of adult patients receiving extracorporeal membrane oxygenation (ECMO) require an antibiotic therapy during their stay in ICU. Amikacin is the most used aminoglycoside for the treatment of septic shock given its good bactericidal activity on Pseudomonas aeruginosa and the low resistance rate observed with the other Gram-negative bacilli. However, data on pharmacokinetics of amikacin in patients receiving ECMO support have scantly been studied in these conditions. We aimed to report on the distribution of peak serum concentration (Cmax) of amikacin and on factors associated with insufficient Cmax in a population of critically ill patients receiving ECMO.

Materials and methods We conducted an observational single-center study of patients admitted to a general ICU. Inclusion criteria were: (1) age ≥18 years and (2) a suspected Gram-negative infection requiring a loading dose of intravenous amikacin. Amikacin was routinely administered intravenously at a dose of 25 mg/kg of total body weight over a 30-min infusion time. Cmax was measured 30 min after the end of infusion and trough serum concentration 24 h after the end of infusion (Cmin). ECMO patients were matched with critically ill patients without ECMO support (controls) for characteristics that had been previously identified as independent predictors for insufficient Cmax, namely body mass index (BMI), cirrhosis and 24-h fluid balance. An amikacin Cmax >60 mg/L and <80 mg/l was considered the target concentration (adequate drug concentrations). An amikacin Cmax <60 mg/L was considered as an insufficient drug concentration, and a Cmax >80 mg/l was considered as an excessive one. The primary outcome was the distribution of Cmax (mg/L) between ECMO patients and controls.

Results A total of 50 episodes in 46 ECMO patients were matched with 50 controls. Baseline characteristics did not differ between groups, except for age (61 vs. 63.5 years), reason for ICU admission and SOFA score (12 [10–14] vs. 9 [6–11]). At the day of Cmax, in ECMO patients the SOFA score was 9 [5.3–9.8], and 100 % episodes required mechanical ventilation, 96 % vasopressors and 44 % renal replacement therapy. There were no significant differences in serum concentrations between ECMO and non-ECMO patients: The amikacin Cmax was 71.7 [58.9–79.7] mg/l for ECMO patients versus 68.4 [53–81] mg/l for non-ECMO patients (p = 0.36). The proportion of insufficient [13/50 (26 %) episodes vs. 17/50 (34 %) episodes), adequate (25/50 (50 %) episodes vs. 18/50 (36 %) episodes] and excessive [12/50 (24 %) episodes vs. 15/50 (30 %) episodes] drug concentrations was similar in ECMO and non-ECMO patients (Table 1). Cmin was measured in 43 (86 %) episodes, and 28/43 (65 %) measurements were above the toxic threshold of 5 mg/L in ECMO patients versus 30/50 (60 %) in non-ECMO patients.

Table 1 Episodes characteristics at ICU admission and time of C max

Conclusion ECMO therapy did not significantly impact on peak and trough serum concentrations of amikacin.

Competing interests None.


1. Schmidt M, Brechot N, Hariri S, Guiguet M, Luyt CE, Makri R, et al. Nosocomial Infections in Adult Cardiogenic Shock Patients Supported by Venoarterial Extracorporeal Membrane Oxygenation. Clin Infect Dis. 2012 Dec 15;55(12):1633–41.

2. De Montmollin E, Bouadma L, Gault N, Mourvillier B, Mariotte E, Chemam S, et al. Predictors of insufficient amikacin peak concentration in critically ill patients receiving a 25 mg/kg total body weight regimen. Intensive Care Med. 2014 Jul;40(7):998–1005.

O9 A high aminoglycoside regimen associated with renal replacement therapy for the treatment of multi-drug-resistant pathogens

Correspondence: Alexandre Brasseur -

Annals of Intensive Care 2016, 6(Suppl 1):O9

Introduction Infections caused by multi-drug-resistant (MDR) Gram-negative (GN) organisms in critically ill patients are a therapeutic challenge. The administration of high-dose aminoglycoside (HDA) therapy coupled with high-flow continuous venovenous hemodiafiltration (CVVHDF) could allow required high drug peaks to be achieved with acceptable drug elimination.

Patients and methods All adult patients present on the intensive care unit (ICU) between October 2009 and July 2014 who had MDR-GN sepsis were considered for HDA and high-flow (>45 mL/kg/h) CVVHDF when an isolated pathogen was susceptible or intermediate susceptible to aminoglycosides, and the patient’s condition was not improving with conventional therapy. Optimal antibacterial activity was defined as a peak concentration of at least eight times the minimal inhibitory concentration.

Results Fifteen patients infected with MDR-GN pathogens (11 with Pseudomonas aeruginosa; ten with abdominal and five with respiratory infections) were treated with amikacin (n = 11), gentamicin (n = 3) or tobramycin (n = 1) and high-flow CVVHDF. A favorable clinical response was observed in eight (53 %) patients, including three in whom microbial eradication was obtained. Six patients were discharged alive from the ICU and five from the hospital. No renal toxicity was observed among survivors.

Conclusion In this cohort of septic patients with MDR-GN infections, HDA combined with high-flow CVVHDF represented a valuable therapeutic option. The effectiveness of this approach should be further evaluated in larger studies.

Competing interests None.

O10 Optimization of administration of vancomycin in septic patients: a prospective randomized study

Correspondence: Valerie Van Ruychevelt -

Annals of Intensive Care 2016, 6(Suppl 1):O10

Introduction Continuous intravenous administration of vancomycin (CIV) in intensive care unit (ICU) patients requires dose adjustment according to weight and renal function, but its assessment remains difficult. The aim of this study was to validate a new dose regimen of CIV correlated with renal function assessed by creatinine clearance (Creat Cl).

Materials and methods In this prospective randomized study, we evaluated septic critically ill patients in whom CIV was given for more than 48 h in the absence of renal replacement therapy. The CIV protocol was developed based on recommendations of IDSA for the loading dose (30 mg/kg), but a new protocol for the maintenance dose was tested (depending on weight and daily Creat Cl). The serum target vancomycin (vanco) concentration was 25–30 μg/ml. Patients were randomized in three groups by daily assessment of renal function by a Creat Cl of 4 h (from 4 to 8 a.m.) or 12 h of Creat Cl (from 8 p.m. to 4 a.m.) or by the MDRD. Acute kidney injury (AKI) was defined as a daily urine output <0.5 ml/kg/h and/or an increase in the serum creatinine of ≥0.3 mg/dL from baseline levels. Multivariable logistic regression analysis was performed to identify predictors of CIV under-/over-dose and predictors of AKI.

Results During 23 months, in a medico-surgical 36 beds ICU (CHU Charleroi, Belgium), 67 patients were randomized: 21 (31 %) in the MDRD group, 21 (31 %) in the 4 h Creat Cl and 25 (38 %) in the 12 h Creat Cl. The three groups were comparable in demographic terms, severity score, kidney function before CIV and bacteriology. For 7/67 (10 %) patients, CIV was targeted on bacteriological findings (one MRSA and six Enterococcus faecium). In the other patients, vanco was introduced empirically for clinically evolving sepsis. Vanco concentration <25 μg/ml on day 1 was found in 15 patients (22 %) in the MDRD group, 11 (16 %) in 4 h Creat Cl and 15 (22 %) in 12 h Creat Cl (p = 0.5), and in 12 (17 %), 12 (17 %) and 13 (19 %), respectively, on day 2 (p = 0.8). In logistic regression analysis, BMI <30 and an elevated Creat Cl >90 were found independent predictors of under-dosing vanco (<25 μg/ml) at day 1 with an OR of 1.098 [1.01–1.19] and 0.935 [0.8–1], respectively. The day 1–day 2 CIV overdose was found to be an independent risk factor for AKI with an OR of 7.5 [2.2–23.7].

Discussion Our protocol cannot optimally achieve target levels on days 1–2. However, the chosen therapeutic target (25–30 μg/ml) seems a posteriori not justified given the local epidemiology. With the target proposed in the literature (20–30 μg/ml), vanco concentration at day 1 was <20 μg/ml in nine patients (43 %) in MDRD group, four (19 %) in 4 h CCU and nine (36 %) in 12 h CCU, and in 3 (14 %), 3 (14 %) and 5 (20 %), respectively, at day 2. It would therefore be preferable to target therapeutic levels between 25 and 30 μg/ml only for patients in the presence of multi-resistant germs demonstrated. For empirical antibiotic choice, therapeutic target [20–30 μg/ml] is probably enough and more careful. The assessment of renal function using three different methods failed to highlight difference in Creat Cl. The evaluation of Creat Cl by the MDRD formula may be sufficient.

Conclusion Insufficient vancomycin concentration is a frequent finding at days 1–2. Predictors were BMI and high Creat Cl. Vanco level >30 μg/ml at days 1–2 is an independent predictive factor for AKI with an OR of 7.5.

Competing interests None.


1. Rybak, M. et al. Therapeutic monitoring of vancomycin in adult patients: a consensus review of the American Society of Health-System Pharmacists, the Infectious Diseases Society of America, and the Society of Infectious Diseases Pharmacists. Am. J. Health-Syst. Pharm. AJHP Off. J. Am. Soc. Health-Syst. Pharm. 66, 82–98 (2009).

2. Ocampos-Martinez, E. et al. Determinants of early inadequate vancomycin concentrations during continuous infusion in septic patients. Int. J. Antimicrob. Agents 39, 332–337 (2012).

O11 Impact of elevated intra-abdominal pressure on the ability of dynamic parameters to predict fluid responsiveness

Correspondence: Antoine Vieillard-Baron -

Annals of Intensive Care 2016, 6(Suppl 1):O11

Introduction Elevated intra-abdominal pressure (IAP) has been suggested to decrease the accuracy of dynamic parameters of fluid responsiveness, especially for the distensibility index of the inferior vena cava (∆IVC). The goal of our study was to confirm this limitation in a large series of unselected patients.

Materials and methods This is a sub-study of an observational, prospective and multicenter study (Hemopred) that we conducted between November 2012 and November 2014. All sedated patients under mechanical ventilation who required an echocardiographic assessment for an acute circulatory failure, with an inserted central venous catheter and arterial catheter, were eligible. Acute circulatory failure was defined as a sustained hypotension (systolic blood pressure <90 mmHg or mean blood pressure <65 mmHg) and/or the presence of clinical signs of hypoperfusion (e.g., mottled skin, oliguria), metabolic acidosis (pH <7.35 and base excess less than −5 mmol/L), elevated lactate (>2 mmol/L) or decreased central venous oxygen saturation (ScvO2 <70 %). Dynamic parameters used to predict fluid responsiveness assessed in this study were pulse pressure variations (∆PP) and the following echocardiographic indices: respiratory variations of aortic Doppler maximal velocity (∆Vmax), collapsibility index of superior vena cava (∆SVC) and ∆IVC.

At the time of echocardiography, a systematic measurement of the bladder pressure was performed as recommended and elevated IAP was defined as a pressure ≥12 mmHg.

Response to fluids was defined by an increase in cardiac output of more than 10 % during a passive leg raising (PLR), as previously validated. Area under the ROC curve (AUC) was built for each recorded dynamic parameter.

Results Among the 540 patients included in the study, 174 (32 %) had an elevated IAP (group 1). No difference was observed compared with the remaining patients without elevated IAP (group 2), except for patients with an abdominal surgery (32 vs 23 %, p = 0.03). Among group 1, 81 patients had an IAP between 12 and 15 mmHg and 93 patients had an IAP >15 mmHg. Central venous pressure (CVP) was increased in group 1 compared with group 2 (11 ± 5 vs 9 ± 5 mmHg, p < 0.001), as well as plateau pressure (Pplat) (20 ± 5 vs 18 ± 5 cmH2O, p < 0.001). In group 1, 81 (46 %) patients were classified as responders by the PLR, compared with 124 (40 %) in group 2 (p = 0.18). No difference between the AUC for ∆IVC (0.64 vs 0.59, p = 0.46), ∆PP (0.71 vs 0.66, p = 0.47) and ∆Vmax (0.83 vs 0.71, p = 0.06) was observed between groups 1 and 2, respectively, although AUC was increased in group 1 for ∆SVC (0.72 vs 0.82, p = 0.02). No difference was observed regarding the feasibility between the two groups for any parameters, although feasibility for ∆IVC was 74 % in group 1 compared with 80 % in group 2 (p = 0.1).

Conclusion Thirty-two percent of patients in our series of unselected ventilated patient with shock had an elevated IAP. They demonstrated higher CVP and Pplat and more benefited from abdominal surgery. Elevated IAP did not have significant effect on accuracy of dynamic parameters of fluid responsiveness, even though ∆IVC was less frequently recorded (nonsignificant) and ∆SVC had a better prediction.

Competing interests None.

O12 Passive leg raising for predicting fluid responsiveness: a systematic review and meta-analysis

Correspondence: Xavier Monnet -

Annals of Intensive Care 2016, 6(Suppl 1):O12

Introduction We performed a systematic review and meta-analysis of studies that investigated the passive leg raising (PLR)-induced changes in cardiac output (CO) or surrogates and the PLR-induced changes in arterial pulse pressure (PP) as predictors of fluid responsiveness in adults.

Patients and methods MEDLINE, EMBASE and Cochrane Database of Systematic Reviews were screened for relevant original and review articles.

Results Twenty-one studies (995 patients) were included. CO was measured by echocardiography in six studies, calibrated pulse contour analysis in six studies, bioreactance in four studies, oesophageal Doppler in three studies, either transpulmonary thermodilution or pulmonary artery catheter in one study and suprasternal Doppler in one study. The pooled correlation between the PLR-induced and the volume expansion-induced changes in CO was 0.76 (0.73–0.80). The best threshold was a PLR-induced increase in CO ≥10 ± 2 %. The pooled sensitivity was 0.85 (0.81–0.88), and the pooled specificity was 0.91 (0.88–0.93). The area under the ROC curve was 0.95 ± 0.01. Eight studies (432 patients) met the selection criteria for the PLR-induced changes in PP. For the PLR-induced changes in PP, the pooled sensitivity was 0.56 (0.49–0.53), the pooled specificity was 0.83 (0.77–0.88), and the pooled area under the ROC curve was 0.77 ± 0.05.

Conclusion The changes in CO induced by a PLR test are highly reliable in predicting the response of CO to volume expansion in adult patients with acute circulatory failure. When its effects are assessed by changes in PP, the specificity of the PLR test remains acceptable, but its sensitivity is poor.

Competing interests XM and J-LT are members of the Medical Advisory Board of Maquet.

O13 Predicting volume responsiveness by using combined end-expiratory and end-inspiratory occlusion tests with echocardiography

Correspondence: Mathieu Jozwiak -

Annals of Intensive Care 2016, 6(Suppl 1):O13

Introduction The changes in intrathoracic pressure induced by end-expiratory (1) and end-inspiratory occlusions induce significant changes in cardiac preload that may be used to test preload dependence. We tested whether the effects of respiratory occlusions on the velocity–time integral of the subaortic flow (VTI) could serve as a functional test for predicting volume responsiveness in patients with circulatory failure.

Patients and methods In ten mechanically ventilated patients with circulatory failure in whom volume expansion was planned, we continuously measured VTI (transthoracic echocardiography) during the 5 last seconds of 15-s end-inspiratory and end-expiratory occlusions, separated by 1 min, and after the subsequent infusion of 500-mL saline. Patients in whom volume expansion induced an increase in cardiac index (transpulmonary thermodilution, PICCO2) >15 % were defined as “volume-responders”.

Results Volume expansion increased cardiac index by more than 15 % (2.75 ± 0.66–3.36 ± 0.71 L/min/m2, p < 0.05) in four patients. In these volume-responders before volume expansion, the end-expiratory occlusion increased VTI to a significantly larger extent than in non-volume-responders (9 ± 5 vs. 3 ± 1 % p = 0.02, respectively). Similarly, before volume expansion, the end-inspiratory occlusion decreased VTI to a significantly larger extent in volume-responders than in non-volume-responders (−11 ± 3 vs. −5 ± 2 % p = 0.01, respectively). Volume responsiveness was predicted by an increase in VTI >4 % during the end-expiratory occlusion with a sensitivity of 75 % (95 % CI 19–99 %) and a specificity of 100 % (95 % CI 54–100 %). If the absolute values of the VTI changes recorded during end-expiratory and end-inspiratory occlusions were added, the predictive accuracy of volume responsiveness tended to be better than for the VTI changes recorded during the only end-expiratory occlusion [area under the ROC curve: 0.917 (95 % CI 0.575–0.999) vs. 0.771 (95 % CI 0.415–0.964), p = 0.24, respectively].

Conclusion The sum of the absolute values of changes in VTI recorded during consecutive end-expiratory and end-inspiratory occlusions could reliably predict volume responsiveness. This prediction tends to be better than the changes in VTI during the only end-expiratory occlusion. More patients will be included in order to confirm these preliminary results.

Competing interests None.


1. Monnet et al. Crit Care Med. 2009; 37:951–956.

O14 Early dynamic left intraventricular obstruction is associated with hypovolemia and hight mortality in septic shock patients

Correspondence: Jean-Louis Chauvet

Annals of Intensive Care 2016, 6(Suppl 1):O14

Introduction Based on previously published case reports demonstrating dynamic left intraventricular obstruction triggered by hypovolemia or catecholamines, this study aimed to establish: (1) IVO occurrence in septic shock patients; (2) correlation between the intraventricular gradient and volume status and fluid responsiveness; and (3) mortality rate.

Patients and methods We prospectively analyzed patients with septic shock admitted to a general ICU over a 28-month period who presented Doppler signs of left intraventricular obstruction. Clinical characteristics and hemodynamic parameters as well as echocardiographic data regarding left ventricular function, size, and calculated mass, and left ventricular outflow Doppler pattern and velocity before and after fluid infusions were recorded.

Results During the study period, 218 patients with septic shock were admitted to our ICU. Hemodynamic and echocardiographic characteristics of patients with intraventricular obstruction before and after fluid infusion are given in Table 2. Left intraventricular obstruction was observed in 47 (22 %) patients. Mortality rate at 28 days was found to be higher in patients with than in patients without left intraventricular obstruction (55 vs 33 %, p < 0.01). Small, hypercontractile left ventricles (end-diastolic left ventricular surface 4.7 ± 2.1 cm2/m2 and ejection fraction 82 ± 12 %), and frequent pseudohypertrophy were found in these patients. A rise ≥12 % in stroke index was found in 87 % of patients with left intraventricular obstruction, with a drop of 47 % in left intraventricular obstruction after fluid infusion.

Table 2 Hemodynamic and echocardiographic characteristics of patients with intraventricular obstruction before and after fluid infusion

A late velocity of diastolic mitral flow, A′ late velocity of diastolic mitral annulus motion, E early diastolic velocity of mitral flow, E′ early velocity of diastolic mitral annulus motion, ns not significant.

Discussion The presence of left intraventricular obstruction is associated with high mortality rate. The data showed an association between the presence of this obstructive flow pattern on left ventricular Doppler flow and the presence of hypovolemia and cardiac hypercontractility and also demonstrated that these patients frequently present small and pseudo-hypertrophic left ventricular. Furthermore, this study also demonstrates the high rate of fluid responsiveness in these patients with left intraventricular obstruction despite the absence of significant pulse pressure variations.

Conclusion In our study, left intraventricular obstruction is a frequent event in septic shock patients bearing an important correlation with fluid responsiveness. Mortality rate was found to be higher in these patients in comparison with patients without obstruction.

Competing interests None.

O15 Predictive factors for poor hemodynamic tolerance to fluid removal in ICU: the DepleRea study

Correspondence: Julien Maizel -

Annals of Intensive Care 2016, 6(Suppl 1):O15

Introduction In the intensive care unit (ICU), the presence of edema after the initial phase of fluid resuscitation is an independent poor prognostic factor. The fluid removal (using diuretics or during renal replacement therapy) is an important component of care. However, the depletion can decrease cardiac output and provoke hypotension. Therefore, it can be harmful. Our objective was to determine factors for poor hemodynamic tolerance to fluid removal.

Patients and methods We conducted a single-center prospective observational study in the medical ICU of the University Hospital of Amiens in France from March to September 2015. Inclusion criteria were critical patients initially requiring fluid therapy and then hemodynamically stabilized, in whom fluid removal was decided. Exclusion criterion was failure to perform a thoracic (or lung) or a cardiac ultrasonography. We collected the main clinical and biological characteristics. Hemodynamic status before the initiation of depletion was assessed by transthoracic echocardiography. Thoracic ultrasound was performed to estimate extravascular lung water. Our primary outcome was poor hemodynamic tolerance in the 24 h following the initiation of fluid, corresponding to at least one episode of hypotension (mean arterial pressure <65 mmHg) or initiation/increase of catecholamines.

Results Sixteen patients were included in the study. The median age was 69 years (IQR 16.3), SAPS II was 52 (IQR 25.8), and the proportion of men was 56.3 %. Median time to initiation of fluid removal was 4 days (IQR 4.5). Fluid removal was performed in nine cases by loop diuretics and in eight cases with renal replacement therapy. The median depletion was 1617.5 ml (IQR 2025.8). Fifty percent were intubated and ventilated. Seven out of eight patients had spontaneous ventilatory cycles. Pulse pressure variation could be calculated in only one patient. Nine patients (56.2 %) did not tolerate fluid removal. The median SOFA at inclusion was significantly higher in patients who had poor hemodynamic tolerance to depletion compared with patients who well tolerated the depletion (9 ± 3 vs 5 ± 5.5, p = 0.036). All patients with catecholamine therapy (n = 4; median posology 0.087 μg/kg/min) did not hemodynamically tolerate fluid removal. Initial median cardiac index in patients who developed arterial hypotension was 2.7 ± 1.6 l/min/m2 and 3.6 ± 3 l/min/m2 among those who did not develop hypotension (p = 0.088). All patients having cardiac index lower than 2.7 l/min/m2 developed hypotension. Diastolic blood pressure of patients with hemodynamic instability was significantly lower (56 ± 10.5 vs 72 ± 18 mmHg, p = 0.042). Four out of five (80 %) of the patients with positive passive leg raising test showed hemodynamic intolerance to fluid removal, while only one patient (16.7 %) had a positive passive leg raising test in the well-tolerated depletion group (p = 0.58). Patients with well hemodynamic tolerance to fluid removal had approximately twice as many B lines than patients having poorly tolerated fluid removal (14 ± 17 vs 7 ± 13.3, p = 0.152).

Discussion Our results suggest that a preload dependence (shown by a positive passive leg raising test) and a lower accumulation of extravascular lung water (shown by fewer B lines) are associated with a poor hemodynamic tolerance to fluid removal. A precarious hemodynamic state (cardiac index lower than 2.7 l/min/m2, catecholamines therapy and lower diastolic arterial pressure) also appears predictive of a poor fluid removal hemodynamic tolerance. Patients with poorly tolerated fluid depletion had a higher SOFA score, indicating a greater severity of these patients. These results must be confirmed in a larger cohort.

Conclusion Our results suggest that poor hemodynamic tolerance to fluid removal is associated with a lower diastolic pressure, the presence of catecholamine and a preload dependency as suggested by a low cardiac index and a positive passive leg raising test. The presence of many B lines assessed by thoracic ultrasound seems to be protective of poor hemodynamic tolerance.

Competing interests None.

O16 High-flow nasal cannula: first-line treatment of noninvasive ventilation for infants with bronchiolitis. Applicability and risk factors for failure

Correspondence: Stéphane Leteurtre -

Annals of Intensive Care 2016, 6(Suppl 1):O16

Introduction Viral bronchiolitis is an acute infection disease of the lower respiratory tract. In the most severe cases, acute respiratory failure occurs. High-flow nasal cannula (HFNC) therapy is a relatively new noninvasive ventilator support that can improve respiratory distress for infants with bronchiolitis. The objectives of this study were (1) to evaluate a study design suggesting the use of HFNC as the first-line treatment for infants with bronchiolitis in a pediatric intensive and intermediate care unit (PI&ICU) and (2) to identify risk factors for HFNC failure.

Patients and methods This study was prospective during two consecutive seasons (2013–2014 and 2014–2015) in a PI&ICU (18 beds). Inclusion criteria were patients <2 years with bronchiolitis needing invasive or noninvasive ventilation. During the 2013–2014 season, no recommendation between HFNC and continuous positive airway pressure (CPAP) was proposed. During the 2014–2015 season, a study design suggested the use of the HFNC as noninvasive ventilation as first-line tratment (before C PAP) for bronchiolitis. Primary outcome was the percentage of HFNC (comparison between the two periods). For the secondary outcome, failure of HFNC was defined by the need of CPAP or invasive ventilation. Clinical and biological variables were collected during the PI&ICU stay at different times (before ventilation, H1–H2, H4–H6, H10–H12 and H24). Comparisons of percentage and logistic regression were made with SAS software (SAS Institute, Cary, NC, USA).

Results During the 2013–2014 season, 34 % (14/41 infants) were treated by HFNC at admission compared with 90 % (55/61 infants) during the 2014–2015 season (p < 10−4) (Fig. 3). The intubation rate at admission and during the hospitalization was not different between the two seasons.

Fig. 3
figure 3

Characteristics of ventilation modalities during the 2 periods

During the 2014–2015 season, HFNC therapy failed in 21 infants (38 %; 21/55). In the responders group, the use of HFNC therapy decreased the respiratory rate, the heart rate (HR) and the CO2 (p < 0.001). With multivariate analysis, non-responders patients to HFNC therapy have a hypercapnia higher (OR 1.136; 95 % CI 1.04–1.24; p < 0.01) and a tachycardia lesser (OR 0.95; 95 % CI 0.91–0.99; p = 0.02) compared with responders patients.

Discussion The use of HNCF as first-line treatment for patients with bronchiolitis has been frequently possible and accepted by all the team in our PI&ICU. No differences for complications are observed during these two periods.

Conclusion HFNC can be used as a first-line treatment for moderate or severe bronchiolitis hospitalized in our PI&ICU. HFNC requires a monitoring to determine the failure criteria. A multicenter study will be necessary to confirm these results.

Competing interests None.

O17 Is high-flow nasal cannula better than nasal continuous positive airway pressure for bronchiolitis management in pediatric intensive care unit?

Correspondence: Aurélie Collins -

Annals of Intensive Care 2016, 6(Suppl 1):O17

Introduction Bronchiolitis is the first pathology encountered in pediatric intensive care unit during winter. In the management of respiratory failure, noninvasive ventilation may improve work of breathing and has been suggested to be effective in preventing intubation. High-flow nasal cannula had increasingly been used and tends to replace nasal continuous positive airway pressure in this indication despite the lack of evidence. Nowadays, no randomized control trial had compared nasal continuous positive airway pressure and high-flow nasal cannula in this indication. Only observational studies exist. Most authors have described a superiority of high-flow nasal cannula against nasal continuous positive airway pressure but with some controversies. In this study, we compared retrospectively nasal continuous positive airway pressure and high-flow nasal cannula in patients with bronchiolitis admitted to pediatric intensive care unit during two seasons in our tertiary hospital.

Patients and methods We included 77 patients, in which 13 treated with nasal continuous positive airway pressure and 24 treated with high-flow nasal cannula. We compared vital signs at H0, H1, H2, H3, H6, H9 and H12 after device positioning, length of stay and rescue intubation.

Results There was a significantly lower respiratory rate at H6; fraction of inspired oxygen was significantly lower at H2 and H6, with comparable pulse oximeter oxygen saturation; length of stay tended to be shorter and intubation rate lower in the high-flow nasal cannula group.

Conclusion Our results tend to show a small advantage to use high-flow nasal cannula over nasal continuous positive airway pressure in respiratory failure in bronchiolitis. Larger prospective randomized clinical trials are needed to compare these two techniques.

Competing interests None.

O18 Interest and risk of high-flow cannula during acute hypoxemic pneumonia in children: a retrospective study

Correspondence: Robin Pouyau -

Annals of Intensive Care 2016, 6(Suppl 1):O18

Introduction Pneumonia remains the world’s leading cause of death among children under 5 years old. Treatment of hypoxemic pneumonia is based on oxygen therapy. High-flow nasal cannula (HFNC) is a technique whereby heated and humidified oxygen can be delivered to the nose at high flow rates. It generates a low level of positive pressure in the upper airways and decreases the work of breathing. This technique might be useful to hypoxemic pneumonia treatment and could reduce the need of invasive or noninvasive ventilation in pediatric patients.

Patients and methods This retrospective monocentric study included all the children from 1 to 18 years old, admitted in pediatric intensive care unit between January 1, 2010, and June 1, 2014, and affected by hypoxemic pneumonia. Children affected by bronchiolitis or asthma were excluded. The goals of our study were to determine predictive factors for success or failure of HFNC. The failure group was defined as the group of patients which required the switch of HFNC to another oxygen delivery system (oxygen mask, noninvasive ventilation (NIV) or invasive ventilation). We studied clinical, biological and radiological criteria at different times for the first 48 h. Then, we compared these criteria between the failure group and the success group.

Results Ninety-two children were included. HFNC was a success for 62 patients (67 %). No side effects were reported and the tolerance was good for all the patients. There was no demographic difference between the two groups of patients. The risk of failure was an initial higher FiO2 [mean of 0.69 in failure group vs 0.46 in the other group (p = 0.001)], an initial lower ratio SpO2/FiO2 [164.6 vs 260.2 (p = 0.001)] and a higher PELOD score at admission [5.27 vs 3.35 (p = 0.042)]. This risk of failure of HFNC persisted for the first 48 h. This risk was significantly higher for patients who had three or more affected quadrants in pulmonary chest X-ray (p = 0.006). Immunocompromised patients (n = 26) were at higher risk of failure of HFNC in our population (p = 0.047). In the failure group (n = 30), 15 patients needed NIV, 15 patients were intubated, and four died. There was no hypercapnia in either group and no modification of the capnia at any time. We finally defined criteria and threshold for patients at higher risk of failure of HFNC: immunodepression, FiO2 ≥0.6, ratio SpO2/FiO2 ≤194, PELOD ≥6, three or more affected quadrants in pulmonary chest X-ray.

Conclusion Our study is one of the first on hypoxemic pneumonia in children. HFNC could be secure, useful and effective in this indication. The successful rate is 67 %. Identification of criteria associated with failure could help the physician to optimize the timing of secondary support, mainly intubation. Those criteria should be validated in multicenter and prospective studies. HFNC seems to be a promising technique of ventilation for patients with hypoxemic pneumonia. The rate of failure may be higher in immunocompromised children, patients with a severe hypoxemia or with an other organ failure (Fig. 4).

Fig. 4
figure 4

See text for description

Competing interests None.


1.Frat J-P, Thille AW, Mercat A, Girault C, Ragot S, Perbet S, et al. High-flow oxygen through nasal cannula in acute hypoxemic respiratory failure. N Engl J Med.2015;372(23):2185–96.

O19 Interest of high-flow nasal cannula (HFNC) versus nasal continuous positive airway pressure (nCPAP) during the initial management of severe bronchiolitis in infants: a multicenter randomized controlled trial

Correspondence: Christophe Milesi -

Annals of Intensive Care 2016, 6(Suppl 1):O19

Introduction Concerning the management of severe bronchiolitis, nCPAP is the gold standard. HFNC is increasingly used in this indication even if there are no randomized studies evaluating its efficacy and safety in this population. Our objective was to evaluate the failure rate of HFNC compared with the nCPAP during the initial management of severe bronchiolitis.

Patients and methods Prospective, controlled, randomized, multicenter study was conducted in five French pediatric intensive care units (PICU). Infants from 0 to 6 months old, admitted in PICU for severe bronchiolitis (modified Wood score >3), were randomized into two groups: “HFNC” (2 l/kg/min) and “nCPAP” (6 cm H2O) during 24 h. The primary endpoint was the percentage of failure in both arms during the first 24 h. Failure was define as an increase in (1) clinical score for respiratory distress (modified Wood score) or (2) respiratory rate (RR) or (3) discomfort (EDIN score) or (4) apnea.

Results {median (SD) and %(IC)} A total of 142 infants were included from November 2014 to March 2015. Their weight was 4100 (1200) g, and they were 40 (35) days old, the RR: 53 (13) breath/min, FiO2: 30 (12), modified Wood score: 4.3 (1), PCO2: 59 (14) mmHg, pH: 7.28 (0.1). Seventy one were included in HFNC group and 71 in the nCPAP group. There was no difference between the two groups (weight, age, HR, BP, RR, FIO2, modified Wood score, EDIN, PCO2). The failure rate was higher in the HFNC group 49 (39–62) vs 31 (20–42) % [OR 0.49 (0.2–0.9), p = 0.018]. We did not find any predictive failure criteria. The main cause of failure in the HFNC group was a raise of the respiratory distress (58 %) and discomfort in the nCPAP group (59 %). Failure occurred rapidly during the first 18 h in the two groups. We did not find any difference between the two groups concerning the length of stay [6.2 (6) vs 7.5 (13) days; p = 0.46] or the number of intubation (3/71 vs 5/71; p = 0.51).

Conclusion During severe bronchiolitis management, the failure rate is higher with the high flow compared with nasal CPAP (Fig. 5).

Fig. 5
figure 5

See text for description

Competing interests None.

O20 Outcome of severe cerebral venous thrombosis in intensive care unit: a cohort study

Correspondence: Benjamin Soyer -

Annals of Intensive Care 2016, 6(Suppl 1):O20

Introduction Severity of cerebral venous thrombosis (CVT) may require the transfer to intensive care unit (ICU). This report described the context for CVT transfer to ICU, the strategy of care, and the outcome after 1 year.

Patients and methods We studied a monocentric cohort of 41 consecutive CVT in a French tertiary hospital. Data collected were as follows: demographic data, clinical course, incidence of craniectomy and/or endovascular procedures, outcome in ICU, after 3, and 12 months.


1—Characteristics 73.2 % were female, 47 years old (IQ 26–53), having a SAPS II 41 (32–45), GCS 7 (5–8), and at least one episode of mydriasis in 48.8 %. 48.8 % had seizures before ICU admission and 43.9 % during ICU stay.

2—Lesions Thrombosis location was 80.5 % in lateral sinus and 53.7 % in superior sagittal sinus; intracranial hematoma was present in 78.0 %, signs of intracranial hypertension in 60.9 %, cerebral edema in 58.5 %, and venous ischemia in 43.9 %.

3—Monitoring All patients had an imaging technique (TDM, angio-TDM, and MRI). All patients were closely monitored for CO2 (PCO2 and end-tidal CO2). When feasible, transcranial Doppler was performed in 78 % of patients. Despite the risk of bleeding, eight patients had a monitoring of intracranial pressure.

4—Therapies All patients received heparin therapy, and nine cases had endovascular treatment (21.9 %), osmotherapy (53.7 %), and decompressive craniectomy (16 cases, 39 %) were necessary to control intracranial hypertension. 63.4 % received norepinephrine to improve brain perfusion, and 90.2 % of patients were under mechanical ventilation. Thirty-one patients received antiepileptic drugs.

5—Outcome Ten out of 41 patients (24.4 %) died in ICU, and 18/31 (58.1 %) discharged from ICU had a good outcome (mRS 0–3). After 12 months, 92 % of survivors (23/25) had a mRS between 0 and 3. The proportion of death was 31.7 % at 1 year.

Compared with ICU deaths, survivors were: predominance of female (p = 0.039), a worse initial GCS [4 (3–6); p = 0.0004], a more frequent mydriasis (100 vs. 32.3 %, p = 0.0001) with more frequent signs of intracranial hypertension on imaging techniques (p = 0.003), with a trend for more frequent use of norepinephrine (90 and 54.8 %, respectively) and an anticoagulation more difficult to be adequate (50 vs. 80.6 %, respectively). Comparing good outcome with poor outcome after 1-year evolution, patients mRS 0–3 appeared with a higher platelets count at day 1 (p = 0.03) and day 5 (p = 0.06) during ICU stay, as a lower D-dimer level after initiation of anticoagulation (p = 0.03).

Conclusion The large proportion of acceptable outcome in survivors, which continue to functionally improve after 1 year, motivates the hospitalization in ICU for severe CVT. Persistence of fibrinolysis and activation of coagulation and difficulty to reach efficient anticoagulation during the initial phase might lead to a poor outcome.

Competing interests None.

O21 Brain lesion spectrum characterization in an experimental model of infective endocarditis

Correspondence: Marie Cantier -

Annals of Intensive Care 2016, 6(Suppl 1):O21

Introduction Infective endocarditis (IE) is a severe disease, associated with a high mortality. Neurological events are the most frequent complications of left-sided endocarditis, occurring in 20–40 % of patients. Embolic events from vegetations are commonly accepted as the main mechanism responsible for brain lesions, but other hypotheses were highlighted, including a potential subacute cerebral vasculitis. The main objective of the present study was to define the cerebral lesion spectrum in an IE rat model.

Materials and methods Wistar rats underwent an experimental aortic Staphylococcus aureus or Enterococcus faecalis IE. During a surgical procedure, a catheter was inserted into the left ventricle via the right carotid artery. Twenty-four hours later, animals were inoculated with bacterial strain. A systematic brain magnetic resonance imaging (MRI) was performed 2 days after bacterial inoculation. After killing, brains and vegetations were removed. Brains were stained with Masson’s trichrome, hematoxylin and eosin, diaminobenzidine-enhanced Perls, and Gram stainings. Immunohistochemistry was performed on brains, including 4′,6′-diamidino-2-phenylindole, myeloperoxidase, glial fibrillary acidic protein, and ionized calcium-binding adapter molecule 1 immunostaining. Finally, DNA extraction and polymerase chain reaction (PCR) were performed on the brains and vegetations for specific bacterial DNA amplification. Three other animals groups underwent the same analyses and served as controls: rats of the sham A group were only inoculated with S. aureus or E. faecalis without any surgical procedure; rats of the sham B group underwent the surgical procedure without bacterial inoculation; and rats of the control group were free of any procedure.

Results Cerebrovascular lesions were detectable on MRI in 12/23 (52 %) animals with IE. Lesions included mostly cerebral micro-bleeds (n = 8, 35 %) and brain infarcts (n = 9, 39 %). We observed no difference related to the involved microorganism. None of the sham A and control rats revealed MRI lesions. In contrast, the radiological lesion type was similar in the sham B and IE rats. Neuropathological findings showed a cerebral lesion load larger than the one observed on brain MRI. In IE rats, a scattered brain injury pattern was documented made of brain infarcts (n = 12, 52 %), micro-hemorrhages (n = 10, 44 %), and micro-abscesses (n = 11, 48 %), associated with a patchy inflammatory perivascular and small vessel wall infiltrate. Reactional gliosis was present around the lesion areas. The IE rats had significantly higher pathological lesions compared with the sham A rats. In contrast, no significant difference was observed between the IE and sham B groups, except for microglial activation, higher in the IE group (p = 0.02). Of note, bacterial studies including Gram staining and PCR for bacterial DNA in brains were inconclusive.

Discussion We report here the first rat model of brain injury in the setting of IE. Brain lesion spectrum was relevant to clinical findings in IE. We did not find strong arguments in favor of septic embolism. Moreover, lesion spectrum found is similar to those observed in cerebral vasculitis in human. Therefore, embolism from vegetations may not be the unique mechanism involved in IE-related brain injury. We found similar lesion type in IE and sham B groups, but higher microglial activation in the IE group. Taken together, these results suggest that sepsis and valvular lesion are synergistic conditions involved in brain injury. In this hypothesis, sepsis appears to amplify observed brain injury and reactional gliosis. Finally, this model appears to be adapted for cerebral micro-bleeds detection and analysis, whose physiopathology needs to be further investigated.

Conclusion In this rat model of IE, we characterized brain injury as scattered infarcts, micro-abscesses, micro-bleeds, and small vessel wall inflammation. This lesion spectrum probably results from a combination of thromboembolism and cerebral vasculitis, promoted by additional sepsis and systemic inflammation.

Competing interests None.

O22 Outcome of patients with Parkinson’s disease admitted in intensive care unit

Correspondence: Omar Ben Hadj Salem -

Annals of Intensive Care 2016, 6(Suppl 1):O22

Introduction Parkinson’s disease is associated with an increased risk of severe septic complications that can lead to intensive care unit (ICU) admissions. Despite this, no prognosis data for Parkinson’s disease (PD) patients in ICU are available.

Patients and methods The objective was to describe PD patients characteristics admitted to ICU and assess mortality and prognosis at the end of hospitalisation and up to 18 months after discharge. To this purpose, data on 62 PD patients admitted between 1999 and 2014 in two university hospital ICUs have been analysed.

Results The admitted patients were of variable age profiles, duration of evolution and severity of Parkinson’s disease. The admission reason was sepsis for 59 % of patients (n = 61), and 85 % had at least one sepsis during their hospitalisation (n = 60). Mortality was 12.5 % (n = 55) after the ICU stay and tended to increase over the 18 months of follow-up. A Hoehn and Yahr score >3 was associated with an extra mortality at M1 (OR 13.22; 95 % CI [1.4–124.9]). The HY average score increased at an early stage from M1 (2.95–3.9; p = 0.031) (n = 21) and lasting until M18 (2.7–3.6; p = 0.049) (n = 11). The proportion of subjects with cognitive disorders increased to reach over 30 % at M18. Thirty-eight percentage of patients had returned home at M18.

Conclusion The ICU hospitalisation, mostly for a sepsis, marks a turning point in the development of Parkinson’s disease. Broader prospective studies are required to assess more accurately the impact of sepsis on Parkinson’s disease evolution.

Competing interests None.

O23 Cerebrospinal fluid glucose and lactate concentrations after subarachnoid hemorrhage

Correspondence: Fabio Silvio Taccone -

Annals of Intensive Care 2016, 6(Suppl 1):O23

Introduction In patients with subarachnoid hemorrhage (SAH), the monitoring of brain metabolism through microdialysis catheters showed a decrease in brain glucose concentrations and an increase in lactate levels as a marker of cellular distress. However, the use of such catheters is not routinely applied. No data are available on glucose and lactate concentrations into the cerebrospinal fluid (CSF) of such patients.

Patients and methods We performed a retrospective analysis of patients admitted after a non-traumatic SAH from January 2011 to December 2014. Inclusion criteria were: (a) age >18 years; (b) the presence of an external ventricular drain (EVD) for intracranial pressure monitoring; (c) daily analysis of CSF including glucose and lactate concentrations for at least four consecutive days; and (d) concomitant analysis of glucose and lactate concentrations on the arterial blood gas analysis. Demographics and clinical characteristics were recorded on admission as the development of complications and 3-month neurological outcome (favorable outcome was defined as a Glasgow outcome scale of 4–5).

Results A total of 223 patients with SAH were admitted over the study period; among them, 131 needed a DVE placement and 54 met the inclusion criteria for the final analysis (median age: 57 [49–64] years; male gender: 28/54). A DVE was placed in most of the patients (41/54) on the day of hospital admission and was maintained for a period of 12 [9–16] days. Of the 48 identified aneurysms, 44 were treated with endovascular therapy. Median Glasgow coma scale on admission was 12 [8–14], and CT scan Fisher’s scale was 4 [3–4]. A total of 16 patients developed a delayed neurological ischemic deficit (DNID); ICU mortality was 30 % (16/54), and 20 (37 %) patients had a favorable neurological outcome. Despite a similar CSF red blood cells count, CSF and blood glucose levels than survivors, non-survivors patients had higher blood (1.9 [1.6–2.3] vs. 1.2 [1.0–1.9] mEq/L; p < 0.009) and CSF lactate levels (4.6 [3.6–5.8] vs. 3.2 [2.3–4.1] mEq/L; p = 0.002) and a lower CSF glucose/lactate ratio (21.5 [15.1–25.1] vs. 28.1 [20.9–35.1]; p = 0.01) on the first CSF analysis. Similar results were also found when patients with poor neurological outcome were compared with others.

Conclusion Patients with poor outcome after SAH demonstrated higher CSF lactate levels and a lower CSF glucose/lactate ratio than patients with favorable outcome on the first day of DVE placement. These CSF findings may suggest a more pronounced metabolic brain distress in these patients or a need for an alternative energetic substrate to glucose.

Competing interests None.

O24 Spontaneous angionegative subarachnoidal hemorrhage: neurological outcome based on a retrospective study of 68 patients

Correspondence: Alexis Soummer -

Annals of Intensive Care 2016, 6(Suppl 1):O24

Introduction Non-aneurysmal angionegative subarachnoid hemorrhage (NA-SAH), which concerns about 15 % of patients, is known to be associated with a good prognosis [1, 2]. Thus, systematic prolonged clinical observation in neuro-ICU might not be necessary, and to date, no recommendation exists about this point. The aim of our study was to describe the incidence of neurological complications in NA-SAH and try to identify associated risk factors.

Patients and methods We performed a retrospective analysis of NA-SAH patients in two university-affiliated hospital from 2008 January to 2015 September. NA-SAH was confirmed in all patients with a negative catheter cerebral angiography.

Usual clinical and demographic data and pattern of bleeding on CT scan (perimesencephalic, non-perimesencephalic, Fischer’s score) on admission were recorded. During ICU stay, occurrence of neurological and medical complications was analyzed: rebleeding, hydrocephalus, requirement of external ventricular drain (EVD), symptomatic or not cerebral vasospasm, delayed ischemic deficit (DID), dysnatremia (>145 or <135 mmol/L), cardiac and respiratory complications. Follow-up lasted until hospital discharge, and clinical status was described with Glasgow outcome scale (GOS). Results were presented as mean ± SD or median (first interquartile–third interquartile), and we compared patients with and without complications. Statistical analysis was performed with SPSS®.

Results Sixty-eight patients were included. Mean age was 51 ± 9, SAPS 2 was 17 (15–20), 96 % were WFNS grades 1–2, and one patient was grade 4. Bleeding was strictly limited to perimesencephalic spaces in 32 (47 %) patients, and 36 (53 %) had a non-perimesencephalic pattern.

22 (33 %) patients developed at least one complication. Nine patients had hydrocephalus, and five of them required EVD. Seven had non-symptomatic vasospasm, and no patient developed DID. GOS at discharge was good in all cases.

Factors associated with occurrence of complication were non-perimesencephalic pattern compared with perimesencephalic (48 vs 18 %, p = 0.01), intraventricular hemorrhage (54 vs 23 %, p = 0.01). In perimesencephalic pattern, no patient experienced serious complication: one patient developed hydrocephalus that did not require EVD, one patient had an asymptomatic vasospasm on CT scan, and four patients had mild dysnatremia. Besides neurological complication, patients often needed multimodal analgesia including opioids to control headaches and nausea/vomiting symptoms.

Conclusion Our retrospective analysis confirms previously published results [1, 2] that patients presented a very low rate of complication. Serious complication occurred in <10 % of patients and consisted in requirement of EVD. Perimesencephalic was never associated with early or late serious complication.

Besides, it is admitted that repetition of brain imaging in non-perimenscephalic SAH gets an average 10 % yield for detection of unrecognized vascular abnormality [1, 2], but we do not confirm this in our population.

Further study on larger cohort might lead to a precision of management in patients with a perimesencephalic pattern including a proposition of a short observation in ICU before discharge to the opposite of non-perimesencephalic patients that require a close observation due to the risk of early and late neurological complication.

Competing interests None.


1. Dalyai R et al., Subarachnoid hemorrhage with negative initial catheter angiography: a review of 254 cases evaluating patients clinical outcome and efficacy of short- long-term repeat angiography. Neurosurgery. 2013 Apr; 72(4): 646–652.

2. Maslehaty H et al., Special features of subarachnoid hemorrhage of unknown origin: a review of a series of 179 cases. Neurol Res. 2012 Jan; 34(1):91–97.

O25 Serious game evaluation of a one-hour training basic life support session for secondary school students: new tools for future bystanders

Correspondence: Stéphane Gaudry -

Annals of Intensive Care 2016, 6(Suppl 1):O25

Introduction Cardiopulmonary resuscitation and immediate activation of emergency medical services are crucial determinants of survival after out-of-hospital cardiac arrest. All citizens should then be taught how to call for help and to perform cardiopulmonary resuscitation. Teaching basic life support to secondary school students has been proved feasible, but not much evaluated. We conducted a study to evaluate the efficacy of a 1-h basic life support training course for secondary school students.

Materials and methods From September 2014 to June 2015, six secondary schools (in two neighboring towns) were included in the study. Two of them received a 1-h basic life support session conducted by an intensivist assisted by three medical students. Our work focused on youngest students (first and second year of secondary school). Sessions contained: a theoretical lecture and a practical simulation-based training (chest compression, use of automated external defibrillators) using low-fidelity manikins (MiniAnnePlus®, Laerdal). Between 2 and 3 months later, both trained and untrained students were assessed using a serious game reproducing real-life cardiac arrest situation (3D real-time simulation software, Stayingalive®, ILUMENS-Dassault system). Primary outcome was the time spent to complete the serious game (from onset of cardiac arrest to shock delivery and recovery). Secondary outcomes included knowledge of emergency call (15, 112 or 18), hand placement during chest compressions and electrodes placement of automated external defibrillator.

Results Among 199 students (none of them have ever played on Stayingalive® Web site) assessed on the serious game, 72 had received the 1-h basic life support session. Children’s ages did not differ between the two groups (12 [11–12] vs 12 [11–12] years old, p = 0.7).

The total game completion time was 169 (±28) s in the trained group versus 185 (±32) s in the non-trained group (p = 0.0006; see Fig. 6).

Fig. 6
figure 6

See text for description

Knowledge of emergency call did not differ between the two groups (92 % in trained group vs 86 % in non-trained group, p = 0.2). Hands were more often properly placed during chest compression in the trained group (92 vs 73 %, p = 0.002). Electrodes of automated external defibrillator were more often properly placed in the trained group (44 vs 10 %, p = 0.0001).

Conclusion A 1-h basic life support training session with low-fidelity manikin improved significantly young students’ performance during a 3D real-life simulation session. This study forms the basis of a large educational health program in Hauts-de-Seine department in France.

Competing interests Manikins were provided by Laerdal Medical without any compensation.

O26 Refractory out-of-hospital refractory cardiac arrest treated by veno-arterial extracorporeal membrane oxygenation. A retrospective single-center experience from 2012 to 2015. CARECMO program (Cardiac ARrest Extra Corporeal Membrane Oxygenation)

Correspondence: Antoine Kimmoun -

Annals of Intensive Care 2016, 6(Suppl 1):O26

Introduction There is a growing interest in emergency extracorporeal life support for the management of out-of-hospital refractory cardiac arrest (OHRCA). However, preliminary results are discrepancy and many questions remain unresolved. We describe here the preliminary results of our CARECMO program in Nancy, France.

Patients and methods We evaluated retrospectively our monocenter experience between September 2012 and September 2015. Patients were eligible to receive a vaECMO if they presented an out-of-hospital cardiac arrest respecting the CARECMO criteria: (1) witnessed cardiac arrest; (2) no flow <2 min; (3) patient located at <15 min to the ECMO center; (3) no ROSC despite advanced cardiac life support (ACLS) provided by the medical mobile team unit; (4) first call to the ECMO center within the 15 min after the beginning of ACLS; and (5) transfer to the ECMO center under mechanical chest compressor.

Results Thirty-two patients presented a OHRCA respecting CARECMO criteria. Thirteen out of 32 patients had prior cardiovascular history, while 8/32 had no prior medical history. Median age was 43 [IQR 34–48] and sex ratio F/M 1:4. Surgical ECMO implantation was successful in 32 patients (100 %). Immediate coronarography was performed in 28/32 (88 %) patients, and at least one coronary artery was revascularized in 20/28 (71 %) patients. Eight out of 28 (28 %) patients had no coronary artery disease, among which none survived. A total of 11/32 (34 %) patients were weaned from vaECMO, and 8/32 (25 %) were discharged from intensive care unit. In survivors, median length of stay in ICU was 30 days [10–43]. All of these patients were alive at 6 months, and seven of them had a Glasgow outcome score at 5. Initial cardiac rhythm, prothrombin time within the 4 h following admission (survivors 48 % [IQR 39–76] vs. non-survivors 41 % [IQR 37–55], p = 0.29) and duration of low flow (survivors 68 [IQR 56–76] vs. non-survivors 76 min [IQR 67–93], p = 0.16) was not associated with the outcome. No difference was found between survivors and non-survivors for pH (7.27 [7.12–7.32] vs. 7.18 [7.12–7.30], p = 0.5) and lactate (7.2 [3.4–13] vs. 8.4 [4.5–14], p = 0.6) at admission in ICU.

Conclusion Following rigorous inclusion criteria, ECMO should be considered in out-of-hospital refractory cardiac arrest.

Competing interests None.

O27 Decreased monocyte HLA-DR expression after out-of-hospital cardiac arrest

Correspondence: Laurent Argaud -

Annals of Intensive Care 2016, 6(Suppl 1):O27

Introduction Post-cardiac arrest syndrome leads to a systemic inflammatory response that shares many similarities with the one observed in sepsis. As a decrease in monocyte HLA-DR expression (mHLA-DR) has been repeatedly described in septic patients with an increase risk of death, we tested whether this immune alteration could also be observed after out-of-hospital cardiac arrest (OHCA).

Patients and methods Fifty-five OHCA patients sampled at day 0 (D0: within 4 h after OHCA), D1 (the next day), and D3 (after two additional days) were included. CD4 + lymphocyte count and mHLA-DR were evaluated by flow cytometry.

Results We observed a marked decrease in mHLA-DR as early as D0 in patients compared with normal values. This decrease persisted till D3 and was associated with a moderate decrease in the number of CD4+ lymphocytes. No correlations were identified between mHLA-DR and both usual prognostic markers after OHCA and sequential organ failure assessment (SOFA) score. However, overtime evolution in mHLA-DR values appeared different between survivors and non-survivors with a quasi-systematic decrease between D1 and D3 in non-survivors versus an increase in survivors.

Conclusion This preliminary pilot study describes the occurrence of OHCA-induced immune alterations as illustrated by a decreased mHLA-DR and CD4+ lymphopenia. CD4+ count and mHLA-DR could be simple tools to assess the systemic inflammatory response after OHCA.

Competing interests None.

O28 Is emergent PCI associated with a clinical benefit in post-cardiac arrest patients without ST-segment elevation pattern? Insights from the PROCAT II registry

Correspondence: Florence Dumas -

Annals of Intensive Care 2016, 6(Suppl 1):O28

Introduction Emergent coronary angiogram and reperfusion are currently a standard of care in patients resuscitated from an out-of-hospital cardiac arrest (OHCA) with ST-segment elevation (STE). However, using a similar invasive strategy is still debated in patients without ST-segment elevation (non-STE). In a large cohort of OHCA patients without STE, we assessed the relationship between the use of an early invasive strategy and patient outcome.

Materials and methods In the absence of an obvious extra-cardiac cause, since many years our practice is to perform an emergent coronary angiogram (CAG) in all OHCA patients (STE and non-STE) at admission, followed by percutaneous coronary intervention (PCI) when required. All patients’ characteristics are prospectively collected in a database (PROCAT). Focusing on non-STE patients and using logistical regression, we investigated the association between early PCI and favorable outcome (CPC1–2 at discharge), and we searched predictive factors for PCI requirement.

Results During the study period (2004–2013), we investigated 958 OHCA patients with an emergent CAG. Among them, 695/958 (73 %), mostly male (76 %) and average age of 60 years, had no evidence of STE on the post-resuscitation ECG. A PCI was deemed necessary in 199/695 (29 %). A favorable outcome was observed in 87/200 (43 %) in patients with PCI compared with 164/495 (33 %) in patients without PCI (p = 0.02). After adjustment, PCI was associated with a better outcome (ORadj = 1.80 [1.09–2.97], p = 0.02). The other predictive factors were a shorter resuscitation length (<20 min), an initial shockable rhythm and a lower dose of epinephrine (p < 0.001). An initial shockable rhythm (ORadj = 2.83 [1.84–4.36], p < 0.001) was an independent indicator for PCI requirement.

Conclusion A culprit coronary lesion requiring PCI was found in nearly one-third of OHCA patients without STE. In these patients, emergent PCI was associated with a nearly twofold increase in the rate of favorable outcome. These findings support the use of an invasive strategy in these patients, particularly in those resuscitated from a shockable rhythm.

Competing interests None.

O29 Predictors of long-term quality of life after cardiac arrest: insights from the Parisian registry

Correspondence: Guillaume Geri -

Annals of Intensive Care 2016, 6(Suppl 1):O29

Introduction Determinants of health-related quality of life (HRQOL) are poorly known in patients resuscitated from out-of-hospital cardiac arrest (OHCA).We aimed at comparing long-term HRQOL between OHCA survivors and controls from the general population and to assess factors associated with HRQOL.

Patients and methods HRQOL was evaluated by telephone interview using SF-36 questionnaire in a large cohort of OHCA survivors. As a comparison group, each OHCA case was age- and gender-matched with four controls from the French general population. Association between pre- and in-hospital Utstein variables and early interventions with the different dimensions of the SF-36 questionnaire was investigated using MANCOVA.

Results During the study period (2000–2013), 1829 OHCA patients were admitted in our unit and 602 discharged alive. SF-36 interview took place 50 months [min, max: 22–93] after cardiac arrest in 255 patients (137 patients died and 210 were lost of follow-up) with a median age of 55 years [45, 64] and including 73.7 % of males. Global physical and mental components did not differ between OHCA survivors and controls (47.0 vs. 47.1, p = 0.88 and 46.4 vs. 46.9, p = 0.45). Physical functioning (74.1 vs. 78.4, p = 0.02) and vitality (50.7 vs. 56.2, p < 0.01) were significantly altered in OHCA survivors. Younger age, male gender and initial shockable rhythm were associated with almost all SF-36 dimensions. Early coronary reperfusion was associated with better physical functioning (+7.0, p = 0.06), general health (+7.3, p = 0.02) and vitality (+4.4, p = 0.08).

Conclusion HRQOL of OHCA survivors appeared similar to that of the general population, but some physical components of SF36 scale were significantly altered. Younger age, male gender, initial shockable rhythm and early coronary reperfusion were independently associated with a better HRQOL.

Competing interests None.

O30 Red blood cell transfusions in early resuscitation of severe sepsis and septic shock in patients with hematological malignancies

Correspondence: Adrien Mirouse -

Annals of Intensive Care 2016, 6(Suppl 1):O30

Introduction Severe sepsis and septic shock are common complications for patients with hematological malignancies. Anemia is a frequent underlying condition in such patients who are then likely to receive red blood cell transfusion as part of early hemodynamic resuscitation. However, the optimal threshold and the eventual benefit of red blood cell transfusion in septic acute circulatory failure remain debated. In this study, we investigated the indications of red blood cell transfusion and its impact on outcome in a large cohort of patients with hematological malignancies who were admitted to the ICU for the main diagnosis of severe sepsis and septic shock.

Patients and methods TRIAL-OH is a prospective, multicenter observational study that included 1011 patients with hematological malignancies who required ICU admission in 2010–2011 in 17 French and Belgian centers (1). We focused on patients admitted in ICU with a definite diagnosis of severe sepsis or septic shock according to the Surviving Sepsis Campaign definitions. We addressed the requirements for red blood cell transfusion during the first 48 h as part of initial resuscitation. Continuous variables were expressed as median [interquartile range]. Endpoints were vital status at 7 days, ICU and hospital discharge. Multivariate logistic regression was used to assess the impact of transfusion on mortality.

Results Among the 1011 patients of the whole TRIAL-OH cohort, 631 (62.4 %) were admitted to the ICU for severe sepsis (55 %) or septic shock (45 %). Patients were mostly males (64.3 %) with an age of 60 [49–70] years. The admission SOFA score was 7 [4–10]. The hemoglobin level at the time of ICU admission was 9.1 [7.9–10.4] g/dL. One hundred and forty-eight (23.5 %) and 89 (14.5 %) patients received red blood cell transfusion at days 1 and 2, respectively. Overall, 210 (33.3 %) patients received 2 [1–3] packed red cells during the first 48 h. Hemoglobin levels were lower in transfused patients at day 1 (7.8 [7.1–8.8] vs. 9.7 [8.7–11] g/dL, p < 0.001) and day 2 (8.9 [7.9–9.9] vs. 9.3 [8.4–10.2] g/dL, p = 0.001) and became similar to those of non-transfused patients at day 3 (8.9 [8.2–9.8] vs. 8.9 [8–10]). Early red blood cell transfusion was more likely in patients with myeloid neoplasms and neutropenia. Transfused patients appeared with more severe presentations as assessed by higher admission SOFA scores (8 [5–11] vs. 6 [4–9], p < 0.001) and blood lactate levels (2.4 [1.6–5.6] vs. 1.9 [1.1–3.6] mmol/L, p < 0.001) and the requirements for organ failure supports including vasopressive drugs (78.9 vs. 49.7 %, p < 0.001), invasive mechanical ventilation (75.5 vs. 33.5 %, p < 0.001) and renal replacement therapy (38.7 vs. 15.4 %, p < 0.001). In univariate analysis, red blood cell transfusion within the first 48 h was associated with increased mortality at day 7 (20.5 vs. 13.3 %, p = 0.02), in the ICU (39 vs. 25.2 %, p < 0.001) and in the hospital (51 vs. 36.6 %, p < 0.001). After adjustment with other significant variables, red blood cell transfusion was no longer associated with day 7 mortality (OR 1.18 [0.78–1.98], p = 0.52), but tended to be associated with in-ICU and in-hospital mortality (OR 1.48 [0.96–2.26], p = 0.07 and OR 1.54 [1.04–2.29], p = 0.03, respectively).

Discussion Although early red blood transfusion was provided to the most severe septic patients, it remained associated with a poor prognosis in the multivariate analysis. Whether transfusion may represent a potent marker of severity in this setting or may favor subsequent side effects in septic patients deserves further investigations.

Conclusion Red blood cell transfusion is commonly used in early resuscitation of hematological patients with severe sepsis, and particularly in those with septic shock. The poor prognostic value was associated with transfusion calls for prospective studies in this high-risk subgroup of patients.

Competing interests None.


1. Azoulay et al. J Clin Oncol 2013; 31(22): 2810–8.

O31 Causal effect of transfusion on mortality and other adverse events among critically ill septic patients: an observational study with a marginal structural model

Correspondence: Claire Dupuis -

Annals of Intensive Care 2016, 6(Suppl 1):O31

Introduction Septic patients tend to be anemic with limited tolerance in intensive care unit. Red blood cell transfusion (RBCT) is the main treatment but can be deleterious. In that context, one RCT(1) concluded that 7 g/dl is a good target but excluded patients with overt bleeding or low hemoglobin level during the day of admission and showed significant increase in ischemic events. Moreover, many cohort studies estimated the impact of RBCT, but reached opposite results. One of their main limits was they did not handle time-dependent covariates and past history of treatment. In that context, the aim of our study was to estimate the causal effect on outcomes and adverse events of RBCT in septic shock patients thanks to a marginal structural model.

Patients and methods We used the prospective French multicentric Outcomerea database. Four outcomes were assessed: death, the occurrence of nosocomial infection, ischemic event and acute respiratory distress syndrome. In each analysis, we included patients with sepsis severe and septic shock since 04/01/1998 and excluded patients with hemorrhagic shock on admission. Only patients staying for more than 48 h were analyzed to estimate the risk of nosocomial infection. Similarly, patients having ischemic events during the first 48 h were not included in the estimation of the risk of ischemic event. A marginal structural cox model with ponderation on the inverse probability of treatment weight (IPTW) and inverse probability of censoring weight (IPCW) was then used to estimate the impact of transfusion with a censoring at day 30. Subgroups with similar hematocrit evolution pattern were determined thanks to a trajectory modeling. Impact of transfusion on death was assessed among those subgroups.

Results Finally, 6016 patients were included, and 1940 (32.2 %) of them were transfused during their ICU stay and received 3 [2–5] (median [IQR]) red cell pack per patient transfused per ICU stay. RBCT was not associated with death [HR 0.8, 95 % confidence interval (CI) 0.544–1.176, p = 0.255], with the occurrence of ischemic event (HR 0.774, 95 % CI, 0.343–1.749, p = 0.538) or with the occurrence of ARDS (HR 0.948, 95 % CI, 0.811–1.107, p = 0.498). However, RBCT was associated with the occurrence of nosocomial infections (HR 2.149, 95 % CI, 1.662–2.779, p < 0.001). Four subgroups were identified thanks to our trajectory modeling. In the subgroup with the lowest hematocrit level (hematocrit 26 [24–28]), transfusion had a beneficial effect (HR 0.375, 95 % CI 0.225–0.625, p < 0.001), whereas in the subgroup with a hematocrit at 30 [28–31.2], transfusion had no more impact on outcome (HR 0.942, 95 % CI 0.601–1.477, p = 0.795).

Conclusion Transfusion was not associated with death in the entire cohort. Modification of transfusion practice such as systematic deleukocytation and lower transfusion thresholds might have improve the tolerance of transfusion. The protective effect in the subgroup with the lower hematocrit level might reflect the limits of a drastic lowering of the transfusion thresholds. RBCT still remained associated with nosocomial infections. Others immunomodulating factors than leukocytes might explain this association.

Competing interests None.


1. Holst LB, Haase N, Wetterslev J, Wernerman J, Guttormsen AB, Karlsson S, et al. Lower versus Higher Hemoglobin Threshold for Transfusion in Septic Shock. N Engl J Med. 2014 Oct 9;371(15):1381–91.

O32 Autoimmune hemolytic anemia in the intensive care unit

Correspondence: Antoine Lafarge -

Annals of Intensive Care 2016, 6(Suppl 1):O32

Introduction Autoimmune hemolytic anemia (AIHA) is a rare, heterogeneous and potentially life-threatening disease. Data on AIHA patients requiring admission to the intensive care unit (ICU) are scarce. The aim of our study was to describe the characteristics and outcome of patients with AIHA in the ICU.

Patients and methods Patients with AIHA admitted to the ICU of Saint Louis University Hospital between years 2002 and 2015 were retrospectively identified using a computerized database. Clinical and biological data including past medical history, etiologic diagnosis, type of direct antiglobulin test (DAT), organ failures, treatments and follow-up were abstracted from the patients’ charts. Primary outcome was ICU mortality. Data are presented as numbers (%) or medians (interquartile range). Survivors (S) and decedents (D) were compared using Chi squared test for binary variables and Mann–Whitney test for continuous variables.

Results During the study period, 270 patients with hemolytic anemia required ICU admission, including 44 (16.3 %) patients with AIHA. Twenty patients presented with a DAT positive for immunoglobulin G (IgG) only (45.5 %), eight with a DAT positive for both IgG and complement (36.3 %) and 16 with a DAT positive for complement only (18.1 %). There were 24 men (54.5 %) with an age of 57 years (44.5–69). Twenty-five patients (56.8 %) had a past history of malignant hemopathy, five (11.4 %) had solid cancer, and five (11.4 %) had HIV infection. Median delay from first symptoms of AIHA to hospitalization was 14 days (2–38). Patients were mostly hospitalized for hematological (23 [52.3 %]) or respiratory (11 [25 %]) failure. Median sequential organ failure assessment score (SOFA) at day 1 was 6 (3.8–8). Hemoglobin rate at admission was 6.8 g/dl (4.6–9.1). Corticosteroids were administered to 39 patients (88.6 %) and anti-CD20 monoclonal antibody rituximab to 11 patients (25.6 %). ICU stay was 2.5 days (2–6), and hospital stay was 15 days (9–42). Follow-up was 74 days (11.5–695.2).

Thirteen patients (29.5 %) died in the ICU, and none of the survivors died during post-ICU hospitalization. There was no difference between S and D regarding associated comorbidities like hemopathy (18 [58.1 %] vs 7 [53.8 %], p = 0.80), solid cancer (3 [9.7 %] vs 2 [15.4 %], p = 0.59) or HIV (4 [12.9 %] vs 1 [7.7 %], p = 0.62). In D, age was higher (72 years old [57.8–76.3] vs 50 years old [34.3–64], p < 0.01) and organ dysfunctions were more severe at day 1 (SOFA 8 [7–11] vs 5.5 [3–7], p < 0.01). Patients with a DAT positive for IgG only (DAT-IgG) displayed poorer outcome in comparison with patients with a Coombs test positive for IgG+ complement or complement only (DAT-IgG + C/C) (ICU mortality 69.2 vs 35.5 %, p = 0.04). In comparison with patients with DAT-IgG + C/C, DAT-IgG patients were older (65 years old [52–73.5] vs 49 years old [34.5–59.5], p = 0.01), presented with higher rates of hemoglobin at admission (8.3 [6.7–9.8] vs 5.5 [3.7–7.5], p < 0.01) and were less likely to receive rituximab in the ICU (2 [10 %] vs 9 [39.1 %], p = 0.03).

Discussion Mortality rate of AIHA patients requiring ICU admission is higher than previously reported and appears to be impacted by age, organ failures and DAT-IgG. Those results are tempered by evident recruitment biases (over-representation of hemopathies, paucity of cold hemagglutinin disease), confounding factors (age, rituximab) and the small sample size. Further larger studies are needed to confirm DAT-IgG as an independent prognostic factor and explain its pathophysiology.

Conclusion The mortality rate of AIHA patients in ICU is consequential. Besides organ failures, IgG-positive DAT may be a substantial prognostic factor.

Competing interests None.

O33 Pre-ICU location, lead time bias and outcomes in patients with thrombotic microangiopathies

Correspondence: Damien Vimpere -

Annals of Intensive Care 2016, 6(Suppl 1):O33

Introduction The thrombotic microangiopathy (TMA) syndrome encompasses a wide spectrum pathologies including thrombotic thrombocytopenic purpura (TTP) and complement-mediated TMA [formerly known as atypical hemolytic uremic syndrome (HUS)]. The aim of the present study was to describe the first clinical symptoms in critically ill TMA patients and to evaluate the impact of pre-intensive care unit (ICU) location on the patients’ outcomes.

Patients and methods Patients with TMA admitted to a 12-bed medical university hospital ICU between 2009 and 2015 were identified using a computerized database. Data concerning the pre-ICU period and clinico-biological parameters were abstracted from medical charts. Data are expressed as numbers (%) and medians (interquartile range). Comparisons were made using the Chi square test or the Mann–Whitney U test as appropriate.

Results Among the 74 TMA patients admitted in our ICU (43 [58 %] females, aged 42 [32–55] years), 51 patients (68.9 %) had TTP, 13 (17.6 %) HUS and 10 (13.5 %) other TMA. At ICU admission, hemoglobin rate was 8.2 g/dl (6.8–9.7), platelet rate 17 G/l (9–31), LDH level 1743 U/l (1064–2581) and creatininemia 113.5 μmol/l (75–184), and 43 patients (64.2 %) displayed elevated troponin levels. Day 1 SAPS2 score was 21.5 (12–32) and SOFA score was 6 (4–8). ICU length of stay was 7 days (4–11), and hospital length of stay was 16.5 days (12.5–28.5). Five patients (9.4 %) died during hospital stay. Earlier TMA manifestation was neurologic in 25 cases (33.8 %), digestive in 20 cases (27 %), uro-nephrologic in 14 cases (18.9 %), dermatologic in 12 cases (16.2 %), cardiologic in ten cases (13.5 %) and respiratory in nine cases (12.2 %). Sixteen patients (21.6 %) reported simultaneous occurrence of more tha one first symptom. The occurrence of diarrhea or abdominal pains as first symptoms of TMA was more frequent in HUS patients in comparison with TTP and other TMA patients (respectively, 8 [61.5 %] vs 11 [21.6 %] vs 1 [10 %], p < 0.001). Time from first clinical symptom to first medical contact was 4 days (0–8.5). This time was longer when the first alleged symptom was dyspnea (14 days [10.5–18.5] vs 3.5 days [0–7], p = 0.03). First medical contact consisted in hospital consultation in 58 cases (78.4 %) including 45 emergency department (ED) consultations, six direct admissions to specialized ICUs (three neurologic and three cardiologic) and three direct admissions to medico-surgical wards. In 16 (21.6 %) cases, the first medical contact was community based (15 general practice and one cardiology practice). Patients were admitted in one hospital ward (range 0–5) before medical ICU referral. Only 14 patients (18.9 %) were directly admitted to a medical ICU performing PEX, after first medical contact with an ED (11 cases) or hospital-based consultations (three cases). None of these patients died. Prior to ICU admission, 21 patients were managed in the ED (27 cases [34.5 %]), another medical ICU in nine cases (15 %) or a medico-surgical ward (24 cases [40 %]). Secondary referral to ICU where PEX was available was similar in patients first admitted to ED or medico-surgical wards (63 vs 70.8 %), but time to ICU transfer was shorter for patients admitted first to ED than for patients directly admitted to wards (1 days [0–1] vs 3 days [1.5–7], p < 0.01). Patients admitted first to another ICU were secondarily referred to our ICU for refractory TMA within 6 days (0.5–10.5).

Discussion Our data suggest that family physicians and ED specialists must maintain a high level of suspicion for the diagnosis to appropriately refer these patients to specialized ICUs. Indeed, no patient directly admitted to ICU died, suggesting that early recognition of TMA syndrome and quick transfer to specialized centers may improve TMA outcomes. However, we found no association of pre-ICU characteristics and mortality, likely because of the low case fatality rate.

Conclusion TMA patients still display relatively long pre-ICU trajectories before referral to specialized ICUs, with inappropriate referral resulting in longer delays for ICU admission in centers where PEX and specific management is available. These data need to be repeated at large scale and with outcomes such as morbidity, length of stay, unresponsive TMAs and long-term outcomes.

Competing interests None.

O34 Septic shocks with no early etiological diagnosis: a multicenter prospective cohort study (the shock 24 study)

Correspondence: Damien Contou -

Annals of Intensive Care 2016, 6(Suppl 1):O34

Introduction The lack of a patent source of infection and microbiological documentation is not uncommon during the first 24 h management of a septic shock, raising the possibility of non-infectious causes.

Patients and methods Among patients admitted in the intensive care unit (ICU) with a “clinical phenotype” of septic shock, we aimed to determine the prevalence of shock with no early etiological diagnosis, to identify their main causes and to compare their outcome to those of patients with an “early-diagnosed” septic shock. We conducted a prospective, multicenter, observational cohort study in ten ICUs of public hospitals in France.

Results From November 2014 to June 2015, 508 patients were admitted for a suspicion of septic shock. Among them, 374 (74 %) had a proven diagnosis of septic shock made during the first 24 h (“early” septic shock, ESS), while the 134 (26 %) others had no source of infection nor microbiological documentation identified within the first 24 h. These patients had either septic shock confirmed during the first 24 h (“late” septic shock, LSS; n = 37/134, 28 %) or a severe systemic inflammatory response syndrome (SIRS) (n = 97/134, 72 %) encompassing patients having a non-septic severe SIRS (NS-SIRS) (i.e., a non-infectious cause was identified; n = 59/134, 44 %) or a severe SIRS of unknown origin (SIRS-UO) (n = 38/134, 28 %). The leading causes of non-septic severe SIRS were adverse drug reactions, acute mesenteric ischemia, malignancies and inflammatory diseases. There were no differences regarding the in-ICU mortality (n = 139/374, 37 % vs. 49/134, 37 %; p = 0.90) and the median duration of ICU stay (7 vs. 7 days, p = 0.69), of tracheal intubation (4 vs. 4 days, p = 0.19) and of vasopressor support (3 vs. 3 days, p = 0.55) between patients with “early” septic shock (ESS) and others. A multivariable Cox model with adjustment on variables significantly associated with day-60 mortality showed that patients with a severe SIRS of unknown origin had a higher risk of mortality than others (Fig. 7).

Fig. 7
figure 7

See text for description

Conclusion One-quarter of the patients admitted in the ICU for a suspicion of septic shock had no infection identified 24 h after vasopressors introduction, and almost three-quarters of these had a severe SIRS that mimicked sepsis. Patient with a severe SIRS of unknown origin had a poorer outcome (Fig. 8).

Fig. 8
figure 8

See text for description

Competing interests None.


1. Brun-Buisson C, Meshaka P, Pinton P, Vallet B, (2004) EPISEPSIS: a reappraisal of the epidemiology and outcome of severe sepsis in French intensive care units. Intensive Care Med 30: 580–588.

2. Heffner AC, Horton JM, Marchick MR, Jones AE, (2010) Etiology of illness in patients with severe sepsis admitted to the hospital from the emergency department. Clin Infect Dis 50: 814–820.

O35 Respective contribution of diaphragm and limbs muscles weakness on weaning from mechanical ventilation outcome

Correspondence: Martin Dres -

Annals of Intensive Care 2016, 6(Suppl 1):O35

Introduction The role of intensive care-acquired weakness (ICU-AW) and diaphragm weakness (DW) is questioned in case of prolonged or failed weaning from mechanical ventilation. The relative contribution of DW and ICU-AW on this outcome is unclear. We aimed at assessing the respective prevalence and impact of DW and ICU-AW on the success (WS) or failure (WF) of the weaning process. We combined the use of twitch airway occlusion pressure and ultrasound to evaluate diaphragm stength.

Patients and methods In ICU patients undergoing a first spontaneous breathing trial, diaphragm function was evaluated by twitch tracheal pressure in response to bilateral anterior magnetic phrenic nerve stimulation (Ptr,stim) and by diaphragm ultrasonography [end-expiratory thickness (TdiEXPI), thickening fraction (TFdi), excursion (EXdi). MRC score below 48 defined ICU-AW and Ptr, stim below 11 cmH2O defined DD.

Results Seventy-six patients were included: age: 60 years (47–69) and duration of mechanical ventilation: 5 days (3–9). DW and ICU-AW were, respectively, present in 48 (63 %) and 26 (34 %) patients. The prevalence of DW was similar among patients with or without ICU-AW (62 vs. 64 %, p = 0.83) as well was the median MRC score in patients with or without DW [51 (43–56) vs. 51 (44–59), p = 0.30]. The proportion of patients with DW but without ICU-AW was 42 %, whereas the proportion of patients with both DW and ICU-AW was 21 % (p < 0.05). By contrast, the proportion of patients with ICU-AW but without DW was 13 %. The proportion of patients without ICU-AW neither DW was 24 %. DW was associated with WF (OR 23.7, 95 % CI 5.0–112.2, p < 0.001) but not ICU-AW (OR 1.9, 95 % CI 0.73–5.0, p = 0.23). Patients with both DW and ICU-AW had a significantly higher ICU mortality as compared to patients with ICU-AW without DW (38 vs. 0 %, p < 0.001). The areas under the receiver operating characteristic curves of Ptr, stim, TFdi and MRC score to predict weaning outcome were 0.92, 0.90 and 0.67, respectively.

Conclusion Diaphragm weakness had higher prevalence and worst impact on weaning outcome than ICU-AW. Diaphragm ultrasound predicted weaning outcome with good performance. Determination of the respective risk factor for DW and ICU-AW deserves now further studies.

Competing interests None.

O36 Impact of sleep quality on the duration of weaning from mechanical ventilation

Correspondence: Ludivine Rousseau -

Annals of Intensive Care 2016, 6(Suppl 1):O36

Introduction In difficult-to-wean ICU patients under mechanical ventilation, the influence of sleep quality on weaning duration has never been studied. We aimed to compare sleep quality between patients with a short weaning duration and those with a prolonged weaning duration.

Patients and methods Prospective physiological study performed in a French teaching hospital. All patients intubated at least 24 h and difficult-to-wean, i.e., those who experienced at least one weaning trial failure, could be included. Patients with continuous sedation, central nervous system or psychiatric disease, or peripheral neuromuscular disease were excluded. A complete polysomnography was performed as soon as possible the night following the first weaning trial failure. Peripheral muscle strength, maximal inspiratory pressure and delirium were measured at time of polysomnography. Weaning duration was defined as the time from polysomnography to extubation. Prolonged weaning was defined as a weaning duration more than or equal to 3 days (failure of at least three weaning trials) according to the international conference consensus of weaning [1].

Results Over the first 6 months of the study, 25 patients intubated for at least 24 h were considered as difficult-to-wean. Among the 15 patients included in the study, six patients (40 %) had short weaning period and nine patients (60 %) had prolonged weaning (≥3 days). Patients with prolonged weaning group had less rapid eye movement (REM) sleep episodes and shorter REM sleep duration than patients with short weaning (0.0 episodes [0.0–2.0] vs 6.0 [4.5–6.8] and 0 min [0–6] vs 43 [33–77], p = 0.02 for both). Atypical sleep and/or pathological wake [2] was more likely to be found in the prolonged weaning group: 56 % of the patients (5/9) versus 17 % (1/6) although this difference was not significant (p = 0.29). Patients with absent or altered electroencephalographic reactivity (n = 10/15) were more likely to have prolonged weaning than the others (80 % of the patients with absent/altered reactivity had prolonged weaning vs 20 % of the patients with normal reactivity, p = 0.03). Maximal inspiratory pressure was similar in the two groups as well proportion of delirium. The only difference was a lower peripheral muscular strength in the prolonged weaning group as indicated by a lower MRC score (32 [28–55] vs 59 [55–60], p = 0.02).

Conclusion Our preliminary results show that decreased REM sleep time and altered electroencephalographic reactivity measured by polysomnography seem to be associated with prolonged weaning duration in ICU patients.

Competing interests None.


1. Boles JM, Bion J, Connors A, Herridge M, Marsh B, Melot C, Pearl R, Silverman H, Stanchina M, Vieillard-Baron A, Welte T (2007) Weaning from mechanical ventilation. Eur Respir J 29:1033–1056.

2. Drouot X, Roche-Campo F, Thille AW, Cabello B, Galia F, Margarit L, d’Ortho M-P, Brochard L (2012) A new classification for sleep analysis in critically ill patients. Sleep medicine 13:7–14.

O37 Integrative ultrasound assessment of lung, cardiac and diaphragm function during a successful weaning trial predicts postextubation distress

Correspondence: Stein Silva -

Annals of Intensive Care 2016, 6(Suppl 1):O37

Introduction Establishing the correct time of weaning from mechanical ventilation (MV) is a crucial issue in the critical care practice. We hypothesize that the use of an integrative thoracic ultrasound assessment, encompassing bedside respiratory, cardiac and diaphragm sonographic data, could significantly improve weaning decision process. The aim of the present study was to prospectively compare the predictive value of an integrated thoracic ultrasound (TUS) applied in this setting, against a standard approach (CB) elaborated from clinical and biological data.

Patients and methods We prospectively recruited consecutive patients from three ICU of an university teaching hospital. TUS assessments were determined before (H0) at the end of a 60-min (H1) spontaneous breathing trial. Clinical and biological (arterial blood gas, plasma BNP) data were recorded concomitantly to create a standard predictive mode. We applied machine-learning methods to improve the accuracy of the related predictive assessments and explore the specific predictive load of each clinical, biological and ultrasonographic parameters.

Results One hundred and thirty-four patients who succeeded a SBT test were extubated and included in the study. Among these patients, 103 were definitely weaned and 31 suffered from postextubation distress. The most relevant clinical/biological parameters (CB model) were the spontaneous respiratory rate and the PaCO2. Regarding the whole ultrasound dataset (TUS model), only lung parameters (A, B1 and B2 profiles) and the echocardiographic estimation of left ventricular filling pressures (E/e) emerged as reliable predictors. Crucially, the predictive value of TUS was significantly greater than CB during both H0 and H1 assessments (Fig. 9).

Fig. 9
figure 9

See text for description

Conclusion An integrative thoracic ultrasound assessment during a successful breathing trial appears to be a most accurate predictor of weaning from MV than a standard clinical and biological approach.

Competing interests None.

O38 ETCO2 improved outcome prediction of mechanical ventilation weaning

Correspondence: Hafid Ait-Oufella -

Annals of Intensive Care 2016, 6(Suppl 1):O38

Introduction Weaning from mechanical ventilation (MV) is a crucial step in the management of intensive care patients. Weaning failure is associated with prolonged duration of MV leading to morbidities, increased mortality as well as increased hospital expenses. However, the prediction of weaning outcome remains challenging. We aimed to investigate the end-tidal CO2 (ETCO2) and hypothesized that the product “ETCO2 × Respiratory Rate,” called ETCO2 × RR, reflects alveolar ventilation and could predict MV weaning outcome.

Patients and methods We conducted aprospective observational study. All patients under MV that met the criteria for weaning as defined by the International Consensus Conference for Weaning from Mechanical Ventilation were enrolled. Patients underwent a T-piece spontaneous breath trial (SBT) for 30 min, and the usual respiratory and hemodynamic clinical parameters were recorded every 5 min. ETCO2 was recorded blindly by an independent physician using capnography. Patients ventilated through tracheostomy were excluded. SBT failure was defined as one or more of the clinical and biological criteria from International Consensus Conference. PaO2, PaCO2 and pH were determined at the beginning (i) and at the end of the SBT (f).

Results During a 10-month period, 310 SBTs were analyzed, 196 patients underwent 1 SBT, 57 did two trials and 48 did three or more. For the first SBT, we analyzed 196 first SBT with three groups: success (n = 139), hypoxemia-related failure (n = 11) and non-hypoxemia-related failure (n = 46). Mean age was 57 ± 17 years, 63 % were men, and SAPS II was 47 ± 16. Duration of mechanical ventilation was 2.0 (1.0–3.4) days before starting SBT, FiO2 was 38 ± 18 %, and PEEP level was 4 ± 1 cm H2O. After 30 min of SBT, ETCO2, respiratory rate and ETCO2 × RR were predictive of MV weaning outcome as AUC was, respectively, 71 % [75–94], 80 [74–89] and 88 % [80–93] (all, P < 0.0001). A threshold of ETCO2 × RR of 860 predicted weaning failure with a sensitivity of 78 % (CI95 % [61–88]) and a specificity of 81 % (CI95 % [72–87]). We found a significant correlation between delta PaCO2 (f–i) and delta ETCO2 × RR (f–i) (r = 0.35, P < 0.0001). During the second SBT, we analyzed 57 patients according to weaning outcome: success (n = 31), hypoxemia-related failure (n = 5) and non-hypoxemia-related failure (n = 21). After 30 min of SBT, the predictive performance of ETCO2*RR was better (AUC 84 % [73–96]), P = 0.0001) than ETCO2 alone (AUC 75 % [59–90]), P = 0.0002) or RR (68 % [52–84], P = 0.03).

Conclusion In unselected mechanically ventilated patients undergoing SBT, ETCO2*RR predicts weaning outcome.

Competing interests None.

O39 Characteristics of 150 cases of weaning-induced pulmonary oedema and effects of diuretics

Correspondence: Xavier Monnet -

* Both first authors contributed to this work

Annals of Intensive Care 2016, 6(Suppl 1):O39

Introduction Weaning-induced pulmonary oedema (WIPO) has not been investigated in large series of patients. Although diuretics are widely used to treat it, the dose that must be administered is unknown. Our aim was to describe the incidence of weaning-induced pulmonary oedema (WIPO) in a population of critically ill patients, to compare the characteristics of patients with and without WIPO and to detail the effects of diuretics on WIPO.

Patients and methods We systematically monitored all consecutive spontaneous breathing trials (SBT, 1-h trial, T-tube trial) that were performed in our unit. In the patients for whom cardiac index (CI) was monitored (PiCCO device), a passive leg raising (PLR) test was systematically performed before SBT in order to assess preload dependence. The diagnosis of WIPO was established a posteriori by four experts who based their diagnosis on clinical, echocardiographic and biological (haemoconcentration) data and on the evolution after SBTs. Experts were unaware of the effects of PLR test in the patients in whom it had been performed.

Results From April to September 2015, 150 SBTs were performed in 36 patients. SBT failed in 67 cases (35 % of all SBTs). According to the experts, the diagnosis of WIPO was positive in 44 cases (66 % of failing cases), negative in 98 cases and inconclusive in five cases. The incidence of previous COPD was 33 % in patients who experienced at least one WIPO and 0 % in patients who experienced no WIPO (p = 0.03). The incidence of previous cardiopathy, hypertension or supraventricular arrhythmia was 78 % in patients who experienced at least one WIPO and 50 % in patients who experienced no WIPO (p = 0.16). Patients with and without WIPO were similar in terms of left ventricular ejection fraction or of mitral E/E′ ratio at baseline and in terms of fluid balance of the previous day. Compared with patients without WIPO, patients with WIPO had a larger gain in weight from admission. Myocardial ischaemia was not detected during any cases WIPO. Among the 34 patients with cardiac output monitoring, the effects of PLR on CI were significantly larger in patients without than with WIPO (changes in CI 13 ± 15 vs. 5 ± 14 %, respectively, p = 0.04). This confirmed our previous finding that the presence of preload dependence is associated with a better cardiac tolerance to the SBT. All patients with WIPO received diuretics. Among the six cases with WIPO in which cardiac output was monitored, the PLR was negative. The PLR test performed after diuretics administration on the following day was positive in two of these six cases. These two cases succeeded to the next SBT. The PLR performed after diuretics administration was negative in the remaining four cases. Three of these cases experienced WIPO during the next SBT.

Conclusion In this population of critically ill patients, WIPO was responsible for two-third of weaning failures. WIPO was associated with preload independence, indicating the incapacity of the heart to adapt to the changes in loading conditions during SBT. After a failing SBT, when diuretics had changed the PLR test from negative to positive, the following SBT was very likely to succeed.

Competing interests XM and J-LT are members of the Medical Advisory Board of Maquet.

O40 Antifungal de-escalation was not associated with adverse outcome in critically ill patients treated for invasive candidiasis

Correspondence: Sébastien Bailly -

Annals of Intensive Care 2016, 6(Suppl 1):O40

Introduction Systemic antifungal therapy of invasive candidiasis needs to be initiated immediately upon clinical suspicion. Unfortunately, no diagnostic tests are available to firmly confirm or discard the diagnosis of IC in the absence of positive blood cultures or non-contaminated positive sample from a sterile site. Controversies exist about adequate time and potential harms of antifungal de-escalation in documented and suspected candidiasis in intensive care unit patients. Our objective was to investigate whether de-escalation within 5 days of antifungal initiation is associated with an increase of the 28-day death in systemic antifungal therapy treated non-neutropenic adult intensive care unit patients.

Patients and methods From the 835 non-neutropenic adults recruited in a multicenter prospective observational study, we selected the patients receiving systemic antifungal therapy for a documented or suspected invasive candidiasis in the intensive care units and who were still alive 5 days after systemic antifungal therapy initiation. They were included into two groups according to the occurrence of observed systemic antifungal therapy de-escalation before day 6. The average causal systemic antifungal therapy de-escalation effect on 28-day death was evaluated by using a double-robust inverse probability of treatment weight estimator which is a causal inference method based on observational data. The objective of this estimator is to balance the distribution of baseline confounders across de-escalation groups, in order to reach the condition of a randomized controlled trial.

Results Among the 647 included patients, early de-escalation at day 5 after antifungal initiation occurred in 142 patients (22 %), including 48 (34 %) patients whose systemic antifungal therapy was stopped before day 6. Patients in the de-escalation group were younger and had a shorter previous intensive care unit stay, but their SAPSII or SOFA score at intensive care unit admission was similar. The rate of proven invasive candidiasis was not different between de-escalation and no de-escalation groups. After adjustment on the baseline confounders, early systemic antifungal therapy de-escalation was not associated with increased 28-day mortality (RR 1.12, 95 % CI [0.76–1.66]). Subgroup analyses did not show any effect of early de-escalation on mortality, and there was no effect of early stopping on morality either (RR 0.98 [0.71–1.35]) (Fig. 10). Finally, the number of days alive without SAT at day 28 was higher in the de-escalation group (14 days [5; 23]) than in the no de-escalation group (10 days [2; 17] p < 0.01) leading to a median cost difference of 1100 euros in favor of de-escalation group.

Fig. 10
figure 10

See text for description

Conclusion Our causal analysis based on a large prospective observational multicenter study showed that systemic antifungal de-escalation in case of suspected or documented IC in non-neutropenic ICU patients occurred in only 22 % of the cases. In non-neutropenic critically ill adult patients with documented or suspected invasive candidiasis, SAT de-escalation within 5 days was not related to increased day-28 mortality, but it was associated with a subsequent and significant decrease in the antifungal consumption. These results remained valid on different subgroups and were confirmed by sensitivity analysis. Early stop in the absence of proven invasive candidiasis or de-escalation to fluconazole at day 5 could be done safely. The latter has to be confirmed in randomized controlled trials.

Competing interests None.

O41 Intensive care acquired pneumonia due to extended-spectrum β-lactamase-producing Enterobacteriaceae (ESBL-PE) among patients with prior colonization: Frequency, risk factors and prognosis

Correspondence: Keyvan Razazi -

Annals of Intensive Care 2016, 6(Suppl 1):O41

Introduction Pneumonia is the most frequent ICU-acquired infection. The increasing prevalence of ESBL-PE carriage on ICU admission raises important questions on empiric therapy policies in patients presenting with ICU-acquired pneumonia, which may include the use of a carbapenem as first-line therapy. The incidence of and risk factors for ICU-acquired pneumonia (ICUAP) among patients with ESBL-PE carriage are unknown.

Patients and methods This 6-year prospective study (May 2009–January 2015) was conducted in the medical intensive care unit of an university-affiliated hospital. All patients with ESBL colonization or infection were prospectively recorded. A detailed clinical profile of each patient was established, and variables associated with ESBL-PE pneumonia after colonization was analyzed.

Results A total of 6168 patients were admitted to our medical ICU during the study period. Nine patients were excluded because they developed ESBL-PE ICUAP before ESBL-PE colonization. Of these patients, 826 (13.4 %) were detected as ESBL-PE carriers on admission screening samples or acquired ESBL carriage while in the ICU. One hundred and six (12.8 % of carriers) patients developed ICUAP after colonization, 47 (44 %) with ESBL-PE, mostly Enterobacter spp. and K. pneumoniae. By multivariable analysis, SAPS II ≥43 (OR 2.6 [1.02–6.8]), more than one episode of ICUAP before the index pneumonia (OR 5.4 [1.2–24.6]), and colonization with E.cloacae or K. pneumoniae rather than with E. coli (OR 7.2 [1.8–28.9]) were independent predictive factors for ESBL-PE ICUAP with prior colonization. More than 2 days of fluoroquinolone within the past 3 months fell short of statistical significance (P = 0.058) Amoxicillin/clavulanic acid during ICU was a protective factor for ESBL-PE ICUAP (OR 0.29 [0.10–0.82]), thus predictive for another microorganism than ESBL-PE, mostly Pseudomonas. All patients but one (98 %) with ICU-acquired infection caused by ESBL-PE had rectal carriage of ESBL-PE with the same species. Appropriate antibiotherapy was given to most patients without difference between ESBL-PE infections and others, even if aminoglycosides were not considered. ESBL-PE ICUAP had higher SOFA score, a trend to more frequent septic shock at ICU-AP onset and worse outcome [57 % died in ICU vs 41 % (p = 0.09)]. Twenty-four patients had ICUAP caused by a carbapenem-resistant microorganism after colonization, most of which (87 %) were nonfermenting gram-negative bacilli. Carbapenem exposure before or after ICU admission was associated with carbapenem-resistant microorganism.

Conclusion This study highlights the high rate (44 %) of ESBL-PE pneumonia acquired in ICU in patients with prior colonization. We identified new risk factors for ESBL-PE ICUAP after colonization. These data may be useful for identifying which patients may warrant empiric ESBL-targeted antimicrobial drug therapy, as improving carbapenem use is currently a major challenge for intensivists.

Competing interests None.

O42 Epidemiology, characteristics and outcomes of septic critically ill patients after the removal of a totally implantable venous-access ports

Correspondence: Marie Lecronier -

Annals of Intensive Care 2016, 6(Suppl 1):O42

Introduction International guidelines support the need to remove totally implantable venous-access ports in septic patients admitted to intensive care unit. Nevertheless, the clinical relevance of this attitude is debated. Therefore, we undertook the present study to determine the proportion of confirmed totally implantable venous-access ports-related infection in case of removal. We also assessed the epidemiological, clinical and microbiological characteristics of these patients.

Patients and methods We conducted a multicenter, retrospective, observational study from 2012 to 2014. All patients with life-threatening sepsis in whom a decision of totally implantable venous-access ports removal was taken by the clinician in charge were included. Catheter-related infection was defined by the association of general or local signs of infection and bacteremia and/or positive culture of the catheter or differential time to positivity of simultaneous blood cultures or good evolution 48 h after device removal and after antibiotics initiation.

Results One hundred and fifty one patients were included (57 ± 14 years old, 61.5 % of male). Totally implantable venous-access ports-related infection was confirmed in 75 patients (45 %). IGS2 was 47 ± 15 and 56 ± 17 (p < 0.05) in patients with and without catheter-related infection, respectively. On admission, temperature was 38.6 ± 1.5 and 38.2 ± 1.5 °C (p > 0.05) and mean arterial pressure was 69 ± 23 and 70 ± 18 mmHg (p > 0.05) in patients with and without catheter-related infection, respectively. Local signs of sepsis were present in 27 (18 %) and 6 (4 %) patients with and without catheter-related infection, respectively (p < 0.05). There was no difference between patients with and without catheter-related infection regarding aplasia on admission (30 and 37 %, respectively, p > 0.05). The microbiological findings identified enterobacteria as the most common pathogen (37.5 %). Coagulase-negative staphylococci and Staphylococcus aureus were the others frequent pathogens (30.3 and 16 %, respectively).

Intensive care unit, day 28, day 90 and 6 months mortality were, respectively, for patients with and without catheter-related infection 9 and 40 % (p < 0.01), 22 and 48 % (p < 0.01), 32 and 53 % (p < 0.05) and 40 and 57 % (p > 0.05).

Discussion Almost half of our population had a confirmed totally implantable venous-access ports infection. It is noteworthy that patients with or without catheter-related infection had similar characteristics on admission but a clear difference regarding the short- and long-term survival. These findings support the removal of the catheter in case of life threatening.

Conclusion Almost half of the totally implantable venous-access ports removed in patients admitted for life threatening were really infected. The clinical course of these patients was better than patients with sepsis from other origin.

Competing interests None.


1. Mermel LA, Allon M, Bouza E, Craven DE, Flynn P, O’Grady NP, et al. Clinical practice guidelines for the diagnosis and management of intravascular catheter-related infection: 2009 Update by the Infectious Diseases Society of America. Clin Infect Dis Off Publ Infect Dis Soc Am. 2009 Jul 1;49(1):1–45.

2. Lebeaux D, Fernández-Hidalgo N, Chauhan A, Lee S, Ghigo J-M, Almirante B, et al. Management of infections related to totally implantable venous-access ports: challenges and perspectives. Lancet Infect Dis. 2014 Feb;14(2):146–59.

O43 Genetic risk of Pseudomonas aeruginosa ventilator-associated pneumonia: the EXTENSE study

Correspondence: Jean-Paul Mira -

Annals of Intensive Care 2016, 6(Suppl 1):O43

Introduction Genetic association studies and genome-wide association studies have demonstrated the role of human genetic variations in the epidemiology of infectious diseases in ICU. However, no study has used the power of genome or exome sequencing in this clinical condition. Hence, we design an exome study to search for genetic factors involved in the risk of ventilator-associated pneumonia (VAP) due to Pseudomonas aeruginosa (Pa).

Patients and methods Patients had been previously participated in Pneumagene study that included 3200 ICU patients, which should be on invasive mechanical ventilation for more than 2 days and had no immunodepression, and no risk factor for Pseudomonas aeruginosa VAP (COPD, Pa colonization). For PaVAP group, inclusion criteria included the presence of two or more VAP due to Pa during the same ICU stay. Control group (C) included patients that had no VAP and have not been colonized by Pa during mechanical ventilation (MV) and that can be matched with one PaVAP patient on values of age, SAPS2, duration of MV, reason for ICU admission, potential risk factors for VAP (coma, ARDS, head trauma) and outcome. The Knome Company performed exome sequencing and analysis with the SKAT Common Rare algorithm. Multiple testing was corrected by bootstrap resampling technique with 1000 resamples and a FWER = 0.05.

Results Fifty patients were included in each group. Median age, SAPS2, gender and Pa risk factors were not different between the groups. The control group had higher duration of MV that PaVAP group (59 vs 52.6, respectively, p < 0.05). One thousand genetic variants were finally detected and analyzed. After multiple corrections, 178 loci remain statistically associated with PaVAP risk. Interestingly, two areas appear to be of high interest on the same part of the genome chr 16 p11.2 (26 SNPs and 20 SNPs, respectively, for area 1 and area 2). These two areas are gene deserts but have already been reported in GWAs and seem to be involved in gene regulation.

Conclusion This is the first exome study in ICU patients that did not reveal clear significant genetic variant that can be associated with risk of PaVAP, despite the presence of two promising areas on the chromosome 16. This powerful approach may offer new opportunities for understanding infectious disease susceptibility, severity, treatment, control and prevention.

Competing interests None.

O44 Cytomegalovirus reactivation enhances the virulence of a Staphylococcus aureus pneumonia in a mouse model

Correspondence: Sami Hraiech -

Annals of Intensive Care 2016, 6(Suppl 1):O44

Introduction Cytomegalovirus (CMV) reactivation is common in immunocompetent mechanically ventilated patients. Lungs are a frequent site of reactivation. CMV reactivation may be responsible for higher mortality, length of stay and mechanical ventilation. A higher incidence of bacteremia and nosocomial pneumonia following CMV reactivation may in part explain the worse prognosis of these patients. The aim of this study was to assess the virulence of a staphylococcal pneumonia developed during CMV reactivation in a mouse model.

Materials and methods The study was approved by our local ethic committee. Female BALB/c mice were used in all experiments. CMV primo-infection was obtained by intraperitoneal inoculation of 2 × 104 PFU of murine CMV (MCMV) Smith strain. Seropositivity was confirmed by immunofluorescence in serum. MCMV was considered to be latent 4 months later. Reactivation was triggered by cecal ligature and puncture (CLP). Mice were considered to have a CMV reactivation 2 weeks after CLP. After this, 32 MCMV-positive mice underwent an intranasal inoculation with 5 × 108 CFU of methicillin-susceptible Staphylococcus aureus (MSSA) to induce pneumonia. Twenty-nine MCMV-negative BALB/c mice were treated according to the same protocol, including CLP (control group). Daily weight, signs of sepsis and spontaneous mortality were noted. After 15 days, surviving mice were euthanized. Blood and lung were collected for bacterial culture. Interferon alpha and gamma were assessed in the serum of both groups. MCMV reactivation was assessed by PCR in the lungs. A second cohort of 26 mice was treated according to the same protocol (13 MCMV positive and 13 control) but were killed at day 2 and day 5 after pneumonia. The same samples were performed after killing.

Results No mortality from staphylococcal pneumonia was observed in the control group, whereas the mortality rate was of 9 % in the MCMV group (p = 0.09). The weight loss at day 3 after pneumonia was higher in MCMV mice than in the control group (2.2 vs 0.7 g, respectively; p = 0.005). Macroscopic observation and bacteriological analysis of lungs showed staphylococcal abscesses in 5/32 mice in MCMV group as compared to 0/29 in control group at day 15. At day 5, 3/8 mice had lung abscesses in MCMV group as compared to 0/6 in control group. Overall, 8/40 (20 %) mice had lung staphylococcal abscesses in MCMV group as compared to 0 % in control group (p = 0.011). Lung bacterial count (median (interquartile range) was significantly higher in MCMV-positive mice as compared to control mice at day 2 [5 × 103 (103–3 × 105) CFU/lung vs 102 (0–4 × 102) CFU/lung; p = 0.007] and day 5 [2.5 × 104 (1.6 × 103–6.5 × 105) vs 15 (10–40); p = 0.005]. Two days after staphylococcal pneumonia, MCMV-positive mice had higher interferon alpha serum levels (p = 0.023). No difference between the two groups was observed concerning interferon gamma levels.

Conclusion In a mouse model, CMV reactivation leads to the switch from a non-lethal to a lethal staphylococcal pneumonia, increases bacterial lung count and favors the occurrence of staphylococcal lung abscesses. The role of the immune response and especially the way of interferon alpha seems to be determining.

Competing interests None.


1. Jaber et al., Chest. 2005 Jan;127(1):233–41.

2. Cook et al., J Infect Dis. 2002 May 15;185(10):1395–400.

O45 Families looking back: meaning-making after a decision to accept or refuse organ donation. A qualitative approach

Correspondence: Nancy Kentish-Barnes -

Annals of Intensive Care 2016, 6(Suppl 1):O45

Introduction In the context of brain death, a survey led by the French Institution, the Agence de Biomédecine, in 2007, showed that for 34 % of caregivers, organ donation could increase the pain of the family. The family is at the heart of the organ donation process: if the patient is not registered as a non-donor, then discussions are led with his/her relatives. In a recent study focused on relatives’ experience and grieving process after request for organ donation, relatives were followed up for 9 months to complete instruments measuring the risk of presenting anxiety, depression and post-traumatic stress symptoms as well as complicated grief. After this follow-up, in-depth interviews were led with voluntary relatives in order to permit a better understanding of their experience: understand how they construct and frame their experience 1 year after the patient’s death; analyse the meaning they give to their decision; draw parallels, regarding experiences and feelings, between the experience of donor versus non-donor relatives; and qualify the impact of the decision on the grieving process.

Patients and methods In the main study, 202 relatives were followed up three times over 9 months. After this period, 24 relatives (saturation obtained at 24) participated in an in-depth interview (16 relatives whose loved-one was a donor, eight whose loved-one was not a donor). Interviews were recorded and fully transcribed. Medium length of interview was 1 h 10 min. We privileged an inductive approach (grounded theory) and obtained four themes through thematic analysis: relationship to the patient in the context of brain death; meaning-making regarding organ donation and the decision; search for continuity (follow-up of transplantation process); and impacts on the grieving process (for oneself and for others).


Relationship to the patient in the context of brain death is complex and ambiguous. Relatives question the patient’s condition and wonder whether he/she can hear. Whatever their subjective answer to this question, they need to talk to the patient. Talking to the patient is a ritual that permits closure. Relatives need the clinicians’ support to be able to express their feelings and say good-bye to the patient who remains a person until he/she is taken to the operating theatre.

Meaning-making regarding organ donation and the decision relatives express a strong responsibility about organ donation decision-making. Three justifications of the decision can be put forward: respect of the patient’s choices; altruism; and the hope of life in spite of death.

Search for continuity Relatives stress the importance of being able to follow-up the transplantation process. If some relatives want news, they must get it; if others do not, they need to know they can. Ambiguity is central, as relatives are incapable of considering a possible failure of the organ donation process. With failure comes the risk of dehumanising the organ donation process.

Impact on the grieving process Organ donation decision impacts on relatives’ experience at two levels. First, on a personal level, accepting organ donation can help give meaning to the patient’s death (“He did not die for nothing”) but can also be experienced as a moral duty (“I respected my son, there was no choice”). Second, on a social level, relatives who accepted become organ donation “ambassadors” by encouraging discussions about organ donation and by expressing their positive perspective on the subject.

Conclusion Relatives invest strong meaning into organ donation decision-making. The decision requires reflection prior to the experience and is made with respect to the patient and to the relative him/herself. Relatives cannot afford ambiguity as the decision impacts on their grieving process and on the meaning they give to the patient’s death.

Competing interests None.

O46 The letter of condolence at the heart of family centered care. A qualitative approach of relatives’ experience

Correspondence: Nancy Kentish-Barnes -

Annals of Intensive Care 2016, 6(Suppl 1):O46

Introduction After the death of a loved-one in the ICU, relatives are particularly vulnerable and at high risk of presenting symptoms of anxiety, depression, posttraumatic stress and complicated grief. Moreover, these relatives may experience persistent incomprehension and a feeling of abandonment regarding the ICU team. Many relatives also express the need to contact the ICU team after the death of the patient. In their official report in 2009, the French institution IGAS recommended to prevent complicated grief, emphasizing on the fact that “care of the family must not end with the death of the patient.” In this perspective, a condolence letter (CL) could help recognize both the relative’s pain and his/her strong link to the ICU team. The CL could also help decrease the risk of presenting post-ICU syndrome. In a recent study by the Famiréa group aimed at measuring the impacts of the CL on relatives’ grieving process—a randomized multicentre trial—242 relatives were followed up over 6 months, half of whom had received a CL. The objective of this presentation is to analyse spontaneous feedback by relatives having received a CL. This feedback was gathered orally during telephone follow-up and also using the letters written by relatives to the ICU teams.

Patients and methods Relatives in the intervention group received a CL 15 days after the patient’s death. They were contacted over the phone twice, 1 and 6 months after the loss. During telephone calls, the interviewers noted all spontaneous feedback regarding the CL. Additionally, all letters spontaneously written to the clinician after receiving the CL were retrieved and analyzed. All feedback was transcribed verbatim. This corpus was analyzed using thematic analysis. Three themes will be put forward: humanization of medical care; reflection on care received by the patient; and closure.


Humanization of medical care Relatives are touched by the time spent by physicians in writing the CL. They say that the letters are personalized and feel that they are not “just a number,” but are recognized for the role they played in caring for the patient and in their pain after the loss. The CL helps to give a positive image of the institution: “With this letter you give me the hope that medical care can be humanized and that means a lot to me.”

Reflection on care received by the patient After receiving a CL, relatives want to thank the ICU team for the patient’s care and want to renew their feeling of trust, more specifically regarding pain management. Many relatives also validate decisions made by the ICU team and show their understanding of the situation. Last, the CL helps to think the patient as a whole, with a specific trajectory and personality, thus reaffirming, from the relative’s perspective, the quality of care. “Thank you for your kind letter. I am convinced that my husband received the best care during his stay. I know that he passed away without any pain and this is a great comfort to me.”

Closure The CL is a way of remembering the person who died with the team that cared for him/her. It also confirms the relationship between the clinicians and the relatives. Remembering, sharing and expressing emotions permit closure that, in turn, helps bereaved relatives to accept the loss and to start healing. “Your letter has been a tremendous help, it really did help to ease our pain.”

Conclusion The CL helps relatives give meaning to the patient’s trajectory. Emphasis is put of personalization, showing that these letters cannot be systematic. Writing a CL must also have meaning for the clinician. More importantly, offering the possibility of a last contact with the team is soothing for the relatives and has positive impacts of their grieving process.

Competing interests None.

O47 Use of a children information booklet to support young relatives visiting a critically ill adult: impact on ICU practices

Correspondence: Lionel Kerhuel -

Annals of Intensive Care 2016, 6(Suppl 1):O47

Introduction According to the 2009 SFAR/SRLF guidelines for improving hospital conditions in ICU, the admission of children who wish to visit a critically ill relative “should be facilitated and supervised” by ICU staff members. Yet, the difficulties faced by caregivers in this challenging setting potentially induce restrictions in children visits. The objective of this work was to evaluate the impact on ICU staff practices of a specifically designed children information booklet.

Materials and methods The booklet was designed by a multidisciplinary team (nurses, intensivists, psychiatrists, cartoonist) and included both child-centred explanations (5–12 years old) and advices for accompanying adults. During the study period, booklets were provided freely to patients’ relatives and ICU staff members as a practical tool to help accompanying children who were likely to visit a patient. We used a prospective before–after study design with a questionnaire at baseline and 1 year post-intervention to assess the impact of the booklet on practices in ICU staff members (nurses, auxiliary nurses and physicians). The participants were split into an intervention centre (booklet available) and a control centre (booklet not available). Both centres were general ICUs and were part of a same university hospital. These ICUs were comparable regarding the nurse-to-patient ratio, the number of visiting hours and the admission policy for family visits. Primary endpoints were the proportion of participants who supported visiting children over the 3 months prior to the questionnaire, and the Moral Distress Scale Revised (MDS-R) to evaluate the emotional experience of staff members in this setting.

Results Fifty-seven booklets were distributed in the intervention ICU (8 % of admitted patients). During the study period (07/2014–07/2015), 93 children were likely to visit their relative. A total of 195 staff members participated in the study (75 % of ICU staff). They were nurses, auxiliary nurses and physicians in 56, 27 and 14 % of cases, respectively. Participants were equally recruited at the intervention ICU (87 participants, 74 % of ICU staff) and the control ICU (108 participants, 76 % of ICU staff). In the intervention ICU, the proportion of staff members who supported at least one visiting child over the 3 months prior to the questionnaire increased from 41 % before the intervention to 69 % after the intervention (p < 0.001), whereas it was not significantly different in the control ICU (38 vs 39 %, p = 0.97). A multivariate analysis confirmed that the use of the booklet was an independent factor for increasing the rate of caregivers who supported visiting children [OR 4.0, 95 % CI [2.0; 7.9], p < 0.001]. No significant effect of occupation (OR 0.9, p = 0.89), age (OR 1.7, p = 0.15) and years of experience in critical care (OR 0.7, p = 0.34) emerged from this analysis. In the intervention ICU, the proportion of participants who disagreed with child visits decreased from 35 to 14 % (p = 0.002), whereas it remained stable in the control ICU (30 vs 34 %, p = 0.51). In the intervention ICU, the proportion of participants who claimed that the admission policy for child visits had to be improved decreased from 94 to 76 % (p = 0.001), whereas it remained stable in the control ICU (91 vs 91 %, p = 0.94). There were no significant differences in MDS-R before and after the intervention in both the intervention ICU (21 [12; 30] vs 20 [9; 30], p = 0.44) and the control ICU (18 [10; 29] vs 21 [8; 33], p = 0.13).

Conclusion We reported a substantial increase in visiting children admission in ICU when a specifically designed children information booklet was available. Despite the fact that no emotional effect was identified, the implementation of booklets seemed useful to standardize practices regarding child visits policy among staff members. Further work would be necessary to assess the utility of the booklet regarding children experience during their visit in the ICU.

Competing interests None.

O48 Withdrawal of life-support therapies reporting in ICU randomized controlled trial: Let us move to a transparent reporting—a systematic review

Correspondence: Jonathan Messika -

Annals of Intensive Care 2016, 6(Suppl 1):O48

Introduction Intensivists’ decision to withdraw or withhold life-sustaining therapies may have considerable impact on patient outcome since data suggest that one death out of two in the intensive care unit ensures such a decision. The vast majority of decisions (including withdrawal/withholding life-sustaining therapies) in the intensive care are evidence-based, stemming from ICU trials. We therefore sought to investigate how decisions of withholding/withdrawing life-sustaining therapies are addressed in randomized controlled trials in critically ill patients.

Materials and methods This is an ancillary analysis of a systematic review of randomized clinical trials in critically ill patients. These randomized controlled trials were identified through an exhaustive search conducted in Medline (PubMed®) over a 1-year period (2013). Data extraction was performed by two independent authors (senior intensivists). Whether or not withdrawing/withholding of life-sustaining therapies was reported in these trials was systematically reported. Data are expressed in median [IQR], or n (%) as appropriate.

Results Among the 150 randomized control trials performed on critically ill patients retrieved from the search, only 7 (4.6 %) reported the number of patients for whom decisions of withdrawal/withholding of life-sustaining therapies had been taken. In the subset of studies that had mortality as primary outcome (n = 25), only four reported the number of patients concerned by these end-of-life decisions.

The seven trials involved a total of 2,620 subjects (300 [80–548] patients per trial), totalizing 1055 death (150 [8–225] per trial) for which 413 (5 [4–118]–39 %) were preceded by a limitation of therapeutic effort. The median survival rate was 52 %.

The interventions assessed were drugs for four trials (balanced vs chloride-rich solution in brain injury patients; enoxaparin for thromboprophylaxis; vasopressin, steroids and epinephrine in cardiac arrest; and recombinant human activated protein C in septic shock) and therapeutic strategies for three (high-frequency oscillation ventilation; blood glucose control algorithm; and targeted temperature management after cardiac arrest). All studies reported figures regarding these patients in the results section, including one in a table, but none of them mentioned this information in the abstract. Only one trial indicated in Methods section the way these data were obtained. In the trial assessing target temperature management in cardiac arrest, pre-specified criteria for withdrawal of life-sustaining therapy were established.

Discussion Withdrawal/withholding of life support therapies is very rarely reported in randomized controlled studies in critically ill patients since <5 % numbered this information.

This is prejudicial when mortality is the primary outcome, given the huge impact on mortality when such decisions are taken in patients. We suggest that guidelines for reporting of randomized controlled trials in critically ill patients include reporting on withdrawal/withholding of life-sustaining therapies, at least in studies with mortality as primary outcome, in Methods and Results sections.

Conclusion Although accounting for a number of deaths of intensive care patients, the number of patients with withdrawal/withholding of life-sustaining therapies decisions are scarcely reported in critical care trials. Whether reporting these figures could alter conclusions of the some of these studies remains to be investigated.

Competing interests None.

O49 Gasp during children end-of-life: healthcare providers’ feelings and knowledge

Correspondence: Charlotte Pierron -

Annals of Intensive Care 2016, 6(Suppl 1):O49

Introduction Gasp onset is frequent during children end-of-life and may provoke distress in the witnessing parents. The knowledge and the feelings of healthcare providers confronted to the gasps have never been studied.

Materials and methods We conducted a prospective study in two neonatal and two pediatric intensive care units in Paris, France. We developed a 12-item survey, about the feelings of the healthcare providers witnessing gasps during end-of-life, their knowledge about these events and the information that should be given to the families. The survey was distributed to all healthcare providers in contact with patients.

Results Fifty-one percent of the 488 healthcare providers of the four wards responded to the survey. Among them, 77 % mentioned a discomfort when confronted to gasps, when 43 and 91 % of the respondents think that gasps are uncomfortable for the dying child and for the parents, respectively. Consequently, 83 % of the respondents declare that gasps should be ceased, primarily for the parents or the child discomfort, but also for the healthcare providers. According to the respondents, the efficient treatments against gasps are sedation, analgesia, and neuromuscular blockers. About one-third and two-thirds of the respondents reported that gasps are interfering with their accompaniment of the dying child and of their parents, respectively. Finally, one-third of the respondents declare that they are personally emotionally affected by the occurrence of gasps. Most of the respondents (78 %) state that parents should be informed before the onset of gasps. Avoiding words with negative meanings, they suggest explaining that gasps are not associated with a willingness to survive and are not respiratory movements and further that their occurrence and their length are unpredictable and above all that they are not uncomfortable or painful. Thirty-eight percent of the respondents declare not knowing the mechanism leading to gasps when 60 % think that gasps are secondary to severe brainstem hypoxia or decerebration.

Discussion The prevalence of gasps is not found in the literature but seems to be high according to pediatricians and neonatologists. Gasp is a reflex of auto-resuscitation that follows a functional decerebration induced by a severe brainstem hypoxia. It is believed that gasps are not painful or uncomfortable for the dying patient although this is debated. Nonetheless, in our study, 43 % of the healthcare providers declare that gasps are uncomfortable for the patient, which is indicating a treatment to stop them. Moreover, 91 % think that they provoke discomfort for the parents and 77 % for the healthcare providers. There exist some case reports about the negative feelings of the parents witnessing gasps, but no studies addressing how healthcare providers confronted to sign of end-of-life in dying child react. Even if gasps are not uncomfortable for the dying child, a treatment would seem to be required to avoid distress of the witnessing parents. However, neuromuscular blockers, the only active compounds against gasps, are related to active euthanasia in ventilator-free children. No official instructions exist on the timing and the content of information to be given to the parents. Some recommendations about the sign of end-of-life suggest informing the family before their onset. Only one-third of the respondents, although working in intensive care units, followed a specific formation for palliative cares or end-of-life. This could explain that more than one-third does not know the mechanisms leading to gasps, resuscitating in part their discomfort.

Conclusion Gasps seem to be frequent at the time of end-of-life, and they can be very impressive for the parents. In our study, we show that gasps provoke discomfort to the healthcare providers too and complicate their accompaniment of the parents and the dying child. Almost half of the healthcare providers think that gasps should be prevented, mainly for the well-being of the parents. A lack of specific formation is likely a source of the difficulties of the healthcare providers. A dedicated formation would allow the healthcare providers to better assist the parents and diminish their distress. Finally, studies on the prevalence of gasps in pediatric population and on how the parents are dealing with them, depending upon the information given by healthcare providers would help us to limit parental and professional distress.

Competing interests None.

O50 Prognosis of malignant primary brain tumors in intensive care unit

Correspondence: Maxens Decavèle -

Annals of Intensive Care 2016, 6(Suppl 1):O50

Introduction In the two past decades, the admission of cancer patients to the intensive care unit (ICU) has increased steadily and accounts for 10–20 % of all ICU admissions nowadays. Malignant primary brain tumors (MPBT) are rare tumors representing <2 % of all new cancer cases in Europe in 2012. Interestingly, if the prognosis improvement of hematological malignancies or solid tumors has been well documented over the past 20 years, few data exist concerning the prognosis of MPBT in ICU. Data on survival of these patients when admitted to the ICU could help accurate triage and management decision. The aim of our study was to describe the profile of MPBT patient admitted to the ICU, to assess ICU and 90-day survival and to identify factors associated with ICU and 90-day mortality in a large cohort of critical MPBT patients.

Patients and methods We performed a bi-center retrospective study of a consecutive cohort of patients with MPBT admitted to ICU over a 19-year period (March 1995–May 2014). Only non-metastatic malignant primary brain tumors were included. Were excluded brain metastases of solid cancers, diffuse lymphomas with cerebral localization and benign brain tumors. The median and the interquartile range were calculated for continuous variables and the absolute and relative frequencies for categorical variables. Statistical analysis consisted of univariate and multivariate logistic regression analysis on ICU and 90-day mortality.

Results A total of 197 patients were included. Acute respiratory failure (ARF) was the main reason for ICU admission (45 %) followed by seizures (25 %) non-epileptic comas (14 %) and shock (12 %). Pneumocystis pneumonia and pulmonary embolism accounted, respectively, for 18 and 9 % of ARF. Admission for epilepsy was more common in patients with glial lesions (84.0 vs. 66.7 %, p = 0.019), and patients with primary brain lymphoma were more frequently admitted for shock (41.6 vs. 18.4 %, p = 0.015). Leukopenia was more commonly found in patients with primary brain lymphoma (50.0 vs. 18.9 %, p = 0.005). ICU and 90-day mortality were, respectively, 23 and 51 %. Date of admission did not impact on mortality. The introduction of invasive mechanical ventilation (MV) in epileptic patients did not affect ICU mortality. In multivariate analysis, factors associated with ICU mortality were the reason for admission for seizure, cancer progression, respiratory rate and Glasgow coma scale (GCS). Factors associated with 90-day mortality were the admission for seizure, cancer progression, the need for mechanical ventilation, systolic blood pressure, respiratory rate, GCS and the Charlson comorbidity index (Table 3). A long-term (1 year) survival was not rare and was observed in around 30 % of cases.

Table 3 See text for description

The median and the interquartile range were calculated for continuous variables and the absolute and relative frequencies for categorical variables

OR odds ratio, CI confidence interval, MV mechanical ventilation

Conclusion ICU mortality of MPBT approaches that of the other solid tumors. Among MPBT patients admitted to ICU, half are alive at 90 days. Epilepsy is associated with a particularly good short-term prognosis. Today, the existence of a MPBT does not appear to be sufficient for ICU recusal admission.

Competing interests None.

O51 Impact of early immunomodulating treatment on outcome of adult patients with anti-N-methyl-d-aspartate receptor encephalitis requiring intensive care: a multicentre study with prospective long-term follow-up

Correspondence: Etienne De Montmollin -

Annals of Intensive Care 2016, 6(Suppl 1):O51

Introduction Encephalitis due to anti-N-methyl-d-aspartate receptor (NMDAR) antibodies is a rare neurological syndrome, representing 4 % of all encephalitis. Despite the fact that intensive care need is both prognostic of a bad neurological outcome and frequent in the course of the disease (75 % of patients), there are presently no data on this specific population. We aimed to describe patients admitted to an intensive care unit (ICU) with anti-NMDAR encephalitis and to identify prognostic factors for good neurological outcome.

Patients and methods In this multicentre study, we included consecutive patients meeting all the following criteria: (1) age >15 years, (2) admission to an ICU with a diagnostic of encephalitis, and (3) positive testing for anti-NMDAR antibodies by immunochemistry and cell-based assay in the cerebrospinal fluid (CSF) at the French National Reference Centre. The primary outcome was a good neurological status at 6 months, defined by a modified Rankin score ≤2. Data on ICU management were retrospectively collected, and neurological outcome was prospectively assessed up to 24 months after ICU admission. Early treatment was defined as administration of the treatment <9 days after intensive care unit admission. Data are presented as numbers (percentages) and medians [inter-quartiles]. Clinically relevant variables with a p value ≤0.1 in univariate analysis were included in the multivariate analysis.

Results We identified 77 patients between 1 January 2007 and 31 December 2014 from 52 hospitals and three countries (France n = 73, Belgium n = 2, Switzerland n = 2). One patient was excluded from analysis because of missing data on the primary outcome.

Patients (aged 24 [20–31] years, female gender 68 (88 %)) were admitted to the ICU mainly for coma (n = 31, 40 %) and seizures (n = 31, 40 %), with a Simplified Acute Physiology Score (SAPS) II of 27 [16–38.5]. The delay between first neurological symptoms and ICU admission was 10 [5–26] days. A tumour was diagnosed in 31 patients (41 %) and consisted in mature teratoma (n = 23, 30 %), immature teratoma (n = 6, 8 %) and small cell lung carcinoma (n = 2, 3 %).

During ICU stay, 70 patients (92 %) experienced seizures and 35 (46 %) presented a status epilepticus. ICU length of stay was 55 [24–88] days. First-line immunomodulating treatment consisted in steroids (n = 61/74, 82 %), intravenous immunoglobulins (n = 71/74, 96 %) and plasmapheresis (n = 17/74, 23 %). Second-line therapy was necessary for 46/74 patients (62 %) and included cyclophosphamide (n = 6/45, 13 %), rituximab (n = 24/45, 53 %) or both (n = 15/45, 33 %). At 6, 12 and 24 months, good neurological outcome was achieved in, respectively, 43/76 (57 %), 59/74 (80 %) and 49/56 (87 %) patients. At 6 months, 3 (4 %) patients were dead.

Variables tested in the multivariate model were: age at admission, delay from ICU admission to tumour imaging, delay from ICU admission to first immunomodulating treatment, type of first treatment and white blood cell (WBC) count in the first CSF examination. By contrast, SAPS II, extra-neurological organ failures, status epilepticus, the presence of a tumour and type of second-line treatment were not associated with outcome. Low WBC count on first CSF examination and early combined first-line treatment were independent factors for good neurological outcome at 6 months (Table 4).

Table 4 Multivariate analysis of factors for good neurologic outcome at 6 months

Conclusion We report in this study the largest cohort to date of intensive care patients with anti-NMDAR encephalitis. We show that a low CSF inflammation and early combined immunomodulating treatment by steroids and intravenous immunoglobulins are independent prognostic factors for good neurological outcome, advocating for prompt diagnosis and aggressive treatment of this rare disease. We also show that severe neurological dysfunction and non-neurological organ failures at admission are not prognostic and that despite prolonged ICU stays, the long-term neurological outcome is good, with improvement up to 24 months after ICU admission.

Competing interests None.

O52 Determinants of outcome in critically ill patients with hematological malignancy and central neurological failure: data from the TRIAL OH study

Correspondence: Stéphane Legriel -

Annals of Intensive Care 2016, 6(Suppl 1):O52

Introduction Neurological failure in patients with hematological malignancy admitted to the intensive care unit has received little attention. We sought to report determinants of outcome in these patients.

Patients and methods We performed a post hoc analysis of the TRIAL OH database. The TRIAL OH study was prospectively carried out in 17 university or university-affiliated centers in France and Belgium (2010–2012). Central neurological failure was defined by any neurological disorder of central origin among impairment of consciousness, seizure with or without status epilepticus, focal neurological signs, encephalopathy, and meningeal symptoms.

Results Of the 1011 patients initially enrolled, 226 (22.4 %) demonstrated central neurological failure. There were 132 males and 94 women, aged 62 years (IQR 53–72). Median SAPS II score was 53 (42–66). Median Charlson comorbidity index was 3 (2–4) at ICU admission, and poor performance status was encountered in 54 (24 %) patients.

Median time diagnosis of malignancy and median time from onset of neurological manifestations at ICU admission were 120 days (8–847) and 0 days (0–1), respectively.

42 (42 %) patients were managed at the diagnosis phase of hematological malignancy, and 47 (21 %) patients were in complete or partial remission. Underlying malignancies were as follows: non-Hodgkin’s lymphoma in 85 (38 %), acute myeloid leukemia in 48 (21 %), myeloma in 22 (10 %), acute lymphocytic leukemia in 17 (8 %), chronic lymphocytic leukemia in 17 (8 %), myelodysplastic syndrome in 11 (5 %), Hodgkin’s disease in 5 (2 %), chronic myeloid leukemia in 5 (2 %) and others in 16 (7 %). Bone marrow or hematopoietic stem cell transplantation were performed in 39 (17 %) patients associating 21 (9 %) autologous and 18 (8 %) allogenic stem cell recipients. A history of cancer chemotherapy within 30 days before ICU admission was noted in 95 (42 %), and neutropenia was found at ICU admission in 52 (23 %) patients.

Presenting manifestations were drowsiness (65 %), coma (32 %), tiredness (26 %), seizures (19 %), nausea/vomiting (15 %), headache (11 %), focal neurological signs (10 %), meningism (6 %) and visual disturbances (5 %).

Median SOFA score at day-1 after ICU admission was 8 (5–12) resulting to a total number of organ failures of 3 (2–3). Median GCS score was 12 (7–15). Mechanical ventilation was required in 124 (55 %) patients, catecholamine support in 93 (41 %) and dialysis in 25 (34 %). Eleven (5 %) received chemotherapy during their ICU stay.

Neuroimaging, lumbar puncture and EEG were performed in 113 (50 %), 73(32 %) and 63 (28 %) patients and were deemed contributive to final diagnosis in 51 (45 %), 18 (25 %) and 24 (38 %) cases, respectively. Finally, a neurosurgical biopsy was performed in only one patient providing the cause of neurological impairment.

Topographic localizations of the cause of neurological failure were as follows: encephalic in 200 (88 %), meningeal in 5 (2 %), brainstem in 2, cerebellar and/or medullary in 1, and undetermined in 17 (8 %). According to the direct/indirect classification, malignancy involvement was sought being direct in 11 (5 %) (central nervous system infiltration), neurological indirect in 114 (50 %) (metabolic in 73, central nervous system infection in 12, vascular damage in 14, iatrogenic in 15), extraneurologic in 77 (34 %) (shock in 52, sepsis in 10 and cardiac arrest in 15) and remained unknown in 24 (11 %).

Among 226 patients with neurological failure, hospital mortality was 50 %. By multivariate analysis, independent positive predictors of hospital mortality were poor performance status [odds ratio (OR) 3.99, 95 % CI 1.82–9.39, p = 0.0009], non-Hodgkin’s lymphoma [OR 2.60, 95 % CI 1.35–5.15, p = 0.005], shock [OR 1.95, 95 % CI 1.04–3.72, p = 0.04] and respiratory failure [OR 2.18, 95 % CI 1.14–4.25, p = 0.02], whereas GCS score at day 1 [OR 0.88/point, 95 % CI 0.81–0.95, p = 0.0009] and autologous stem cell transplantation [OR 0.25, 95 % CI 0.07–0.75, p = 0.02] were negative predictors of hospital mortality. The 2-year probability of survival was markedly lower in patients requiring ICU management with neurological failure than in the absence thereof (HR 1.45, 95 % CI 1.18–1.78, p = 0.0003).

Conclusion In hematological patients requiring ICU management, neurological failure is frequent and associated with poor outcome. Type of underlying hematological malignancy, poor performance status, hemodynamic and respiratory failures and intensity of consciousness impairment are independently associated with poor outcome. Early goal directed management in those patients may offer opportunities for improvement.

Competing interests None.

O53 Ammonemia predicts severity and outcome in cirrhotic patients with hepatic encephalopathy in ICU

Correspondence: Simona Tripon -

Annals of Intensive Care 2016, 6(Suppl 1):O53

Background Hepatic encephalopathy (HE), a complication of cirrhosis, presents clinically from mild neuropsychiatric symptoms to coma. It is a major public health problem, it is associated with a poor prognosis, one-half of the patients presenting HE will die in the following year. The physiopathology of HE is complex and multifactorial. Since decades, a relationship between ammonia and the occurrence HE has been proposed, but remains still controversial. The correlation between the severity of HE and the ammonia levels is thus still debated. Often in clinical practice, the serum ammonia is not measured in cirrhotic patients with altered neurological examination.

Aim The aim of this study was to determine whether ammonia was correlated with the severity of HE episode and with the outcome, in a large series of patients with cirrhosis admitted in Liver Intensive Care Unit.

Methods We included prospectively all cirrhotic patients hospitalised in Liver ICU for severe cirrhosis decompensations. Serum ammonia was measured at admission, during the hospitalisation and at the discharge from ICU. Neurological status was assessed by West-Haven (WH) and Glasgow scores. Overt HE was diagnosed as a West-Haven score ranging between 2 and 4.

Results Ninety-eight cirrhotic patients hospitalised in ICU between May 2014 and September 2015 were included. The main clinical characteristics were as follows: mean age was 58 ± 13 years, 74 patients were male, and the aetiology of the liver cirrhosis was: alcoholic (48 %), virus (12 %), mixed (alcool + virus/metabolic) (22 %) and other (15 %). 91 % of the patients had a decompensated cirrhosis (Child–Pugh B and C). 85 % of patients had an abnormal neurological examination at admission, 27 % were hospitalised for HE, and 17 % had already a treatment for HE. The mean ammonia was 99.1 ± 57.3 μmol/L, and 80 % of patients displayed abnormally elevated ammonia (>50 μmol/L). Ammonia was higher in patients with HE than in those without [108 (95 % CI 86–123) vs 57 (95 % CI 50–81), p = 0.0004]. In patients with HE, ammonia was significantly correlated with severity of HE assessed by GCS (p = 0.001). Overall, 34 HE patients experienced worsening of their neurological status during hospitalisation. Ammonia at admission was significantly higher in those patients [128 (95 % CI 108–147) vs 87 (95 % CI 64–110), p = 0.007]. In multivariate analysis, parameters independently associated with worsening of neurological status were MELD score and ammonia levels at admission (p = 0.01 and 0.02, respectively).

Conclusion Hyperammonia reflects the severity of HE and is correlated with the poor prognosis in cirrhotic patients with HE.

O54 Acute varicella zoster encephalitis admitted to the ICU: a case series of 47 patients

Correspondence: Emmanuel Canet -

Annals of Intensive Care 2016, 6(Suppl 1):O54

Introduction Varicella zoster virus (VZV) is responsible for human infections with various clinical presentations, followed by long-lasting viral latency in the spinal and cranial ganglia. VZV reactivation can be responsible for encephalitis. The purpose of this study is to describe the clinical features and the prognosis of VZV encephalitis in adults patients admitted to the intensive care unit (ICU).

Patients and methods A retrospective multicenter cohort study was performed, including consecutive adult patients admitted to 18 French ICUs between 01/01/1999 and 01/09/2015 with a diagnosis of VZV encephalitis. Data and follow-up were abstracted from the medical charts. Patients were evaluated for vital status at ICU discharge. A univariate analysis was performed to identify factors associated with ICU survival.

Results Forty-seven patients were included in the study. Median age was 54 (inter-quartile range 35–67) years and 22 (47 %) were male. Thirty-nine patients (83 %) reported an underlying cause of immune deficiency. The three main causes of immune deficiency were malignancy (12, 31 %) solid organ transplantation (10, 26 %) and HIV infection (8, 21 %). Long-term exposure to steroids was reported in 19 (49 %) patients. Patients were admitted to the ICU 2 (1–6) days after the onset of the first neurological symptoms. Altered mental status at ICU admission was constant. The other symptoms were focal neurological signs (17, 36 %), seizure (13, 34 %) and status epilepticus (9, 19 %). Thirty-three (70 %) patients had a rash. The median coma Glasgow score (CGS) at ICU admission was 12 (8–15); 12 (27 %) patients had a CGS ≤ 8. Two-third of the patients had fever. The median white blood cell count in the cerebrospinal fluid (CSF) was 59 (16–116)/mm3 with 73 (67–84) % of lymphocytes and 21 (8–67) % of neutrophils. Median CSF protein level was 1.4 (0.7–3.8) g/L, and median CSF glucose level was 3.7 (2.5–5.0) mmol/L. CSF PCR for VZV was tested in 37 (79 %) patients and positive in 89 % of the cases.

Thirty-two (68 %) patients had a central nervous system imaging, 27 (57 %) a computed tomography (CT) and 21 a magnetic resonance imaging (MRI). Twenty (74 %) CTs were reported as normal. Main abnormal findings reported with CT were cerebral edema (2, 7.4 %) and petechial hemorrhages (2, 7.4 %). When performed, most MRI reported abnormal findings (16, 76 %), including brainstem lesions (5, 29 %), arteritis and ischemic infarctions (4, 20 %) and demyelinating lesions (4, 20 %).

The median SOFA score at day 1 was 7 (4–9). Patients were treated intravenously with 10 (10–15) mg/kg/8 h of acyclovir during a median of 14 (9–20) days. In addition, three (6 %) patients received intravenous globulins. During ICU stay, 41 (87 %) patients required invasive mechanical ventilation, 20 (43 %) received vasopressors, and RRT was implemented in 14 (30 %) patients. The median ICU length of stay was 16 (6–39) days. Thirty-six (77 %) patients were alive at ICU discharge.

By univariate analysis, two factors were associated with an increased likelihood of ICU mortality: a SOFA score ≥7 at day 1, odds ratio (OR) 28.8, 95 % confidence interval (1.54–538.4), p < 0.01, and the presence of disseminated intravascular coagulation, OR 7.0, 95 % CI (1.23–39.8), p = 0.03.

Discussion We report the clinical features of 47 patients admitted to the ICU for the management of VZV encephalitis. This diagnosis may be challenging as clinical signs and symptoms may be diminished by an underlying cause of immunosuppression. Indeed, one-third of the patients had no fever and up 30 % had no skin rash. Patients should be investigated for VZV-related cerebral arteritis despite the fact that its treatment remains a matter of debate.

Conclusion Severe VZV encephalitis occurs mainly in patients with a history of immune deficiency. In the ICU setting, about 20 % of the patients experienced status epilepticus. MRI is useful to evaluate patients with VZV encephalitis. ICU mortality is 23 % and correlates with the number of organ dysfunction at ICU admission.

Competing interests None.

O55 High-flow nasal cannula for acute respiratory failure in immunocompromised patients

Correspondence: Virginie Lemiale -

Annals of Intensive Care 2016, 6(Suppl 1):O55

Introduction In immunocompromised patients with acute respiratory failure (ARF), invasive mechanical ventilation remains associated with high mortality. Choosing the adequate device for oxygenation is of the utmost importance in that setting. High-flow nasal cannula (HFNC) has been associated with decreased mortality in ARF. In this study, we assess outcomes of immunocompromised patients treated with HFNC.

Patients and methods We performed a post hoc analysis of iVNIctus study, a multicentre trial of NIV in critically ill immunocompromised patients admitted for ARF in 29 intensive care units in France and Belgium. Only those alive at day 2 were analyzed in this study. A propensity score-based approach was used to assess the impact of HFNC compared with oxygen only on hospital mortality. Propensity score was based on randomization group, severity at admission and etiology of acute respiratory failure. Primary endpoint was mortality at day 28. Secondary endpoints were intubation rate, duration of mechanical ventilation, ICU-acquired infection, ICU and hospital stay lengths.

Results Among 374 patients included in the study, 353 met inclusion criteria. Underlying disease included malignancy (n = 296, 84 %), solid organ transplantation (n = 24, 6.8 %) and other cause of immunosuppression (n = 33, 9.3 %). ARF etiologies were mostly bacterial pulmonary infections (n = 157, 44.4 %) and opportunistic infection (n = 76, 21.5 %). 180 (50 %) patients received NIV. D-28 mortality was 22.6 % (80 deaths). Mechanical ventilation was ultimately needed in 142 (40.2 %) patients.

At ICU admission, 127 patients received HFNC, whereas 226 patients never received HFNC. D-28 mortality was, respectively, 33 (25.9 %) and 47 (20.7 %) (p = 0.26). Ninety patients in each group (HFNC or oxygen only) were matched according the propensity score. Among them, 92 patients received NIV (44/90 in HFNC group vs 48/90 in oxygen group). HFNC was not associated with a decreased D-28 mortality (23.3 vs 25.5 %, p = 0.76). Intubation rate and ICU-acquired infections were, respectively (43.3 vs 50 %, p = 0.47, and 23.3 vs 26.7 %, P = 0.73). ICU lengths of stay and mechanical ventilation duration were not different (respectively, 8 [7–11] vs 7 [6–16 days, p = 0.89, and 13 [4–46] vs 14 [8–33 days, p = 0.72).

Conclusion In this study, HFNC during acute respiratory failure in immunocompromised patients did not improve survival compared with oxygen only. HFNC was associated with neither a decrease intubation rate nor a lower ICU-acquired infection rate.

Competing interests None.

O56 High-flow oxygen therapy through a nasal cannula in immunocompromised patients with acute hypoxemic respiratory failure

Correspondence: Remi Coudroy -

Annals of Intensive Care 2016, 6(Suppl 1):O56

Introduction In the early 2000s, two randomized controlled trials showed that as compared to standard oxygen therapy, noninvasive ventilation could decrease mortality of immunocompromised patients admitted to ICU for acute respiratory failure. However, the benefits of noninvasive ventilation in immunocompetent patients with acute respiratory failure are debated. High-flow oxygen therapy through a nasal cannula may offer an alternative in hypoxemic patients. We recently found in a randomized controlled trial including 310 patients with acute respiratory failure (FLORALI study) that as compared to noninvasive ventilation, high-flow oxygen therapy decreased mortality. Immunocompromised patients could also be included in this study, except for those with profound neutropenia. Therefore, we assessed the benefits of high-flow oxygen therapy or noninvasive ventilation in this subgroup of patients. Our objective was to compare intubation and mortality rates in the subset of immunocompromised patients admitted to ICU for acute respiratory failure.

Patients and methods We performed a subgroup analysis in the framework of the FLORALI study. This study included all patients with non-hypercapnic (PaCO2 ≤45 mmHg) acute respiratory failure while excluding patients with cardiogenic pulmonary edema and those with underlying chronic lung disease. Patients were assigned to three groups according to treatment: high-flow oxygen therapy, standard oxygen therapy or noninvasive ventilation. The primary outcome was the intubation rate, and secondary outcome included 90-day mortality. We focused on the subset of immunocompromised patients included in this study, while patients with profound neutropenia were deliberately excluded.

Results Among the 310 patients with acute respiratory failure, 82 (26 %) were immunocompromised including 26 patients in the high-flow oxygen therapy group, 30 in the standard oxygen group and 26 in the noninvasive ventilation group. Intubation rates were 31, 43 and 55 % in the high-flow oxygen therapy, standard oxygen therapy and noninvasive ventilation groups, respectively (p = 0.04). The 90-day mortality rates were 15, 27 and 46 % in the high-flow oxygen therapy, standard oxygen therapy and noninvasive ventilation groups, respectively (p = 0.046). Ventilator-free days at day 28 were 26 ± 6, 23 ± 10 and 14 ± 13 days in the high-flow oxygen therapy, standard oxygen therapy and noninvasive ventilation groups, respectively (p < 0.0001).

Conclusion In immunocompromised patients admitted to ICU for acute hypoxemic respiratory failure, as compared to standard oxygen therapy or noninvasive ventilation, high-flow oxygen therapy was associated with lower intubation and mortality rates and reduced duration of invasive mechanical ventilation.

Competing interests None.

O57 Prospective validations of the PASTEIL score to assess the clinical pretest probability of Pneumocystis jirovecii pneumonia (PjP) in patients with hematologic malignancies (HMs) and acute respiratory failure (ARF)

Correspondence: Virginie Lemiale -

Annals of Intensive Care 2016, 6(Suppl 1):O57

IntroductionPneumocystis jiroveci pneumonia (PjP) occurs increasingly in non-AIDS immunocompromised patients, chiefly those with HMs. Delayed implementation of trimethoprim–sulfamethoxazole is associated with increased mortality. A clinical score to early assess at the bedside the clinical pretest probability of PjP before proceeding with diagnostic testing would allow avoiding missed diagnosis and delayed treatment.

Patients and methods Last year, we presented at REANIMATION 2015 the PASTEIL score (see Table 5) that was developed in 1092 HM patients with ARF (134 [12.3 %] with proven PjP) admitted to eight ICUs between 2006 and 2012 (“learning set”). First, variables were identified through a multivariable logistic regression model after multiple imputation of missing confounders, and then, the score described below was derived from mean of beta regression coefficients and validated by bootstrap (resampling). Discrimination (i.e., ability of the score can separate PjP from non-PjP patients) was measured by the area under the receiver operating characteristic (ROC) curve (AUC, c statistic). Calibration (i.e., how well predicted probabilities agree with actual observed risk) was measured by the Hosmer–Lemeshow statistic (goodness-of-fit test and calibration curves) that compares the average predicted risk within subgroups to the proportion that actually develops disease. The score was performing very well in the learning set, with a very good discrimination (AUC was 0.87 saying that predicted values for cases are all higher than for non-cases) and a very good calibration (mean goodness of fit of −0.75 and calibration curves suggesting applicability of the models to individual cases in this learning set). Here, we sought to further validate this score in a prospective and independent reliability cohort. We first imputed the missing data using mice (multiple imputation algorithm), based on 30 multiple imputed datasets. On each complete imputed dataset, the score was computed for each patient and averaged; lastly, for each patient, the predicted probability of PjP was derived from the average score and its performance assessed.

Table 5 See text for description

Results To validate the diagnostic score, we collected data from an independent dataset of 238 patients with HMS admitted to 26 ICUs for ARF (iVNIctus trial), including 15 (6.3 %) patients with PjP. When compared to patients with other ARF etiologies, PjP patients were younger (33 aged <50 vs. 20 %, P = 0.02), had more frequently a lymphoproliferative disorder (93.3 vs. 45.7 %, P = 0.0003), never received prophylaxis (vs. 36 % in other ARF etiologies, P < 0.0001) and had more frequently an interstitial pattern on chest imaging (67 vs. 32 %). The three other components of the score, namely time since respiratory symptoms onset, shock at admission and pleural involvement, were not significantly different between PjP patients and those with other ARF etiologies. In this reliability cohort, the score has a very good discrimination [AUC was 0.83 (95 % CI 0.73–0.93, DeLong estimates)]. However, the calibration was suboptimal with predicted probabilities of PjP above the observed proportions of actual PjP in the low deciles of the score.

Conclusion Within hours from ICU admission, bedside clinicians are able to establish the clinical pretest probability of PjP in patients with HMs and ARF. The PASTEIL score has a very good discrimination that provides promising ways to avoid missing the diagnosis and delaying treatment. Interventional studies in high-risk patients are warranted (Fig. 11).

Fig. 11
figure 11

See text for description

Competing interests None.

O58 Bedside contribution of electrical impedance tomography (EIT) to set mechanical ventilation for severe acute respiratory distress syndrome (ARDS) on extracorporeal membrane oxygenation (ECMO)

Correspondence: Guillaume Franchineau -

Annals of Intensive Care 2016, 6(Suppl 1):O58

Introduction Extracorporeal membrane oxygenation (ECMO) has been proposed as a possible therapeutic option for patients with severe acute respiratory distress syndrome (ARDS) who have refractory hypoxemia or excessively high inspiratory airway pressures as well as being unable to tolerate volume and pressure limited strategy. This device also permits “ultraprotective” mechanical ventilation with further reduction in volume and pressure that might ultimately enhance lung protection and improve clinical outcomes of acute respiratory distress syndrome (ARDS) patients. In addition, high level of positive end-expiratory pressure (PEEP) during the first day on ECMO may be associated with better outcome. On the other hand, inappropriate high PEEP level may also lead to overdistension, hyperinflation and definitive lung damage. Because of these considerations, continuous monitoring of harmful effects of PEEP appears relevant. Considering the high severity of these patients, electrical impedance tomography (EIT) could allow an individual, noninvasive, real-time, bedside, radiation-free imaging of the lungs, with global and regional dynamic analysis.

The aims of this study were: (1) to evaluate the ability of EIT to monitor a PEEP trial in patients with severe ARDS on ECMO and ventilated with very low tidal volume and (2) to evaluate the benefits of EIT to provide tools to individualize and match PEEP with the regional distribution of ARDS lesions.

Patients and methods We conducted a monocentric observational prospective study. Inclusions criteria were: (1) sedated patients mechanically ventilated in pressure-controlled mode for severe ARDS that required veno-venous ECMO; (2) delay between ECMO initiation and inclusion <7 days. Exclusion criteria were manufacturer contraindications to use of a thoracic belt, cardiac pacing, pneumothorax or past history of barotrauma and hemodynamic instability. A decremental PEEP trial, using 5 cmH2O 20-min steps, was performed from 20 to 0 cmH2O. Driving pressure and respiratory frequency were maintained constant at 14 cmH2O and 24/min, respectively. At the end of each PEEP level, hemodynamic, pulmonary compliance, blood gases and EIT were recorded. Lung images were divided into four symmetrical frontal to dorsal region of interest (ROI). For each PEEP level, we recorded impedance variation (Δz) in each ROI, end-expiratory lung impedance (EELI) and estimation of the alveolar collapse (CL) and overdistension (OD) area.

Results We included 15 patients [age 57 (range 40–72), six males, SOFA score 15 (12–17)]. Origin of ARDS was viral pneumonia (46 %) and bacterial pneumonia for 20 % of patients. Analysis was conducted after a median time of mechanical ventilation of 8 (6–12) days and 4 (2–6) days of ECMO. Compared with PEEP 20, decremental PEEP trial led to a decrease in EELI of 14, 37, 60 and 96 % at PEEP 15, 10, 5 and 0, respectively. Regional analyses of EELI and Δz illustrated that “ultraprotective” ventilation of the most dorsal region of the lungs did not participate in ventilation regardless the level of PEEP. However, mid-ventral and mid-dorsal regional ventilations were more dependent on PEEP. Meanwhile, pulmonary compliance was 13.4, 18.7, 18.6, 18.7 and 12.1 mL/cmH2O at the five decreasing PEEP levels, respectively, resulting to median tidal volumes ranging from 2.8 to 4 ml/kg. Median overdistension decreased from 37 to 0 % and collapsed area of the lung increased from 0 to 50 % while decreasing PEEP level. Based on the best compromise between low overdistension and low collapse, EIT provided “best PEEP level” for each patient on ECMO. These results were variable, and depending on patients, optimal PEEP was between 5 and 15 cmH2O (see Fig. 12).

Fig. 12
figure 12

See text for description

Conclusion Our study is the first using EIT on severe ARDS with ECMO. Our results indicate that EIT is a simple and feasible technique that could provide good assessment of regional changes during PEEP maneuvers, despite a low spatial resolution and very low tidal volumes. In our population, best PEEP level, which protects from both overdistension and collapse, did not exist. However, EIT may provide a tool to select PEEP level with the best compromise between overdistension and collapse. This level was very variable from patient to patient, illustrating the need to individualize mechanical ventilation settings based on monitoring the effects of PEEP.

Competing interests None.

O59 Liberal oxygenation versus restrictive oxygenation in ICU patients: effects on proinflammatory cytokines, sRAGE and organ dysfunctions - results of an ancillary study

Correspondence: Loïc Barrot -

Annals of Intensive Care 2016, 6(Suppl 1):O59

Introduction There is no clear recommendation on which level of oxygenation is better in intensive care patients. Oxygen (O2) is known as a double-edged sword because of the risk of oxygen toxicity, especially in the lungs in balance with the risk of hypoxemia. In a previous multicenter international randomized study, we tested the safety and the feasibility of management of two levels of oxygenation in ICU patients with one group having oxygen saturation kept between 88 and 92 % (conservative = CO) and the other having oxygen saturation kept higher or equal to 96 % (liberal = LO) (Am J Respir Crit Care Med 2015; 150903080351007). Results of the pilot study confirm the feasibility and the safety of conducting a such study. Here, we report results of the French cohort with measurement of proinflammatory cytokines, soluble receptor for advanced glycation end products (sRAGE) a marker of pulmonary epithelial injury and markers of organ dysfunction.

Patients and methods This is an ancillary study of a multicenter, international, prospective, randomized study. Inclusion criteria were age higher or equal to 18 years, mechanical invasive ventilation begun not more than 24 h, previsible length of ventilation higher than 48 h. Exclusion criteria were pregnancy, legally incapacitated patient, lack of written consent of trusted person or next-of-kind, high risk of death during the next 72 h and clinician in charge of the patient refusal. The FiO2 was set to achieve targets of 88–92 % SpO2 for the CO group or ≥96 % SpO2 for LO group during mechanical ventilation. IL-1, IL-6, TNF-α, IL-12, IL-8 and sRAGE were dosed at days 0, 3 and 6. Troponin, catecholamines, lactates, creatininemia, platelets, fibrinogen, hemoglobin and CRP were measured at days 0, 1, 2, 3, 4, 5, 7 and 10. Hemodynamic instability, new onset of ARDS and PaO2/FiO2 at day 7 was noticed in ICU and in the hospital, at day 90 and in ICU for mortality.

Results A total of 240 patients were screened. Twenty-two patients were recruited from June 2013 to October 2014. There were 11 patients per group. Patients characteristics were similar between the two groups except a greater number of patients with ARDS at the inclusion in CO group. The two groups had different mean PaO2 over time with 68 mmHg [64–79] in CO group versus 89 mmHg [84–106] in LO group. Variations of IL-1, IL-12 and TNF-α were not detectable.

IL-6 (pg/ml) was 129 [50–333] versus 50 [18–142] at day 0, 40 [14–109] versus 25 [8–78] at day 3, 20 [6–61] versus 21 [7–64] at day 6 in, respectively, liberal and conservative groups.

IL-8 (pg/ml) was 43 [20–92] versus 36 [16–84] at day 0, 18 [8–39] versus 25 18 [8–44] at day 3, 14 [6–33] versus 19 [8–45] at day 6 in, respectively, liberal and conservative groups.

IL-10 (pg/ml) was 3.68 [1.08–4.46] versus 4.54 [0.81–5.86] at day 0, 2.38 [0.68–3.19] versus 0.64 [0.17–0.97] at day 3, 1.79 [0.35–4.04] versus 0.28 [0.14–0.77] at day 6 in, respectively, liberal and conservative group.

SRAGE (pg/ml) was 2336 ± 305 versus 2172 ± 334 at day 0, 1756 ± 316 versus 2038 ± 364 at day 3, 2020 ± 376 versus 1297 ± 364 at day 6 in, respectively, liberal and conservative groups.

All show significant variation over time. No significant variation overtime between the two groups was detected despite a tendency to a higher sRAGE elevation in liberal oxygenation group at day 6. There was no difference in mortality (in ICU, in hospital or at day 90). There was no difference in PaO2/FiO2 at day 7, hemodynamic instability. More episodes of hypoxia were detected in the restrictive group as expected. There was no difference between the two groups over time of troponin level, catecholamine doses, creatininemia, platelet, fibrinogen, lactates and CRP. SOFA has nonsignificant tendency to be lower in the CO than in the LO group. There was a nonsignificant tendency for new onset of ARDS in LO group. Hemoglobin was lower in the CO group due to lower basal level.

Conclusion A liberal oxygenation strategy does not induce higher systemic inflammation neither other complication in comparison with a conservative oxygenation strategy in ICU patients. However, despite nonsignificant, sRAGE, a marker of epithelial pulmonary injury, has tendency to stay at high level at day 6, there is tendency to have more ARDS, and SOFA score at day 6 has tendency to be higher in patients exposed to liberal oxygenation strategy in comparison with conservative oxygenation strategy. Further research has to be conducted to find the more safe level of oxygenation in ICU patients, especially for the protection of the lungs.

Competing interests None.


1. Panwar R, Hardie M, Bellomo R, et al.: Conservative Versus Liberal Oxygenation Targets for Mechanically Ventilated Patients – a Pilot Multicenter Randomized Controlled Trial. Am J Respir Crit Care Med 2015; 150903080351007.

2. Kallet RH, Matthay MA: Hyperoxic Acute Lung Injury. Respir Care 2013; 58:123–141.

O60 Immature granulocyte level at the acute phase of sepsis, a potential prognostic marker of clinical deterioration. The Septiflux 2 multicenter trial

Correspondence: Thomas Daix -

Annals of Intensive Care 2016, 6(Suppl 1):O60

Introduction Sepsis remains a major cause of death for intensive care unit (ICU) patients. The numerous experimental data published contrast with the few biological data (mainly neutrophil and platelet count) used by physicians at patients’ bedside in such circumstances. Since there are no specific predictive clinical signs in sepsis, it would be helpful to have a biological tool to predict the progression of the disease but also to understand the pathophysiology and especially the immune status of the patients. Given the great worsening potential, using a prognostic marker to early identify patients at risk of clinical deterioration could enable a more specific management. In that way, “Septiflux 1,” a single-center trial (Guerin et al., Crit Care Med, 2014) on “selected septic patients,” demonstrated that immature granulocyte CD10dim/CD16dim level at acute phase of sepsis can predict early clinical deterioration. We wanted to confirm these promising results in a large multicenter trial. The main objective was to evaluate the ability to predict early clinical worsening of immature granulocytes CD10dim/CD16dim assessed by flow cytometry at the acute phase of sepsis. Secondary objectives were to appraise the capacity of other leukocyte subsets to predict clinical deterioration at D-2 and D-28 death.

Patients and methods We designed a multicenter prospective observational study. Patients hospitalized in the emergency department (ED) or in the ICU with SIRS and a clinically or microbiologically documented bacterial infection developing for <24 h were eligible. Exclusion criteria were pregnancy, progressive neoplastic disease, HIV, history of hematological or inflammatory disease, immunosuppressive treatment and hospitalization in ICU for more than 48 h. Within 8 h after admission, flow cytometry analysis was performed to evaluate immature granulocytes CD10dim/CD16dim but also T-lymphocytes CD3pos, expression of CD24, CD64 and CD11b on granulocytes, and pro-inflammatory monocytes CD16pos. As regards their organ failure, all patients were classified as sepsis, severe sepsis and septic shock using the international sepsis classification. Sepsis deterioration was evaluated 48 h after inclusion and was defined as followed: sepsis developing to severe sepsis or septic shock; severe sepsis developing to septic shock. Death occurring before 48 h was also considered as clinical deterioration. Patients were followed until D-28. SOFA score was also calculated at inclusion and after 48 h.

Results From November 2013 to June 2015, at 11 university sites, 1062 patients were included in this trial. Recruitment has now ended and 780 patients have been kept for final analysis, the remaining being secondary excluded mainly because sepsis was not confirmed by the adjudication committee. At inclusion, 523 (67 %) were hospitalized in the ICU and 257 were from the ED. All the sites have performed the related flow cytometry according to a standardized protocol. The database cleaning process and locking should be done in the coming weeks followed by the statistical analysis. Results will be available in the coming 2 months.

Conclusion Septiflux 2 is the larger multicenter trial that has evaluated the prognostic value of immature granulocytes and the overall interest of flow cytometry at the acute phase of sepsis.

Competing interests None.


1. Guérin E, Orabona M, Raquil MA, et al. Circulating immature granulocytes with T cell killing functions predict sepsis deterioration. Crit Care Med 2014;42(9):2007–18.

O61 Analysis of microbiome and resistome using next-generation sequencing in urine samples from patients with sepsis, severe sepsis or septic shock

Correspondence: Philippe Vignon -

Annals of Intensive Care 2016, 6(Suppl 1):O61

Introduction Early and appropriate antibiotic therapy is a prognostic factor in patients presenting with severe sepsis and septic shock. Incidence of multidrug-resistant bacteria (MRB), especially Gram-negative bacilli (GNB), continuously increases. Next-generation sequencing (NGS) is a powerful technique that enables sequencing the complete genome of bacteria, thus giving access to their precise identification and potential antibiotic resistance genes. However, this technique is currently too cumbersome and expensive to be implemented in routine practice. Accordingly, the feasibility of NGS has been scarcely assessed on biological samples obtained from patients with sepsis. This study aimed at evaluating the feasibility and the diagnostic capacity of NGS in patients with urosepsis using the detection of bacterial microbiome and resistome from urine samples. Results were compared with those of conventional microbiology.

Patients and methods In this pilot study, we analyzed the urine samples of 40 patients with urinary sepsis (n = 17), severe sepsis (n = 8) or septic choc (n = 15) secondary to upper tract urinary infection (17 men; mean age: 71 years). Conventional urine culture was either positive with a single isolated GNB (n = 27) or two GNB (n = 6), being MRB or not, positive with three or more bacteria (n = 3), or negative (n = 4). After total DNA extraction and partial depletion of human DNA, prokaryotic DNA from urine samples was sequenced using IonProton™ technology. The bioinformatic analysis was conducted with the Geneious® software (microbiome) and the Web site of the Center for Genomic Epidemiology (microbiome and resistome). All NGS analyses and their final interpretation were performed by a medical fellow who had received personalized training prior to the study. NGS results were compared to those of conventional culture. Initially, microbiological results from conventional culture were known for comparison with NGS data (Group 1; n = 21). Subsequently, the medical fellow who performed NGS analysis was blinded from the results of conventional cultures (Group 2; n = 19).

Results For the microbiome (Table 6), NGS showed a total concordance for the detection of bacterial species (mainly Escherichia coli), in samples with single isolated bacteria. NGS enabled the identification of bacteria potentially responsible for urosepsis in three out of four urine samples with negative conventional culture. For the resistome, NGS enabled the identification of all genes explaining the acquired resistance phenotypes (n = 18) and especially all ctx-m genes which give MRB phenotype to bacteria (ESBL, n = 11). No relevant difference in NGS diagnostic capacity was observed between groups. However, the study showed that a time-consuming in-depth analysis was still required.

Table 6 See text for description

Conclusion NGS performed on urinary samples from patients with urosepsis appears feasible and provides concordant information with that obtained from conventional culture regarding both the identification of bacteria and associated profiles of antibiotic resistance. Even if the technical requirements, duration of computerized analysis and costs currently preclude its clinical implementation, NGS promises in the near future to provide accurate results faster than conventional bacterial culture to early guide initial empirical antibiotic therapy in patients with urosepsis.

Competing interests None.

O62 Patients metabolomic profiles at intensive care admission

Correspondence: Aurelie Thooft -

Annals of Intensive Care 2016, 6(Suppl 1):O62

Introduction In stress situation, inflammation causes several metabolic modifications. To give energetic substrate to vital tissues, an activation of proteolysis, lipolysis and hepatic gluconeogenesis occurs and, at cell level, the main energy source comes from anaerobic reactions. The metabolomic analysis is an approach that enables to study many products from metabolic pathways [1]. We aimed to study metabolic profile of patients admitted in intensive care unit (ICU) to identify potential biomarkers of patients’ outcome.

Patients and methods This is a prospective study in a medico-surgical ICU. We included adult patients admitted in ICU for <24 h and healthy volunteers. At admission, the sera of the patients were analysed by 1H nuclear magnetic resonance. We first compared patients and volunteers by multivariate analysis. The patients were divided into four groups based on their diseases: sepsis, septic shock, shocks from other origins than sepsis and other pathologies. We realized a Kruskal–Wallis test followed by Dunn correction to analyse the metabolites between the different groups. After identification of relevant metabolites, we search correlations between these and the mortality and the length of stay.

Results A total of 117 consecutive patients were screened, of whom 111 were included in the study after informed consent signature. Mean age was 67 ± 14, mean SOFA score was 5 ± 3, and mean APACHE II score was 20 ± 9. The 90-day mortality was 30 %. We first identified a significant difference in metabolic profile between volunteers and ICU patients (CV-ANOVA, p < 0.001). When septic shock patients were compared with non-septic patients, some significant differences in the metabolome were observed (p < 0.05): pyruvate, lactate, carnitine, creatine, myo-inositol, creatinine, urea, phenylalanine and mannitol. Despite all these changes, there was no correlation between metabolomic profile and patients prognosis.

Discussion These metabolic differences indicate a profound change in energetic metabolism in septic shock patients. The high ratio lactate/pyruvate indicates a trend towards anaerobic glycolysis. The lactate increasing comes from hypoxic area and can serve for hepatic gluconeogenesis via Cori cycle. The increase of carnitine reflects the free fatty acids oxidation to be used as energy source. A reflect of proteolysis is observed by the rise of phenylalanine. The amino acids are used in the Krebs cycle or are degraded in urea, which is also increased in our study. We also observed a rise of creatine—energetic substrate of muscles—and its degradation product, creatinine. The myo-inositol increase reflects an activation of intra-cellular messages. Finally, the mannitol, which does not come from human metabolism, could perhaps reflect bacterial translocation.

Conclusion Metabolomic approach is a new interesting way to characterize metabolic changes in ICU patients at their admission. In the group of septic shock, we observed an increase in anaerobic glycolysis, proteolysis, lipolysis and gluconeogenesis. Nevertheless, the differences are not correlated with prognosis.

Competing interests None.

O63 Lymphocyte phenotype during severe sepsis and septic shock

Correspondence: Matthieu Le Dorze -

Annals of Intensive Care 2016, 6(Suppl 1):O63

Introduction Two major immuno-inflammatory phases are well documented during sepsis: an early active inflammatory phase and a deactivated inflammatory phase leading to immunosuppression. Early death (<7 days) and late death (>7 days) must be then distinguished. The immune deficiency concerned both innate and adaptive system and represents a potential indication for immunostimulation therapy (IFN-γ, IL-7, PD-1 antagonists, etc.). Finding immune parameters to be used as a monitoring of inflammatory status becomes a major goal for research. Lymphocyte as a source of mediators for innate immunity has a phenotype in severe septic patients that is not well characterized. The objective was to describe and compare lymphocyte phenotypes to healthy volunteers, measured early (<7 days) in the course of severe sepsis with or without shock, and to study the relationship of this phenotype with prognosis.

Patients and methods We conducted a monocenter prospective observational study. Patients and groups: Severe Sepsis Group = severe sepsis ± septic shock; Control group = healthy volunteers. Measurements at D2 (D1–D3): demographic, clinical severity, mortality, blood cell count, monocytic HLA-DR expression, lymphocyte phenotype (flow cytometry, CD3+, CD3+CD4+, CD3+CD8+, CD19+, CD16+CD56+).

Results Forty-eight healthy volunteers were in control group: 24M/24F and aged 42 (31–54) years. Sixty-seven patients were in severe sepsis group: 39M/28F and aged 69 (52–85) years. Charlson, SOFA and IGS2 scores were 5 (2–7), 8 (5–13) and 52 (38–67), respectively; sites of sepsis origin are abdominal (33 %), lung (27 %) and others (40 %); early death was at J7 (28 %) and late death was at J7–J28 (20 %) (Table 7).

Table 7 Lymphocyte phenotype and mHLA-DR expression [values are expressed as median (first quartile–third quartile); Student’s t test]

Lymphocyte phenotype was not related to the severity scores (IGS2 SOFA). Lymphocyte phenotype was not related to early (<day 7) and late death (day 28).

Conclusion T lymphopenia (whatever the sepsis severity) is associated with a decreased monocytic HLA-DR expression suggesting a synergistic change in both innate and adaptive immunity. This phenomenon can be physiologic, but could be a marker of immunosuppression associated to secondary infections. Monitoring mHLA-DR expression and lymphocyte phenotype over the course of sepsis may argue in favor of immunostimulation therapy, especially if secondary infections occurred.

Competing interests None.


1. Hotchkiss RS, Monneret G, Payen D. Nature Rev Immunol, 2013.

2. Boomer JS et al. JAMA, 2012.

O64 Myeloid-derived suppressor cells expressing arginase-1 and IDO play a major role in immune dysfunction during septic shock

Correspondence: Fabrice Uhel -

Annals of Intensive Care 2016, 6(Suppl 1):O64

Introduction Patients who survive the early hours of sepsis develop an acquired immune dysfunction responsible for nosocomial infections and late mortality. Myeloid-derived suppressor cells (MDSCs), a heterogeneous population characterized by their ability to suppress T-cell responses, may play a major role. They have been described in numerous tumors as well as in inflammatory and infectious diseases. However, their phenotype and mechanism of action remain unclear. The objective of this study was to broadly assess the presence and mechanisms of suppression of MDSC subsets during septic shock.

Patients and methods In a prospective cohort of patients with septic shock (n = 35) and healthy controls (n = 26), peripheral blood myeloid cell subsets were quantified and phenotypically analyzed by multicolor flow cytometry.

The expression of 45 myeloid genes was studied by qRT-PCT on whole-blood RNA from septic patients (n = 29) and healthy controls (n = 15). Plasma levels of MDSC mediators were quantified by ELISA and high-performance liquid chromatography (HPLC) in 73 septic patients and 19 healthy controls. In vitro, CFSE-labeled peripheral blood mononuclear cells were depleted or not of CD14pos or CD15pos cells and stimulated with anti-CD3/CD28 antibodies; T-cell proliferation was assessed by CFSE dilution. Arginase and indoleamine 2,3-dioxygenase (IDO) activities were determined by HPLC.

Results We identified a suppressive phenotype in circulating myeloid cells from septic patients, characterized by an increase in circulating monocytic M-MDSC (CD14posHLA-DRlo/neg monocytes) count (p < 0.001) and in CD14negCD15pos low-density granulocytes identified as granulocytic G-MDSCs (p < 0.0001). Further phenotypic analyses showed that G-MDSCs include subsets of both immature granulocytes and mature granulocytes expressing high levels of degranulation markers. G-MDSCs were more specifically associated with occurrence of nosocomial infections.

A panel of genes involved in MDSC biology, including S100A8, S100A9, MMP8 and ARG1, was significantly upregulated in the blood of septic patients compared with healthy controls. Plasma levels of MDSC mediators S100A8/A9, S100A12, arginase 1 as well as IDO activity were significantly increased in septic patients compared with healthy controls (p < 0.001).

In vitro, restoration of T-cell proliferation after CD14pos and CD15pos depletion confirmed the suppressive properties of M- and G-MDSCs, respectively. Quantitative analysis of amino acids concentrations in culture supernatants revealed an arginase activity specifically burden by G-MDSCs and an IDO activity burden by both M- and G-MDSCs.

Conclusion In patients with septic shock, M-MDSC and G-MDSC counts are increased and promote the occurrence of nosocomial infections. Those subsets bear IDO and arginase activities and exert a suppressive activity on T-cell proliferation. These new insights highlight the complex mechanisms underlying immune dysfunction in septic patients and could promote future therapeutic approaches.

This work was supported by the 2012 experimental research grant from French Intensive Care Society.

Competing interests None.

O65 Anti-neutrophil cytoplasmic antibody (ANCA)-associated vasculitis admitted to the intensive care unit: a multicenter study

Correspondence: Nicolas Lerolle -

Annals of Intensive Care 2016, 6(Suppl 1):O65

Introduction Data regarding patients with the most severe manifestations of ANCA-associated vasculitis admitted to intensive care unit are scarce. The objective of the present study was to analyze the main features of ANCA-associated vasculitis patients admitted to intensive care unit for vasculitis activity and to study long-term outcome.

Patients and methods We analyzed retrospectively consecutive adult ANCA-associated vasculitis patients from 17 intensive care units admitted for active disease over a 10-year period (2002–2012). This intensive care group was compared to a group of ANCA-associated vasculitis patients admitted to a nephrology department within the same period. ANCA-associated vasculitis patients admitted to intensive care unit or nephrology department for other reasons than disease activity (i.e., complication of treatments such as sepsis) were excluded.

Results A total of 88 patients were included in the intensive care group and 55 patients in the non-intensive care group. The mean age of patients of the intensive care group was 59 ± 16.1 years, 54 patients had granulomatosis with polyangiitis (61 %), 32 had micropolyangiitis (36.4 %), and two had eosinophilic granulomatosis with polyangiitis (2 %). Proteinase-3 ANCAs were predominant (60 %). Seventy-four patients had newly diagnosed ANCA-associated vasculitis, and 14 were cared for a relapse. When compared to the non-intensive care group, the intensive care group had a higher Birmingham Vasculitis Activity Score (23 ± 8 vs 16 ± 4, p < 0.01) and a higher frequency of lung (diffuse alveolar hemorrhage in 64 vs 11 %, p < 0.001), heart, ear–nose–throat and central nervous system involvement. Respiratory assistance, renal replacement therapy and vasopressor amines were used in 72, 54 and 27 % of intensive care patients, respectively. Renal replacement therapy was required in 44 % of the non-intensive group patients (p = 0.254 for comparison with intensive care patients). In the intensive care group, all patients were treated with steroids, 86 % received cyclophosphamide, and 51 % had plasma exchanges as remission induction regimen. Rituximab was used in 3 % in the intensive care group. Fourteen patients (15.9 %) died during intensive care stay. The need for mechanical ventilation, the use of vasopressors and the occurrence of at least one infectious event were significantly associated with intensive care mortality in univariate analysis. The use of plasma exchange was not significantly associated with mortality change (OR 0.69 [0.24–1.98]). After adjustment, only the occurrence of an infectious event was associated with intensive care mortality in the multivariate analysis (OR 11 [1.8–64.2] p < 0.01). Despite a higher in-hospital mortality rate (19.3 vs 3.6 %, p = 0.007), the long-term mortality of the intensive care group was not different from that of the non-intensive care group (respectively, 1-year mortality, 20.4 vs 17.0 %, p = 0.372, and see Fig. 13).

Fig. 13
figure 13

See text for description

Conclusion This study represents the largest cohort of ANCA-associated vasculitis patients with active disease requiring intensive care. Even if respiratory failure appears as the main cause of intensive care admission, we identified sepsis as the main risk factor for intensive care mortality. Our results suggest that surviving patients have a long-term prognosis that is not significantly different from patients whose disease does not require intensive care admission.

Competing interests None.

O66 Continuous renal replacement therapy versus intermittent hemodialysis: impact on mortality and renal recovery

Correspondence: Anne Sophie Truche -

Annals of Intensive Care 2016, 6(Suppl 1):O66

Introduction The optimal initial modality of renal replacement therapy remains controversial. Available literature is limited by several biases such as lack of statistical power, the absence of standardization of renal replacement therapy initiation criterion, lack of information concerning dialysis dose and a high rate of crossover between modalities. Moreover, some area of uncertainties remains including optimal modality of renal replacement therapy during shock and their impact on long-term renal outcome. Additional studies including a sufficient number of patients were necessary.

The purpose of our study was to compare the prognostic impact on both techniques, adjusting for the probability of being treated by continuous renal replacement therapy or intermittent hemodialysis a particular day.

Patients and methods Patients of a French prospective high-quality observational multicenter cohort were included if they underwent at least one renal replacement therapy session between 2004 and 2014. The day of renal replacement therapy initiation for a patient was his study inclusion day.

Baseline and daily patients’ characteristics differences between treatment groups were taken into account by using a marginal structural Cox model, allowing drawing a causal conclusion in observational longitudinal data analysis. The composite primary endpoint was 30-day mortality and dialysis dependency. Following subgroup analyses were planned in the experimental design: chronic renal or heart disease; liver cirrhosis; diabetes; hypertension; hemodynamic status at renal replacement therapy initiation defined according to SOFA hemodynamic component (strictly inferior to 3, and equal or above 3); invasive mechanical ventilation at inclusion; and extent of daily weight gain between intensive care unit admission and inclusion, defined as the upper quartile of the study population.

Secondary objective was to assess 6-month prognosis in term of mortality and persistent renal dysfunction by a logistic model with inverse probability of treatment weighting, considering the treatment most received in the first 7 days.

Results Among 18,834 patients of the cohort 1360 were included, with, respectively, 544 (40.0 %) and 816 (60.0 %) patients initially treated by continuous therapies and intermittent hemodialysis. Median age was 65 [interquartile range: 54; 76], and 204 (15.0 %) patients suffered from chronic kidney disease, 235 (17.3 %) from chronic heart failure and 327 (24.0 %) from diabetes. A medical condition was present among 77.3 % of patients. Median initial SOFA score was 10 [interquartile range: 7; 13]. Invasive mechanical ventilation was required at renal replacement therapy initiation in 930 (68.4 %) patients. Both therapies were administered in 265 (19.5 %) patients.

At day 30, 539 (39.6 %) patients died. Among survivors, 150 (23.8 %) still required renal replacement therapy. There was no difference between initial modalities for the primary endpoint (HR 1.00; 95 % CI 0.77–1.29; p = 0.97).

In patients with higher weight gain at renal replacement therapy initiation, mortality and dialysis dependency were significantly lower with continuous therapies (HR 0.54; 95 % CI 0.29–0.99; p = 0.05). Conversely, this technique was deleterious in patients without hemodynamic instability (HR 2.24; 95 % CI 1.24–4.04; p = 0.01). In patients with shock at inclusion, no improvement in prognosis was noted with continuous therapies.

Six-month mortality and persistent renal dysfunction were not influenced by initial renal replacement therapy modality (OR 0.81; 95 % CI 0.45–1.46; p = 0.48).

Discussion The marginal structural model allowed inclusion of a large number of patients while minimizing biases resulting from our observational design. Hence, variables associated with choice of treatment and outcome was taken into account by this model. Switch between treatment modalities was considered by an as-treated analysis. Lastly, the model was adjusted for timing of renal replacement therapy initiation and dialysis dose.

Main limitation of the study is that even if all known confounding factors were introduced in the model, unknown confounding factors might not have been considered.

Conclusion Continuous renal replacement therapy did not improve 30-day and 6-month patients’ prognosis. It seems beneficial for patients with fluid overload, but might be deleterious in the absence of hemodynamic failure.

Competing interests None.


1. Kidney Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Work Group. KDIGO Clinical Practice Guideline for Acute Kidney Injury. Kidney inter., Suppl. 2012; 2: 1–138.

O67 Hospital survival and need of renal replacement therapy with respect to AKI duration in critically ill patients: results of a multicenter cohort study

Correspondence: Michaël Darmon -

Annals of Intensive Care 2016, 6(Suppl 1):O67

Introduction The classical dichotomy between functional acute kidney injury (AKI) and intrinsic AKI has recently been challenged. Thus, authors suggested that transient and persistent AKI in critically ill patients might share similar pathophysiological mechanisms. A recent study confirmed these findings [1]. However, this previous study suffers a lack of statistical power, and time dependency for renal recovery was unaccounted for.

The primary objective of our study was to assess prognostic impact of AKI duration. Secondary objective was to assess relationship between renal recovery at specific time frame and need for renal replacement therapy.

Patients and methods We performed a retrospective analysis of a large prospective multicenter cohort database involving 20 French ICUs. Every adult patient with AKI at ICU admission according to the KDIGO definition was included.

Renal recovery was defined by a decrease in one stage or more according to the KDIGO classification.

Results are reported as medians (interquartile range, IQR) or numbers (%). Variable of interest was day-28 mortality, and influence of renal recovery was assessed after adjustment in a cause-specific hazard model taking time dependency of this variable into account. Discharge alive from the ICU was used as a competing event.

Results Overall, of 18,684 patients of our database, 5229 had an AKI at ICU admission and were included in this study. 3167 patients were of male gender (60.6 %), median age was of 70.5 [57.7–78.8], and initial severity assessed by SAPS 2 was of 51 [39–70]. Main reasons for ICU admission were shock (41.0 %) and acute respiratory failure (21.2 %).

AKI severity at admission according to KDIGO classification was stage 1 in 2453 patients (46.9 %), stage 2 in 1178 (22.5 %) and stage 3 in 1598 (30.6 %). According to our definition, renal recovery occurred in 65.7, 62.6 and 45.7 % of patients with AKI stages 1, 2 and 3, respectively.

Day-28 mortality was, respectively, of 22.8, 27.8 and 26.3 % for patients with AKI stages 1, 2 and 3. After adjustment for confounders, renal recovery was associated with a favorable outcome (cause-specific HR 0.53; 95 % CI 0.45–0.62). Results were unchanged when assessing influence of a renal recovery defined by renal recovery at day 3.

In patients alive at day 8, rate of renal replacement therapy in patients with persistent and transient AKI defined at various time frames is reported in Fig. 14.

Fig. 14
figure 14

See text for description

Conclusion These data confirm short renal dysfunction reversibility to be potent prognostic marker in AKI patients and therefore to be clinically relevant in defining severity of AKI. Additionally, our results confirm a definition of persistent AKI based upon renal recovery between day 3 and day 7 to be potent predictor for need of RRT. Distinction between transient and persistent AKI might therefore be clinically relevant surrogate outcome variable for diagnostic testing in critically ill patients with AKI.

Competing interests None.


1. Perinel et al. Crit Care Med 2015.

O68 Early versus delayed initiation of renal replacement therapy in septic acute kidney injury: a retrospective study

Correspondence: Antoine Dewitte -

Annals of Intensive Care 2016, 6(Suppl 1):O68

Introduction Acute kidney injury (AKI) is a common complication in critically ill patients. Its incidence in septic shock is around 50 % (1). Among patients with severe AKI, 50–70 % receives renal replacement therapy (RRT). However, the optimal timing of RRT initiation in patients with severe AKI remains controversial (2). This study compared the effect of early versus delayed RRT in patients with septic AKI.

Patients and methods This retrospective study included all patients hospitalized in our intensive care unit (ICU) between January 2008 and December 2013 for severe sepsis or septic shock complicated by AKI requiring RRT. Exclusion criteria were age <18 years, end-stage renal disease and RRT initiation before ICU admission. Reference creatinine was retrieved from blood samples at steady state during the year preceding admission. AKI was defined and classified according to the KDIGO criteria (both diuresis and serum creatinine). Early group was defined as an initiation of RRT within 12H after admission or AKI stages 0–1 at initiation. Delayed group was defined as an initiation of RRT >12H after admission and AKI stages 2–3 at initiation. The primary outcome was in-hospital mortality.

Results A total of 312 patients were treated by RRT during the study period. Two hundred and four patients suffered from severe sepsis or septic shock, and 169 patients were included in the study. In the early group (n = 79), SOFA score (12 vs 10, P = 0.01) and serum creatinine level (174 μmol/L [123–283] vs 154 μmol/L [102–229], P = 0.003) were higher at ICU admission. One hundred and fifty-eight (93 %) patients were receiving norepinephrine and mechanical ventilation when EER was initiated. The KDIGO score at RRT initiation was higher in the delayed group (3 vs 2; P < 0.0001) on both creatinine and diuresis criteria. The crude hospital mortality was 58 % in the early group versus 54 % in the delayed group (P = 0.6). A Cox regression model showed a better probability of survival adjusted to the SOFA score at ICU admission in the early group (P = 0.003, Fig. 15). Risk factors for mortality analyzed by logistic regression were the PaO2/FiO2 ratio and lactate level at initiation of RRT. Survivors had a similar renal recovery in both groups, but the hospital length of stay was longer in the delayed group (51 vs 34 days, P = 0.05).

Fig. 15
figure 15

See text for description

Conclusion Our study reflecting our clinical practice reveals a better outcome when RRT is early initiated in ICU according to its timing after admission and AKI severity. The deleterious effects of fluid overload and hypoximia in case of AKI should be precise in further studies.

Competing interests None.


1. Quenot JP, Binquet C, Kara F, Martinet O, Ganster F, Navellou JC, Castelain V, Barraud D, Cousson J, Louis G, Perez P, Kuteifan K, Noirot A, Badie J, Mezher C, Lessire H, Pavon A: The epidemiology of septic shock in French intensive care units: the prospective multicenter cohort EPISS study. Crit Care 2013, 17:R65.

2. Karvellas CJ, Farhat MR, Sajjad I, Mogensen SS, Leung AA, Wald R, Bagshaw SM: A comparison of early versus late initiation of renal replacement therapy in critically ill patients with acute kidney injury: a systematic review and meta-analysis. Crit Care 2011, 15:R72.

O69 Renal recovery after severe acute kidney injury in critically ill myeloma patients

Correspondence: Adrien Joseph -

Annals of Intensive Care 2016, 6(Suppl 1):O69

Introduction Despite substantial improvements in the management of multiple myeloma, renal failure remains an important burden which tremendously impairs prognosis. The purpose of this study was to describe the characteristics and prognostic factors for renal recovery in myeloma patients admitted to the intensive care unit (ICU) for acute kidney injury (AKI) stage 3 treated with renal replacement therapy (RRT).

Patients and methods A retrospective single-center cohort study was performed, including consecutive myeloma patients admitted to our ICU between 01/01/2007 and 01/09/2015 and treated with RRT. Data and follow-up were abstracted from the medical charts. Patients were evaluated 60 days after ICU discharge and divided into three groups: alive without dialysis, alive and dialysis-dependent or deceased. A univariate analysis was performed to identify factors associated with being alive without dialysis 60 days after ICU discharge.

Results Fifty-five patients were included in the study. Median age was 63 (inter-quartile range 58–70) years, and 33 (60 %) were male. Patients were admitted to the ICU 3 (0–5) years after the diagnosis of myeloma, after 1 (1–3) line of chemotherapy. Twenty-two had already been treated with autologous stem cell transplantation, at least once. The baseline renal function before ICU admission, estimated by the glomerular filtration rate (GFR), was 58 (41–78) ml/min. The median SOFA score at day 1 was 6 (4–8). The three main reasons for ICU admission were AKI (33, 60 %), sepsis (21, 38 %) and acute pulmonary edema (13, 55 %). During ICU stay, 17 (31 %) patients required invasive mechanical ventilation, 13 (27 %) received vasopressors, and RRT was implemented in all patients (55, 100 %) during 13 (5–42) days. The median ICU and hospital length of stay were 13 (5–43) days and 24 (14–35) days, respectively. Forty (73 %) patients were alive at hospital discharge.

At day 60 after ICU discharge, 22 (41 %) patients were alive without dialysis, 19 (36 %) had died, and 12 (23 %) were still undergoing dialysis. Among the 22 patients who recovered, the median time of dialysis was 6 (2–18) days. These patients recovered their previous renal function with a median GFR of 65 (25–74) ml/min at day 60 (p = 0.70).

By univariate analysis, two factors were associated with a decreased likelihood of renal recovery at day 60: a history of autologous stem cell transplantation, odds ratio (OR) 0.27, 95 % confidence interval (0.08–0.93), p = 0.04, and a proteinuria at ICU admission >380 mg/mmol creatininuria, OR 0.22, 95 % CI (0.06–0.71), p = 0.01.

None of the other variables related to the hematological malignancy or to the ICU stay were associated with renal recovery at day 60.

Discussion This study is the first to assess renal recovery in critically ill myeloma patients with severe AKI, in the era of bortezomib therapy. First of all, survival and renal recovery in our study are much higher than previously reported. About two-third of the patients were alive at day 60, and 41 % had recovered their baseline renal function. We identified two prognostic factors which can be considered as surrogates of more aggressive malignancies.

Conclusion AKI stage 3 in critically ill myeloma patients was associated with a hospital mortality <30 %. The main cause of AKI is likely to influence the likelihood of renal recovery. Patients with more aggressive malignancies (history of autologous transplantation, higher proteinuria) had a poorer renal prognosis.

Competing interests None.

O70 Impact of proactive nurse participation in ICU family conferences: a Mixed-Method Study

Correspondence: Maïté Garrouste-Orgeas -

Annals of Intensive Care 2016, 6(Suppl 1):O70

Introduction Poor communication is associated with anxiety, distress and posttraumatic stress-related symptoms in families of both survivors and nonsurvivors. Few studies have assessed the impact on everyday medical decisions of better communication, most notably delivered by a multidisciplinary team. Our objectives are to investigate family perceptions of having a nurse participate in family conferences and to assess the psychological well-being of the same families after ICU discharge.

Patients and methods We designed a parallel-group randomized trial comparing family conferences with versus without the proactive participation of a nurse in family conferences. The study has a mixed-method design with a qualitative study embedded in a single-center randomized study. We included one family member for each consecutive patient who received more than 48 h of mechanical ventilation in the ICU. We scheduled planned proactive participation of a nurse in family conferences led by a physician. In the control group, conferences were led by a physician without a nurse. We evaluated family conferences on days 1 and 3, weekly, during bad news or end-of-life. Audits by checklists were conducted for all conferences. All ICU physicians and nurses received training of communication by role players. Physicians and nurses used conferences guides created before the study. We used the NURSE acronym, the Ask-Tell-Ask concept and the VALUE strategy depending on the type of conferences. The primary outcome was the score on the Impact of Events Scale-revised (IES-R) (for posttraumatic stress-related symptoms) completed by the family member 3 months after the death or discharge of the patient. Secondary outcomes were symptoms of anxiety and depression in family members 3 months after patient death or ICU discharge. We conducted a qualitative evaluation of perceptions by family members of nurse participation in the family conferences at ICU discharge using interview guides. Each interview was transcribed verbatim and evaluated using interpretative phenomenological analysis.

Results Of the 172 eligible family members, 100 (60.2 %) were randomized; among them, 88 underwent semi-structured interviews at ICU discharge and 86 completed the Peritraumatic Dissociative Experiences Questionnaire at ICU discharge and the Impact of Event Scale and Anxiety Depression Questionnaire 3 months later. The number of conferences was 188: 90 on day 1; 57 on day 3; 24 on day 7 and then weekly; and 17 for patients at the end-of-life. There were 96 conferences in the intervention group and 92 in the control group. The audits by checklists showed no significant between-group differences in conference contents on days 1, 3, 7 and then weekly or at the end-of-life. Conference duration was about 30 min at all time points.

Primary and secondary outcomes

The median [IQR] IES-R score was not significantly different between the control and intervention groups (24 [12.5–45] and 21 [9–23], respectively; P = 0.24). IES-R scores indicated posttraumatic stress-related symptoms in 52.2 % and 50 % of family members in the control and intervention groups, respectively (P = 0.83). Although the intervention group had significant lower anxiety (4 [1–9] vs 8 [4.5–12], p = 0.01) and depression (2 [0–6] vs 5.5 [1–11.5], p = 0.04) subscale scores, the prevalence of severe anxiety [14 (33.3 %) vs 23 (52.3 %), p = 0.08] and depression symptoms [10 (23.8 %) vs 17 (38.6 %), p = 0.14] (HADS subscale scores >8) did not differ significantly between the two groups.

Qualitative findings

The qualitative data indicated that the families valued the principle of the conference itself. Perceptions of nurse participation clustered into four main themes: trust that ICU teamwork was effective (50/88, 56.8 %), trust that care was centered on the patient (33/88, 37.5 %), trust in effective dissemination of information (15/88, 17 %) and trust that every effort was made to relieve anxiety in family members (12/88, 13.6 %). In conferences without nurse participation, two themes were identified: satisfaction with the information given by the physician alone (15/45, 33.3 %) and satisfaction with the information given by the nurse in the patient’s room (7/45, 15.5 %).

Conclusion In conclusion, our study adds new insights about the role for repeated, routine, structured family conferences in the ICU. The families greatly appreciated the conferences themselves. Although the participation of a nurse in the conferences did not change significantly the post-ICU burden, it was considered positive by the family members. Our study offers some guidance on how to conduct family conferences in each specific clinical situation.

Competing interests None.

O71 Duration of platelet storage and outcome of critically ill patients

Correspondence: Cécile Aubron -

Annals of Intensive Care 2016, 6(Suppl 1):O71

Introduction Platelets (PLTs) can been stored up to 5–7 days depending on the country’s transfusion policies. Changes in platelets structure and functional variables during the storage have been described in vitro. Although increase in duration of PLTs storage could improve PLTs availability in remote area and decrease waste of PLTs, information on the impact of platelets storage duration is limited and controversial. This study aims to determine whether the storage duration of PLTs transfused to critically ill patients is associated with patient outcomes.

Patients and methods We performed a retrospective analysis of patients admitted to two intensive care units (ICUs) in tertiary hospitals from 2008 to 2014. Comparison of outcomes of patients receiving only PLT with a maximum age higher than 5 days PLT versus patients receiving only PLT with a maximum age <5 days has been made. Outcome variables were hospital mortality and ICU-acquired infection (bacteraemia and/or bacteriuria). Associations between PLT age and outcomes were modelled using multiple logistic regression and linear regression.

Results Among 2250 ICU patients who received one or more PLTs, storage duration of all transfused PLTs was available for 1430 (64 %). These patients had a mean age of 57.5 ± 18.1 years, mean APACHE 3 score at admission was 65.8 ± 29.0, and 964 (67.4 %) were male. The median PLT age was 4 days (IQR 2, range 2–5). In-hospital mortality was 22.1 %, median ICU LOS in survivors 5.0 days (IQR 8.7) and infection rate 13.2 %. When comparing patients who received PLTs with a maximum age of 5 days with a maximum age <5 days, there were no significant differences in demographics, pre-ICU illness severity or number of transfused products. There were also no differences in mortality (21.6 vs. 22.8 %, p = 0.59), bacteraemia (7.5 vs. 6.4 %, p = 0.44) or LOS (median 4.9 vs. 5.1 days, p = 0.57). After adjusting for confounders, age of PLTs was not independently associated with mortality (odds ratio [OR] 1.04, 95 % confidence interval [CI] 0.78–1.40) or infection (OR 1.31, 95 % CI 0.88–1.97). Similar results were found when considering separately bacteraemia and bacteriuria or when analysing the 639 patients who were transfused with only one PLT unit.

Conclusion In this large retrospective study, storage duration of PLTs is not associated with an increased risk of mortality or infection in critically ill patients. Our results support further research investigating safety and benefit to extend PLTs storage duration.

Competing interests None.


1. Inaba K, Branco BC, Rhee P, Blackbourne LH, Holcomb JB, Spinella PC, Shulman I, Nelson J, Demetriades D. Impact of the duration of platelet storage in critically ill trauma patients. J Trauma. 2011 Dec;71(6):1766–73.

2. Triulzi DJ, Assmann SF, Strauss RG, Ness PM, Hess JR, Kaufman RM, Granger S, Slichter SJ. The impact of platelet transfusion characteristics on posttransfusion platelet increments and clinical bleeding in patients with hypoproliferative thrombocytopenia.Blood. 2012 Jun 7;119(23):5553–62.

O72 REA-C-SUR safety culture in intensive care in France: is there a link with morbi-mortality conferences?

Correspondence: Cédric Bretonnière -

Annals of Intensive Care 2016, 6(Suppl 1):O72

Introduction Improving the safety of care is a challenge for the health systems. Development of a safety culture (SC) is one of the objectives.

Morbidity–mortality conferences (MMC) are one tool that could help in improving SC.

The main objective of the national multicenter REA-C-SUR project is to measure the SC in ICUs in France. The secondary objective is to establish a link between SC and the characteristics of MMR.

This project was initiated by the national group ‘RMM, Qualité et Sécurité des Soins en Réanimation.’ It is funded by a national grant AO PREPS 2012.

Materials and methods SC was measured among professionals (doctors, managers, nurses, etc.) through the questionnaire (Hospital Survey on Patient Safety Culture, HSOPS) developed by AHRQ in the USA and validated in French. This questionnaire, through 40 items, explores ten dimensions of SC. A dimension is developed if the score is above 75 % and “to be improved” if the score is <50 %.

For each unit, MMCs’ organizational characteristics were collected.

Results From September 2013 to September 2014, 64 units (adults or pediatric) were included from French hospitals as follows: 40 teaching, 23 non-teaching and one private sector. A total of 36,149 patients, in 2012, were admitted.

Participation rate was excellent: 3692 questionnaires filled out of 4811 (77 %). SC overall is really undeveloped. The scores for perception by professionals in the management support for security (dim 9) or teamwork between departments (dim 10) are extremely bad: <50 % for almost all units. The dimension “non-punitive responseis also very undeveloped (Fig. 16).

Fig. 16
figure 16

See text for description

There is a statistical link between MMR and SC. The more the MMR are structured, the better the scores for the overall perception of safety and support of management. There is also a link between nurses’ involvement and reporting of adverse events, learning organization and human resources. Finally, the presence at MMCs, of an external participant, seems to improve communication openness.

Conclusion This prospective, multicenter study is the first in ICUs, in Europe. It might help in improving SC that is to date really low.

Competing interests None.

O73 Survival of critically ill solid cancer patients: results of a retrospective multicentre study

Correspondence: François Vincent -

Annals of Intensive Care 2016, 6(Suppl 1):O73

Introduction We previously reported factors associated with day-120 outcome in ICU survivors among 1053 patients with solid cancers hospitalized in intensive care unit (ICU) [1]. We performed additional analysis to assess hospital mortality and factors associated with day-120 mortality in the overall population of patients.

Patients and methods We conducted an international, multicenter, retrospective study. All solid cancer patients, except those admitted after scheduled surgery, admitted between 2006 and 2011 in the five participating ICUs were included. Variables of interest were in-hospital and day-120 post-ICU mortality. We performed logistic regression analyses to identify variables statistically significantly associated with hospital mortality and day-120 post-ICU discharge mortality. A prediction score (Oncoscore) was then built and evaluated.

Results A total of 1053 patients were included. The median age was of 63 years (54–71), and 66.8 % of the patients were of male gender. The principal underlying cancers were lung in 202 (19.2 %), colonic in 178 (16.9 %), breast in 129 (12.5 %), and head and neck in 125 (11.9 %).

In-ICU, in-hospital, and day-120 post-ICU discharge mortalities were of, respectively, 41.3, 60.7, and 65.8 %. Factors independently associated with day-120 mortality were systemic extension of the cancer (OR 2.54; 95 % CI 1.87–3.45) and organ support including need for mechanical ventilation (OR 2.54; 95 % CI 1.80–3.45), renal replacement therapy (OR 1.54; 95 % CI 0.99–2.38), or vasopressors (OR 2.35; 95 % CI 1.66–3.29). Conversely, when compared to lung cancer, other types of tumours were found to be protective from poor outcome with odds ratio ranging from 0.25 to 0.52 (P < 0.05 for each).

A prediction model (Oncoscore) was then built ranging from 0 to 11. In the studied population, Oncoscore was 6 (3–7). The overall model area under ROC curve in predicting day-120 outcome was fair (0.74; 95 % CI 0.71–0.77). A score of 4 was found to be sensitive of poor outcome (sensitivity 0.84), while a score of 8 was specific (specificity 0.92). According to these cut-offs, day-120 mortality was of 40 % (n = 111), 70 % (n = 363) and 87 % (n = 216) with Oncoscore <4, of 4–7 and >7, respectively. Cumulative survival is reported in Fig. 17.

Fig. 17
figure 17

See text for description

Conclusion Critically ill patients with solid cancer have a meaningful day-120 survival. Conversely to critically ill patients with haematological malignancy underlying malignancy and extension of the disease seems to be independently associated with day-120 mortality. The Oncoscore developed in this study, although performing fairly in the studied population remains to be validated in an external population of solid cancer patients requiring ICU admission.

Competing interests None.


1. Abstract FC091, S175, SRLF 2014.

O74 Encouraging results of the French controlled donation after circulatory death Maastricht category III Program

Correspondence: Samuel Gay -

Annals of Intensive Care 2016, 6(Suppl 1):O74

Introduction Organs from donors following circulatory death (DCD) have undoubtedly contributed to the expansion of donor organ pool. Recently, the French regulation agency, Agence de BioMédecine (ABM) and the Intensive Care Societies (SRLF/SFAR) made recommendations and authorized “controlled” donation after circulatory death (cDCD) Maastricht category III. Here, we describe the experience of one of the first authorized pilot centers in France.

Patients and methods From December 2014 to September 2015, all cDCD data were prospectively collected, in conformity with the ABM protocol [1].

The different steps of our local procedure were: (1) Withdrawal of life-sustaining therapy (WLST) decision was discussed by intensive care staff including external second opinion. This decision was independent of any discussion to donate organs for transplantation. (2) The relatives were involved in the decision of WLST. The healthcare provider ensured the understanding and acceptance of the decision. (3) Only after WLST decision, the local organ procurement coordination was contacted to indicate that the patient might be eligible for cDCD. (4) Then, a representative assessed the patient for potential donation (health security). (5) If the patient was suitable for donation, a second interview with relatives was done to research the non-opposition for donation. (6) Evaluation of kidney and liver function. (7) WLST determined by the ICU staff and relatives. (8) Declaration of death after circulatory arrest. (9) Percutaneous introduction of in situ regional normothermic recirculation (RNR) in intensive care. (10) Transfer in the operating room for organ procurement.

This prospective, single-center, descriptive study reports the first data from our cohort of cDCD.

Results From 01/12/2014 to 31/08/2015, among 742 intensive care admissions, four potentials donors were identified (three anoxic encephalopathy and one severe traumatic brain injury). All four had no opposition for donation and were eligible for the procedure. Titrated sedation and extubation were always used for WLST. Death occurred in each case in <3 h. There was one cannulation failure of the RNR. For the three other procedures, six kidneys were explanted and grafted. There was no delayed graft function. One liver was explanted and grafted with success. During the same period, our team accomplished 17 donation after brain death and six uncontrolled DCD (Maastricht II).

Conclusion The understanding of the cDCD protocol by relatives was good. The discussion of donation options was frequently initiated by family request for information. There was no procedure failure due to delayed cardiac arrest or prolonged hypoperfusion period. These results could be explicated by our procedures for WLST with titrated sedation, analgesia and comfort care. Our policies and procedures for WLST are the same for both donation and non-donation cases. Our team met technical difficulties for percutaneous RNR. This percutaneous approach needs to be evaluated in the future. cDCD represents 18 % of our recovery activity for transplantation. After an intense preparation phase, essential to the proper understanding of the protocol, the establishment of a cDCD activity in our center led to six kidney and one liver transplants. On this short cohort, we had no grafts dysfunction. These data showed promising results concerning cDCD program but need further evaluation (Fig. 18).

Fig. 18
figure 18

See text for description

Competing interests None.


1. Antoine C, Mourey F, Prada-Bordenave E: How France launched its donation after cardiac death program. Ann Fr Anesth Réanimation 2014, 33:138–143.

O75 Clinical significance of cardiac troponin I release in severe trauma patients

Correspondence: Maxens Decavèle -

Annals of Intensive Care 2016, 6(Suppl 1):O75

Introduction Elevated plasma levels of cardiac troponin I (cTn-I) have been described in about 10 % of cases in the early phase of severe multiple trauma patients. However, apparently specific for myocardial cell membrane disruption, the mechanisms of its increase are still poorly understood. Moreover, if cTn-I elevation appears as an independent prognostic marker in various critical situations, its impact on severe trauma outcome in the intensive care unit (ICU) remains controversial. The main objectives of our study were first to determine the incidence and factors influencing cTn-I release in severe multiple trauma and secondly to assess its prognosis on outcomes in intensive care unit.

Patients and methods We performed a retrospective analysis from a prospective, multicenter, regional scientific database (Traumabase®), of all patients admitted to the Beaujon University Hospital ICU between January 2011 and December 2013. Only patients for whom a troponin assay was performed on arrival were included for analysis (immunological methods of chemiluminescence-ARCHITECT STAT Troponin-I®, positivity threshold 0.05 ng/ml). If more than one value were available, the highest troponin value over the first 24 h was retained for analysis. The primary endpoint was the cTn-I elevation and the ICU mortality. Continuous variable is expressed as median and interquartile range and categorical variables as absolute value and relative frequencies. Univariable and multivariable logistic regression analysis was performed for cTn-I elevation and ICU mortality.

Results A total of 1029 patients were included (age 38 ± 18; Injury Severity Score 19 ± 14). Pre-hospital variables associated with cTn-I elevation in multivariable analysis are reported in Table 8. In a multivariable model including the Injury Severity Score and the Simplified Acute Physiology Score and the presence of shock and cardiac arrest before admission, initial cTn-I elevation is not associated with ICU mortality.

Table 8 See text for description

Conclusion In this series of more than 1000 patients, the initial serum concentration of cTn-I does not appear as an independent predictor of mortality in the ICU. Its initial elevation seems rather to reflect the severity of the trauma. However, its association with various pre-hospital variables may highlight the complex mechanisms of its release.

Competing interests None.

O76 The modified Glasgow prognostic score is helpful in screening lung cancer patient at risk of hospitalization at the emergency department

Correspondence: Julie Gorham -

Annals of Intensive Care 2016, 6(Suppl 1):O76

Introduction We previously found predictive factors for hospital admission and death during hospitalization in lung cancer patients consulting at the emergency department. Systemic inflammation is a prognostic factor in cancer patients and can be studied by inflammation-based prognostic scores [1], such as the modified Glasgow prognostic score (Table 9), a combination of C-reactive protein and albumin, biological parameters easily obtained and measured in routine. This score was initially described in 2003 in patients with inoperable lung cancer and proved to be a prognostic factor for survival [2]. Then, it was studied for its prognostic value in patients with operable cancer and receiving chemo-/radio-therapy. A predictive role for response to antineoplastic treatment was also suggested [3]. The goal of our study was to determine whether systemic inflammation measured using the modified Glasgow prognostic score could improve the predictive value of our previous models, a model for hospitalization and a model for death during hospitalization.

Table 9 The modified Glasgow prognostic score

Patients and methods We conducted a retrospective study including all patients with lung cancer consulting at the emergency department of a cancer hospital between January 1, 2008, and December 31, 2010. In order to assess whether the Glasgow score improves the prediction of the two published models (predicting hospitalization and death during hospitalization), we have refitted the two models in the cases with the Glasgow score available, dropped the variables one per one with a nonsignificant p value, and then added the Glasgow score to both models.

Results Of the 548 emergency department visits, CRP and albumin needed for calculating the modified Glasgow prognostic score were available in 291 cases. Of these, 94 (32 %) involved patients with clinically documented infection. The 291 visits resulted in 222 hospitalizations. The majority of patients had a modified Glasgow prognostic score of one and a stage IV cancer. In multivariate analysis based on the model originally developed, the modified Glasgow prognostic score is an independent predictor of hospital admissions (odds ratio 2.72 per one-unit increase; p value <0.0001) as well as arrival by ambulance (odds ratio 25.93; p value <0.0001) and the presence of physical signs associated with the chief complaint (odds ratio 2.83; p value = 0.001). One hundred and eighty-one consultations (first visit only) were considered for determining prognostic factors for death during hospitalization. The modified Glasgow prognostic score is an independent predictor of death during hospitalization (odds ratio 2.95 per one-unit increase; p value <0.001) as well as arrival by ambulance (odds ratio 19.03; p value = 0.0002). After stratifying patients into two groups according to the presence of an infection or not, this score added significant information in the two strata (p value 0.003 and 0.001, respectively).

Discussion This is an original study. To our knowledge, this score has never been evaluated in emergency departments in cancer patients or even in a general population.

Conclusion The modified Glasgow prognostic score is an independent predictive factor for hospitalization and death during hospitalization in patients with lung cancer consulting at the emergency department.

Competing interests None.


1. Proctor MJ, Morrison DS, Talwar D, et al. A comparison of inflammation-based prognostic scores in patients with cancer. A Glasgow Inflammation Outcome Study. Eur J Cancer 2011; 47: 2633–41.

2. Forrest LM, McMillan DC, McArdle CS, Angerson WJ, Dunlop DJ. Evaluation of cumulative prognostic scores based on the systemic inflammatory response in patients with inoperable non-small-cell lung cancer. Br J Cancer 2003; 89: 1028–1030.

3. McMillan DC. The systemic inflammation-based Glasgow Prognostic Score: a decade of experience in patients with cancer. Cancer treat Rev 2013; 39: 534–40.

O77 Relationship between body mass index classes and massive transfusion needs in trauma patients and predictive performance of the TASH score in obese and non-obese populations: a retrospective study on 910 trauma patients

Correspondence: Audrey De Jong -

Annals of Intensive Care 2016, 6(Suppl 1):O77

Introduction The early detection of patients at risk of massive transfusion (MT) is challenging in bleeding trauma patients [1]. Mortality has been shown to be higher in severely injured obese patients compared with non-obese patients. However, to our knowledge, the risk of MT has never been assessed in obese population. Moreover, although the Trauma-Associated Severe Haemorrhage (TASH) score is a robust scoring system for predicting MT requirements [2], its predictive performance has never been validated in obese patients. The main objective of the study was to compare the MT rate in trauma patients according to obesity status. The secondary objectives were to assess and revalidate the TASH score in predicting MT in population of obese and non-obese trauma patients and to use a grey zone approach to providing relevant thresholds to optimize the ability to predict MT in clinical practice.

Patients and methods All trauma obese and non-obese patients admitted in a Level I Regional Trauma Center were included. Were excluded patients who died immediately after admission or not directly admitted or with important clinical data missing in the database. MT rate and TASH score were assessed in obese [body mass index (BMI) ≥30 kg/m2] and non-obese (BMI <30 kg/m2) patients. Logistic regression was performed to assess the relationship between obesity and MT, before and after adjusting for ISS score. The odds ratio (OR) was provided for obesity status and for an increase of 5 units of BMI with the 95 % confidence interval (CI). Received operating characteristic (ROC) curve analysis was performed for the TASH score in predicting MT in the obese and non-obese groups. Thereafter, thresholds analysis was carried out from the ROC curves using the grey zone method.

Results Between January 2008 and December 2012, 119 obese trauma patients and 791 non-obese trauma patients were included. The MT rate was 10 % (94/910) in the whole population. The MT rate trended to be higher in obese patients versus non-obese patients: 15 % (18/119) versus 10 % (76/791), OR 1.68 [95 % CI 0.97–2.92], p = 0.07. After adjusting for ISS score, obesity was significantly associated with MT rate (OR 1.79 [95 % 1.00–3.21], p = 0.049). BMI as a continuous variable was also significantly associated with MT before (OR 1.24 [95 % CI 1.00–1.54], p = 0.047) and after adjusting for ISS score (OR 1.28 [95 % CI 1.03–1.60], p = 0.028). The TASH score was higher in the obese group than in the non-obese group: 8 ± 5 versus 6 ± 5. The area under the ROC curves of the TASH score in predicting MT (≥10 U of packed red blood cells in the first 24 h) was very high and comparable between the obese and non-obese groups: 0.93 (95 % CI 0.89–0.98) and 0.94 (95 % CI 0.92–0.96), respectively (p = 0.80). The grey zone ranged, respectively, from 10 to 13 and 9 to 12 in obese and non-obese patients and allowed to separate patients into three subgroups using the TASH score: low, intermediate and high risk, significantly associated with ICU and hospital stay (Fig. 19).

Fig. 19
figure 19

Rates of massive transfusion, death, length of ICU and hospital stays according to the TASH score in the obese and non-obese groups

Conclusion Obesity was associated with a higher rate of MT and a higher TASH score, which was a strong and accurate predictor of MT in both obese and non-obese populations. For the first time, the TASH score was validated in obese patients and a grey zone was established for this predictive score, allowing the MT risk to be affirmed or rejected with accuracy.

Competing interests None.


1. Spahn DR, Bouillon B, Cerny V, et al. Management of bleeding and coagulopathy following major trauma: an updated European guideline. Crit Care 2013; 17(2):R76.

2. Yücel N, Lefering R, Maegele M, et al. Trauma Associated Severe Hemorrhage (TASH)-Score: probability of mass transfusion as surrogate for life threatening hemorrhage after multiple trauma. J Trauma 2006; 60(6):1228–1236.

O78 The Eschmann stylet as first tracheal access tool during emergency difficult tracheal intubation: An initial simulation trial followed by a clinical study

Correspondence: Gilles Dhonneur -

Annals of Intensive Care 2016, 6(Suppl 1):O78

Introduction When anesthesia quality is controlled, tracheal intubation (TI) difficulty may be related to: patient’s conditions and characteristics, the environment where the maneuver is performed and the skill of the physician. In case of difficult laryngoscopy, the Eschmann stylet (ES) is proposed as a second-step intubation tool to access the trachea. We have hypothesized that when difficult emergency intubation is encountered, first-step ES (ES-1) may reduce TI delay.

Materials and methods We conducted two prospective successive evaluations. The first one was a simulation trial performed over 3 months. In our airway management laboratory, 24 physicians (12 novices and 12 skilled physicians) were involved. Each of them had to intubate a manikin allowing to simulate simple or difficult TI (Cormack grade = 1 or 3: C1 or C3) placed in three different positions and environment: standard operating table (OT), lying on the ground (LG) and with very restricted access to the head (RA). All the sequences of TI were randomized for C1, C3, OT, LG, RA and to either conventional tracheal intubation technique using the tracheal tube as first tracheal assess tool (TT-1) or ES-1. With the results of this simulation trial and obtention of ethical approval, the 12 skilled physicians conducted the clinical study over 1 year in prehospital conditions. Anesthesia and difficult airway management procedures were standardized, and the patients were randomly allocated to either TT-1 or ES-1 group. Prehospital TI characteristics were recorded. The study was preemptively dimensioned and powered to demonstrate 30 % reduction in TI delay for the patients with difficult laryngoscopy as attested by the low percentage of glottis inlet visible (POGO score) of below 30 % (Ref). In both studies, TI difficulties (anticipated and post-maneuver) and comfort were assessed.

Results A total of 276 tracheal intubations were timed and analyzed for the simulation trial. We demonstrated that ES-1 as compared to TT-1 was a powerful mean to significantly reduce TI delay of: novices when C3 was imposed in OT, LG and RA; and skilled physicians when C3 was imposed for LG and RA. For the clinical trial, 250 patients were included. Details of both groups were similar. We demonstrated in the subgroup of patients with POGO score of <30 % (n = 73) that ES-1 as compared to TT-1 significantly reduced TI delay.

Conclusion We have observed in a difficult TI simulation trial that ES-1, as compared to conventional TI, was a powerful strategy to shorten TI delay when direct laryngoscopy was difficult for novices and skilled physicians. We demonstrated that simulation results translated to clinical conditions. Our clinical data suggest that ES-1 may be useful strategy when emergency difficult TI is anticipated because of the patient’s characteristics or environment. Moreover, we believe that the recommended 2d TI attempt using TT is not necessary and the switch to ES could be proposed at the first difficult laryngoscopy.

Competing interests None.

O79 Maternal mortality risk factors for eclampsia

Correspondence: Chaigar Mohammed Cheikh -

Annals of Intensive Care 2016, 6(Suppl 1):O79

Introduction Eclampsia remains an important cause of maternal and perinatal mortality and morbidity worldwide [1]. The aim of this study was to determine the eclamptic patient’s characteristics and to identify risk factors associated with maternal mortality in patients with eclampsia.

Patients and methods We conducted a retrospective study between January 2002 and December 2013, including all patients diagnosed as eclampsia in the obstetrical intensive care unit of our hospital.

Results During this period, a total of 1130 cases of eclampsia were collected, yielding an overall incidence rate of 1.05 %. Mean age was 27 ± 6.64 (15–48) years. The major proportion of patients was nulliparous (63.4 %), 819 (72.4 %) cases were diagnosed during the antenatal period, whereas 311 (27.6 %) had a postnatal onset. Mean gestational age at onset was 34.5 ± 4.3 (21–43) weeks of gestation. HELLP syndrome was identified as the most frequent complication in our study, associated with 34.3 % of patients, while acute kidney injury, placental abruption and neurological disorders were, respectively, associated in 18.5, 17.7 and 16.5 % of cases. In our study 55 patients died, yielding a case fatality rate of 4.9 %, cerebral hemorrhage was identified as the first cause of death associated with 25.5 % of deceased patients. In univariate analysis, risk factors for maternal mortality in eclampsia were: advanced maternal age (≥ 30 years), diagnosis of eclampsia during the antenatal period, low GCS (≤ 8), the presence of a neurological deficit, oligo-anuria or jaundice, the presence of these biological disorders: low platelets count (<100,000) – ASAT >70 UI/l – ALAT >70 UI/l – LDH >600 UI/l, the occurrence of HELLP syndrome, acute kidney injury, cerebral hemorrhage, acute pulmonary edema, abruption placentae, postpartum hemorrhage, disseminated intravascular coagulation, the use of an anticonvulsant treatment other than the magnesium sulfate, subcapsular liver hematoma and long assisted ventilation (>72 h). According to logistic regression model, we have identified as risk factors for maternal mortality in eclampsia: advanced maternal age (≥30 years), low GCS (≤8), ALAT >70 UI/l, cerebral hemorrhage, acute pulmonary edema, disseminated intravascular coagulation, subcapsular liver hematoma and long assisted ventilation (>72 h).

Conclusion Eclampsia is still a major cause of maternal morbidity and mortality worldwide. It incidence remains high, especially for developing countries in need of health care. Several studies are needed to determine all the risk factors and improve maternal prognosis.

Competing interests None.

O80 Use of aminoglycosides in cirrhotic patients admitted in intensive care for severe sepsis or septic shock: short- and intermediate-term impact on renal function and mortality

Correspondence: Benjamin Zuber -

Annals of Intensive Care 2016, 6(Suppl 1):O80

Introduction Severe sepsis and septic shock are a frequent life-threatening cause of ICU admission of cirrhotic patients. Among the antibiotics frequently used in intensive care, the prescription of aminoglycosides to cirrhotic patients is controversial because of their nephrotoxicity. The aim of this study was to evaluate the use of aminoglycosides administration among cirrhotic patients admitted in intensive care unit for severe sepsis and septic shock and study their impact on mortality and renal function.

Patients and methods Cirrhotic patients with severe sepsis and septic shock were selected over a 18-year period (1997–2014) from a longitudinal prospective French multicenter database. We investigated the factors associated with the prescription of aminoglycosides started between day 1 and day 3 of admission, and we studied the modalities of administration of aminoglycosides. Using a competing risk analysis, we evaluated the impact of aminoglycosides administration on the risk of renal replacement therapy or death at day 28. Among survivors who required dialysis or with a renal failure when leaving ICU, we retrospectively collected renal status at day 90 after leaving intensive care.

Results Among the 332 cirrhotic patients admitted for severe sepsis or septic shock in ICU, 119 (35.8 %) received antibiotic therapy with aminoglycosides. In multivariate analysis, factors associated with aminoglycosides prescription were immunosuppression, SAPS II, and surgical cause for ICU admission. Administration of aminoglycosides was also significantly different between centers. All patients received a once-daily administration of aminoglycosides for a 2-day median time (IQR 1–5). Using a cause-specific model for competing risk adjusted on risk factors for death and renal replacement therapy, administration of aminoglycosides between days 1 and 3 was not associated with renal replacement therapy or death at day 28 (death: cause-specific hazard ratio CSHR 1.41, 95 % CI 0.98–2.02, p = .06; renal replacement therapy: CSHR 1.44, 95 % CI 0.67–3.1, p = .35). Among 195 patients discharged alive from the ICU, 49 patients had persistent kidney failure or needed renal replacement therapy during their ICU stay. Among these patients, 16 patients (33 %) died at day 90 and no patient required dialysis at day 90.

Conclusion The use of aminoglycosides in cirrhotic patients with severe sepsis and septic shock is frequent in intensive care units. Their administration with a daily dose is not associated with renal replacement therapy or death in this study.

Competing interests None.

O81 Diagnostic yield and therapeutic impact of liver biopsy in the intensive care unit

Correspondence: Bertrand Sauneuf -

Annals of Intensive Care 2016, 6(Suppl 1):O81

Introduction The specific diagnostic and therapeutic yield of liver biopsy in patients hospitalized in intensive care unit (ICU) is unknown. The main goal of this study was to evaluate the contribution of liver biopsy to diagnosis and treatment in this population.

Patients and methods An 8-year (2006–2014) retrospective study was conducted in medical and surgical ICU of Caen University Hospital on patients without transplanted liver who underwent liver biopsy during their management. Medical files were retrieved for data analysis.

Results Forty-six patients underwent a liver biopsy mostly by transjugular route (94 %). The liver biopsies were performed for a suspicion of cirrhosis in 20 % of cases, for acute liver failure in 26 %, for a suspicion of severe acute alcoholic hepatitis in 22 % and for a suspicion of blood disease in 28 %. Biopsy-related intraperitoneal hemorrhage occurred in two (4 %) patients requiring an embolization in one case. Liver biopsy established a specific diagnosis in 38 (83 %) patients. Therapeutic modifications followed the results of the biopsy in 18 (39 %) patients. ICU mortality was of 57 % and was not different in the group of patients with a specific diagnosis established by the biopsy and in the group of patients for which the biopsy was non-contributory (21 deaths/38 patients vs 5/7; p = 0.68).

Conclusion In this cohort highly selected by the treating physicians, liver biopsy helped to establish a specific diagnosis in the majority of the patients with a limited complications rate.

Competing interests None.

O82 Critically ill patients requiring liver transplantation: a 7-year monocentric retrospective experience

Correspondence: Benjamin Lebas -

Annals of Intensive Care 2016, 6(Suppl 1):O82

Introduction Fulminant hepatitis or acute on chronic liver failure can result in multiple organ failure. In these settings, liver transplantation is the treatment required for recovery. However, such patients have a pejorative post-transplant prognosis when compared to patients who undergo transplantation with stable chronic liver disease. This results in an unresolved ethical dilemma given the context of graft shortage. Moreover, published data or guidelines to cope with this dilemma are scarce. Our objective was to assess long-term post-transplant prognosis in patients admitted to intensive care unit (ICU) in relation to multiple organ failure and acute liver failure, and to search for possible pre-transplant predictive factors for post-transplant mortality.

Patients and methods We performed a retrospective monocentric study in 84 critically ill patients requiring liver transplantation according to standard worldwide accepted criteria. These patients were admitted in our ICU from 2007 to 2014. We used χ2 or t test for univariate analysis and Cox proportional hazards model for multivariate analysis to identify factors associated with mortality.

Results Pre-transplant median model for end-stage liver disease (MELD) score was 42 and mean SOFA (Sequential Organ Failure Assessment) score 15/24, reflecting, respectively, liver dysfunction and overall severity. In addition, 20 % out of the patients demonstrated septicemia before transplantation (i.e., during ICU stay). Overall post-transplant mortality rate was 33 % with a mean follow-up of 2.3 years. After 1 year, de novo acute liver failure patients had better survival rates than those from the French National Survey average, while the mortality rate of acute on chronic liver failure was 40 %. In addition, 97 % of survivors were autonomous, yet 2 % required chronic dialysis.

Ten pre-transplant factors were significantly predictive of 90-day post-transplant mortality in univariate analysis, among which arterial lactate, low platelet count, SOFA score, SAPS (Simplified Acute Physiology) II score and respiratory failure (PaO2/FiO2 ≤300 mmHg) for the overall cohort (p < 0.05). Severe pre-transplant respiratory failure (PaO2/FiO2 ≤200 mmHg) and a late initiation of invasive ventilation (>24 h after admission to ICU) were strongly predictive of 90 days post-transplant mortality in multivariate analysis (hazard ratio 7.1, p < 0.04, and 7.2, p < 0.05, respectively). In contrast, the need for norepinephrine infusion, renal replacement therapy or the occurrence of septicemia did not influence post-transplant mortality.

Discussion When excluding the seven deaths influenced by pre-transplant acute respiratory failure, the overall remote mortality is decaying to 25 %, which is close to post-transplant survival rates observed nationwide in liver transplantation for stable chronic liver disease. Moreover, in our cohort, long-term morbidity for survivors was low if compared to data from the literature.

Conclusion Our data suggest that liver transplantation can be safe and successful despite critical care criteria and poor underlying conditions (i.e., a context of circulatory failure requiring infusion of high doses of norepinephrine, septicemia or renal replacement therapy). They also suggest that liver transplantation should cautiously be advised in the presence of preoperative acute lung injury criteria. In the latter case, liver transplantation should (at least) temporarily be discouraged.

Competing interests None.

O83 Hospital survival and access to liver transplantation of MARS-treated patients in France, 2004–2008: a retrospective multicenter study (RETROMARS)

Correspondence: Christophe Camus -

Annals of Intensive Care 2016, 6(Suppl 1):O83

Introduction There is no current recommendation for the use of albumin dialysis with the molecular adsorbent recirculating system (MARS) in patients with acute or chronic liver failure. We performed a retrospective survey in France in patients who were treated with MARS, 2004–2008. The aim of the study was to assess the indications of MARS treatments, hospital survival, with or without liver transplantation (LTx), and also long-term follow-up, especially in patients with preexisting liver disease.

Patients and methods All the hospitals in which MARS technique was regularly performed were requested to participate to the study. All patients who had received MARS therapy in adult care setting over the 5-year study period were eligible. A questionnaire was sent to investigators and should be completed for each treated patient. The study was approved by the National Research Consultative Committee and the Institutional Ethics Committee. Variables associated with outcome were identified by multivariate analysis using logistic regression.

Results Among 25 centers, 16 participated to the study. A total of 383 patients received 393 MARS treatments (1091 sessions). Based on the examination of Mars treatment kits consumption records at all French hospitals during the study period, it could be estimated that the study accounted for 78 % of total MARS sessions in adult care setting in France.

The recorded indications for MARS (% of treatments) The indications were: major hyperbilirubinemia (37.1 %); hepatic encephalopathy (HE, 23.6 %); any kidney dysfunction (22.9 %; either hepatorenal syndrome [HRS, 15.3 %] or other renal dysfunction [7.6 %]); refractory pruritus (9.9 %); acute liver failure (ALF, 32.6 %); early nonfunction after LTx (6.1 %); liver failure after liver resection (3.8 %); ≥2 indications (37.1 %).

Hospital survival rate according to the indication (%, 95 % CI) One hundred and thirty-one patients were listed and 86 underwent LTx. Survival was assessed according to whether or not the patients were on the transplant list. (1) Not listed (66.3 %): ALF, 46 % (35–57); hyperbilirubinemia, 22 % (15–31); HE, 14 % (7–24); HRS, 20 % (10–34); refractory pruritus, 85 % (69–95); nonfunction after LTx, 0 % (0–21); liver failure after liver resection, 29 % (10–53); ≥2 indications, 17 % (15–25).

(2) Listed (33.7 %): ALF, 75 % (63–85); hyperbilirubinemia, 63 % (48–76); HE, 41 % (25–59); HRS, 44 % (23–67); refractory pruritus, 100 % (79–100); nonfunction after LTx, 82 % (55–96); ≥2 indications, 59 % (45–73).

Hospital survival rate in patients with nonbiliary cirrhosis (n = 137) The overall rate was 38 % (30–46) (listed for LTx: 56 % [40–71]; not listed: 32 % [23–41]). Variables associated with hospital death were HE (OR 3.00 [1.25–7.19], p = 0.01), hyperbilirubinemia (OR 2.40 [1.08–5.38], p = 0.03) and the absence of listing for LTx (OR 3.18 [1.30–7.76, p = 0.01).

Outcome in ALF patients Thirty-eight ALF patients (29.5 %) underwent LTx during their hospital stay. Listing with the “Superurgence” priority was the single variable associated with hospital survival (OR 3.79 [1.69–8.48], p = 0.001). Hospital survival without transplantation was associated with paracetamol etiology (OR 2.41 [1.04–5.58], p = 0.04) and the number of sessions performed (OR 1.49 [1.11–2.00], p = 0.009).

One-year KaplanMeier survival estimate according to preexisting liver disease (%, 95 % CI) The patients with preexisting liver disease were classified into five main diagnostic categories and were evaluated separately whether or not they underwent LTx. Fifteen patients were lost to follow-up.

(1) Without LTx: nonfunction after LTx, 9 % (−8 to 26); liver failure after liver resection, 0 %; late graft dysfunction, 5 % (−5 to 15); nonbiliary cirrhosis, 15 % (8–22); biliary cirrhosis/sclerosing cholangitis, 67 % (29–greater than 100).

(2) With LTx: nonfunction after LTx, 88 % (65–greater than 100); late graft dysfunction, 80 % (60–100); nonbiliary cirrhosis, 91 % (80–greater than 100); biliary cirrhosis/sclerosing cholangitis, 90 % (71–greater than 100).

Discussion ALF was the recorded indication of one-third of all MARS treatments and was associated with acceptable outcome even for nontransplanted patients. In patients with preexisting liver disease who were not candidates to LTx, the prognosis of nonbiliary cirrhosis was very poor, especially when HE or hyperbilirubinemia was the indication for MARS. Nonfunction after LTx was fatal in all cases without reLTx. Regarding late graft dysfunction and liver failure after liver resection, MARS provided no survival benefit in the absence of LTx perspective.

Conclusion MARS therapy could be used in patients with ALF even in those for whom LTx is not considered. In patients with preexisting liver disease, MARS should be best restricted to those who are suitable for LTx. Prospective studies are required to reassess the benefit of MARS in listed patients awaiting a liver graft.

Competing interests None.


1. Bañares R, Nevens F, Larsen FS, Jalan R, Albillos A, Dollinger M, Saliba F, Sauerbruch T, Klammt S, Ockenga J, Pares A, Wendon J, Brünnler T, Kramer L, Mathurin P, de la Mata M, Gasbarrini A, Müllhaupt B, Wilmer A, Laleman W, Eefsen M, Sen S, Zipprich A, Tenorio T, Pavesi M, Schmidt HH, Mitzner S, Williams R, Arroyo V; RELIEF study group. Extracorporeal albumin dialysis with the molecular adsorbent recirculating system in acute-on-chronic liver failure: the RELIEF trial. Hepatology 2013;57:1153–62.

2. Saliba F, Camus C, Durand F, Mathurin P, Letierce A, Delafosse B, Barange K, Perrigault PF, Belnard M, Ichaï P, Samuel D. Albumin dialysis with a noncell artificial liver support device in patients with acute liver failure: a randomized, controlled trial. Ann Intern Med 2013;159:522–31.Footnote 1

P1 Severe brain injuries: epidemiology and outcome in a Tunisian medical intensive care unit

Correspondence: Hassen Ben Ghezala -

Annals of Intensive Care 2016, 6(Suppl 1):P1

Introduction Severe brain injuries are most often due to road traffic accidents and usually occur in young persons. Despite the progress made, their management remains dependent on the availability of resources. In this study, we tried to establish the profile of severe traumatic brain injury (TBI) in the teaching department of intensive care unit of the regional hospital of Zaghouan and compare it to the literature. Then, we tried to determine outcome factors.

Materials and methods We performed a retrospective study over 04 years from 2011 to 2014. We enrolled all adult patients admitted to intensive care via emergency following a cranial trauma with GCS ≤8 after normalization of hemodynamic status. We excluded from the results analysis patient records with insufficient clinical or para-clinical data.

Results During the study period, 132 patients admitted for TBI were collected representing 3 % of all admissions and 12 % of all head trauma hospitalized in the unit. The most common mechanism was the road traffic accident (78 % of cases). All transport was non-medicalized. So there was no intubation–ventilation at the site of trauma. The takeover delay was 78 min. The average age of patients was 43 ± 16 years with extremes ranging from 19 to 83 years. The sex ratio was 4.3. The state of consciousness was evaluated by the GCS with an average of 5 ± 2. Sixty-six percent of patients had a pupillary anomaly, 19 % had a neurological deficit, and 81 % had associated lesions. The most frequent CT lesions were subarachnoid hemorrhage and cerebral contusions. All patients required intubation and mechanical ventilation with an average duration of hospitalization of 9 ± 2 days. The prognosis, assessed by the Glasgow Outcome Scale (GOS), was correlated with several factors. The takeover delay (p < 0.02), the GCS (p < 0.01), the presence of pupillary anomaly (p < 0.01), the presence of associated lesions (p = 0.01), the ISS score (p < 0.01) and the CT scan score (p < 0.01) were considered as outcome indicators in univariate analysis. A Kaplan–Meier survival curve showed a correlation between the age, the takeover delay and the outcome of the patients. Correlation with secondary systemic insults was not established in our study. Multivariate analysis showed a correlation between the Glasgow coma scale and mortality.

Conclusion The profile of TBI in ICU of Zaghouan is similar to data in recent literature. The prognostic factors identified in our study were: the mean delay of management, the GCS, the presence of pupillary anomalies, the presence of associated injuries, the ISS and the CT score. In order to improve the outcome of TBI, we need to use multimodal monitoring.

Competing interests None.

P2 Stroke prognosis in the ICU of the regional hospital of Bizerte and withhold and withdrawal treatment decisions

Correspondence: Hatem Ghadhoune -

Annals of Intensive Care 2016, 6(Suppl 1):P2

Introduction The short-term prognosis of stroke in ICU remains very reserved. Therefore, the management of this disease remains controversial. Few Tunisian studies have examined this subject, and no withhold and withdrawal treatment decisions were established in the severe cases. The purpose of this study is to explore predictors of stroke mortality in intensive care.

Patients and methods Sixty patients with stroke confirmed by CT scan (either intracerebral hematoma or ischemic stroke) were included in this study from patients admitted into the ICU (six beds) of the regional hospital of Bizerte between January 2010 and August 2015. The socio-demographic, anamnestic, clinical, biological and evaluative data were evaluated by univariate and multivariate analysis.

Results The median age was 60 years (interquartile range 54–66). Twenty (33.3 %) of the patients were women with a sex ratio equal to 2. The diagnosis of recurrent stroke was retained in 20 (33.3 %) patients. Sixty percent of patients had an ischemic stroke. IGS2 and APACHE 2 mean were, respectively, 41.57 ± 15 and 19.5 ± 6.7. One hundred percent of patients with GCS <6 died. The use of mechanical ventilation was required in 76.7 % of patients (38/60). The length of stay was 13.4 ± of 18 days. Mortality was 71.7 %.

In univariate analysis, factors associated with death were a high IGS2 (p = 0.009) and APACHE 2 (p = 0.01), a low coma scale (p < 10−3), a low PaO2/FiO2 ratio (p = 0.003), mechanical ventilation (p = 0.002) and septic shock appeared during evolution (p = 0.04). In multivariate analysis, a low coma scale and the onset of septic shock during the evolution were the only factors associated independently with increased ICU mortality among patients admitted for stroke.

Discussion Withhold and withdrawal treatment decisions in severe stroke are related mainly to neurological status and advanced age (1). The absence of reversible complications and a deep coma (GCS <9) shall qualify the usefulness of the resuscitation (2). Also, the use of mechanical ventilation is associated with high mortality (3). In this study, advanced age and mechanical ventilation were not associated with a significant mortality. For against a Glasgow score of <6 is associated with 100 % mortality. This could be a major criterion in therapeutic decisions limiting serious stroke and help Tunisian health authorities to improve the standards established in the management of severe stroke.

Conclusion The severity of the stroke at admission is related to the depth of the coma which depends on withhold and withdrawal treatment decisions. However, prognosis mainly relies on the improvement in the quality of stroke management and the prevention of its complications.

Competing interests None.


1. Crozier S, Santoli F, Outin H, Aegerter P, Ducrocq X, Bollaert PÉ. Severe stroke: Prognosis, intensive care admission and withhold and withdrawal treatment decisions. Rev Neurol (Paris). 2011 Jun-Jul;167(6–7):468–73.

2. Management of stroke in emergencies. Standards of Tunisian minister of public health. Mai 2006.

3. Crozier S. Withdrawing and withholding treatments in acute severe old stroke patients. La revue du Praticien. vol. 62, n° 9, 2012, pages 1243–1245.

P3 Brain death predictive factors for traumatic brain injury

Correspondence: Ahmed Youssef Kada -

Annals of Intensive Care 2016, 6(Suppl 1):P3

Introduction Encephalic death is one of the traumatic brain injury consequences. The goal of our study case-witnesses was to determine the predictive factors for them.

Patients and methods All patients hospitalized for traumatic brain injury are prospectively included between September 2008 and September 2013, and the only criterion of exclusion was the presence of a confusing factor. If they evolved to encephalic death, we paired them with deceased ones, but not of neurological cause.

Age, sex, Glasgow score, tomodensitometric classification (Marshall classification), medical transport and kidney injury (uremia) were collected.

We carried out an univariate then multivariate analysis to identify predictive factors for encephalic death post-traumatic.

Results A total of 95 cases and 95 witnesses were included in our study. All profited from a therapeutic without limitation of care. The average score at admission was 5.40 ± 2.29 at the cases and 6.96 ± 3.45 at the witnesses (p = 0.01), middle age being of 25.88 ± 15.08 years at the cases and of 38.71 ± 21.53 years at the witnesses (p = 0.01).

Male sex (p = 0.004; OR 3), transport >60 mn (p = 0.02; OR 2), age < 50 years (p = 0.0001; OR 6), a class IV on the scannographic classification of Marshall (Traumatic Coma Data Bank) (p = 0.01; OR 5.47) and high uremia (p = 0.01; OR 0.01) were found as predictors of encephalic death in univariate analysis.

In multivariate analysis, predictors were age lower than 50 years (p = 0.000; OR 8.9) and a unilateral mydriasis (p = 0.025; OR 4.026).

Discussion The aim of the study was to provide predictive clinical elements of passage to encephalic death. Two risk factors are come out from this analysis significantly, age factor (OR 8.9) and unilateral mydriasis (OR 4). These elements will make it possible to anticipate an unfavourable evolution and to adapt the therapeutic protocol consequently.

Conclusion These elements will make it possible to anticipate an unfavourable evolution and to adapt the therapeutic protocol consequently.

Competing interests None.


1. Facteurs pronostics de passage en état de mort cérébrale chez les patients en coma grave admis en réanimation dans le cadre d’une réanimation d’attente pour prélèvement d’organes. J.-B. lascarrou, C. ageneau, T. Fradin, S. Valot, M. Fiancette, I. Vinatier, M. lemarié, J.-C. lacherade, C. lebert, a. Cottereau, a. Yehia, e. Clémenti, J. Reignier, l. Martin lefevre. Réanimation (2012) 21:S21–S24.

2. Facteurs de risque d’évolution vers la mort encéphalique chez les patients hospitalisés en réanimation pour hématome intracérébral.V. Jouffroy, E. Brocas, A. Rodrigues, B. Deschamps, D. Perrin-Gachadoat. Prélèvement d’organes/Annales Françaises d’Anesthésie et de Réanimation 32S (2013) A250–A254.

P4 Guillain–Barré syndrome in intensive care unit: epidemiology and prognostic factors: analysis of 30 patients

Correspondence: Chtara Kamilia -

Annals of Intensive Care 2016, 6(Suppl 1):P4

Introduction Several complications related to Guillain–Barré syndrome (GBS) can lead to admission in ICU. The objective of our study is to describe clinical and laboratory characteristics of patients with GBS in ICU and identify predictive factors for poor prognosis in this patients group.

Patients and methods Retrospective study included all patients over a period of 8 years with GBS confirmed by an electromyogram. Categorical variables were expressed as frequencies and percentage, whereas quantitative variables were expressed as media [quartiles]. Two groups were compared in univariate analysis: survivors and deceased. Multivariate analysis was performed to identify the independent predictors of poor prognosis.

Results During the study period, 30 patients were admitted for GBS. The median age was 42.5 [26.3–59]. The sex ratio (M/W) was 0.6. The main indication to ICU admission was insufficient acute respiratory (86.7 %), shock (10 %) and impaired consciousness (13.3 %). The median SAPSII was 33.5 [29.5–39] points. The median SOFA score was 3 [2–5] points. The median SAPSIII score was 52 [50.3–56.5] points. The motor deficit was present in all patients. An associated sensory loss was observed in 76.7 % of cases. An anomaly of the ocular mobility was observed in 6.7 % of cases. Autonomic disorders were found in only one patient (3.3 %). A albumin-cytological dissociation was observed in 76.7 % of cases. The electromyographic study identified axonal damage in 36.7 % of cases. The mechanical ventilation was reported in 18 patients (60 %) for a median duration of 19 [8–31] days. Noninvasive ventilation has been attempted unsuccessfully in two patients. A tracheotomy was performed in nine patients (50 % of ventilated patients). Plasmapheresis was performed in 21 patients (70 %), while the immunoglobulins were administered in six patients (20 %). The ICU stay was complicated by nosocomial pneumonia in 13 patients (43.3 %) and pulmonary embolism in two patients (6.7 %). The duration median ICU stay was 11.5 [4.8–21.8] days. ICU mortality was 36.7 %. In multivariate analysis, only the occurrence of nosocomial pneumonia was associated with a poor prognosis (OR 44.6; 95 % CI [3.5–560]; p = 0.003).

Conclusion GBS is a serious disease involving life threatening. Mortality of severe forms seems to be conditioned by serious complications occurring during the ICU stay, especially pneumonia nosocomial.

Competing interests None.

P5 Eclampsia: epidemiological and clinical study about 194 cases

Correspondence: Fatma Kaaniche Medhioub -

Annals of Intensive Care 2016, 6(Suppl 1):P5

Introduction Eclampsia is a serious paroxysmal gravidic–puerperal accident complicates vascular and renal conditions of pregnancy. It is characterized by convulsive seizures that can be fatal to both mother and fetus [1]. The aim of our work is to study the epidemiological and clinical characteristics in obstetric patients admitted in ICU for eclampsia

Patients and methods Our study is retrospective and includes all cases of eclampsia collected in the medical ICU of the University Hospital of Sfax over a period of 20 years (1993–2012)

Results During our study period, 725 patients were hospitalized in the ICU. Among these patients, 194 presented eclampsia (26.7 %). Pregnancy has been followed medically in 79 women (40.6 %). The average age of patients was 28 ± 4.3 years. The average parity was 1.7 ± 1.2 with a primiparity from 49.5 %. The mean gestational age was 34.8 ± 2.9 weeks. HELLP syndrome was noted in 30 cases (15.5 %), placental retro hematoma in 15 cases (7.87 %), acute renal failure in 43 cases (22.3 %), disseminated intravascular coagulation in 27 cases (14 %), acute pulmonary edema in 17 cases (8.8 %) and state of shock in seven cases (3.6 %). The cesarean delivery was urgent in 161 cases (83 %). Mechanical ventilation more than 24 h was necessary in 79 cases (40.7 %), and 43 patients (22.2 %) received a transfusion (packed red blood cells and/or fresh frozen plasma and/or platelet button). Hemodialysis was carried in two cases (1 %). Antihypertensive therapy was initiated in 171 patients, magnesium sulfate in 190 patients and anticonvulsant therapy in 187 patients. The average length of ICU stay was 2.7 ± 2.2 days. Mortality hit nine patients (4.6 %) and 25 fetuses (12.8 %).

Conclusion Eclampsia is a serious accident complicating gestational hypertension. It particularly affects the primipara. Obstetric medical management including emergency cesarean section and adequate reanimation reduces maternal and fetal mortality.

Competing interests None.


1. Pottecher T, Luton D, Zupan V, Collet M. Réanimation des formes graves de prééclampsie. Conférence d’expert communes SFAR/CNGOF/SFMP/SFNN; 2009.

P6 Neuroimagery in eclampsia and preclampsia

Correspondence: Chaigar Mohammed Cheikh -

Annals of Intensive Care 2016, 6(Suppl 1):P6

Introduction Eclampsia is a severe complication of preeclampsia liable to an important maternal and fetal mortality, principally in developing countries [1]. The aim of our study was to describe neuroradiological lesions in preeclampsia/eclampsia in obstetric ICU of our hospital and to study clinical and therapeutic characteristics of this pathology as well as the outcome of our patients.

Patients and methods We conducted a retrospective descriptive study spread over 5 years, from January 1, 2010, to December 31, 2014, including all patients with posterior reversible encephalopathy syndrome (PRES), hemorrhagic stroke and cerebral ischemia occurring eclampsia or preeclampsia and hospitalized in obstetric ICU department.

Results Twenty-seven patients were included in the study. The mean age was 28.18 years. Only five patients had a well-attended pregnancy. The average systolic blood pressure was 158 mmHg, and the mean diastolic blood pressure was 96.29 mmHg. The mean Glasgow score at ICU admission was 11.18, and the observed neurological deficits were: hemiparesis (22 %), blindness (19 %), hemiplegia (7 %) and facial paralysis (7 %). Complications were: HELLP syndrome (29.6 %), placental hematoma (25.9 %), acute renal failure (3.7 %) and pulmonary edema (3.7 %). All patients had conventional neuroradiological exploration showing damage in all cases. CT showed two cases of ischemia (7.4 %), seven cases of cerebral edema (six cases of PRES: 22.2 %; one case of diffuse edema: 3.7 %) and three cases of hemorrhagic stroke (11.11 %). MRI diffusion sequence approved in 15 patients has showed a PRES in ten cases (37.03 %), cerebral ischemia in two cases (7.4 %), one case of hemorrhagic stroke and two cases of diffuse edema. The main mode of delivery was cesarean section under general anesthesia for 22 patients (81.48 %). All patients had received antihypertensive treatment and anticonvulsant treatment. The outcome was critical in one patient how died from a septic shock. Perinatal mortality was about 25.9 %.

Conclusion From this study, improving maternal and fetal prognosis requires:

  • Prevention, with early and regular monitoring of pregnant women;

  • An appropriate treatment of pregnancy hypertension;

  • And management of preeclampsia and eclampsia before complications.

Competing interests None.


1. o.collange et Al. Eclampsie. AFAR2010.

P7 Interest of brain oxygen tension measurement (PbrO2) in the diagnosis of cerebral vasospasm after aneurismal subarachnoid hemorrhage

Correspondence: Stephane Welschbillig -

Annals of Intensive Care 2016, 6(Suppl 1):P7

Introduction Delayed brain ischemia as a result of cerebral vasospasm is the most common cause of death and disability after aneurysmal subarachnoid hemorrhage (SAH). Because clinical signs cannot be detected for sedated patients, monitoring tools have been developed. As an example, transcranial Doppler is an easily bedside method to detect middle cerebral artery vasospasm [1], but fails in the detection of others intracranial vessels spasm. Tissue brain oxygen tension measurement (PbrO2) has been described as a useful monitoring tool to detect tissue hypoxia before irreversible neuronal damage in trauma brain-injured patients [2]. The goal of the study was to determinate the accuracy of PbrO2 to detect cerebral vasospasm after severe aneurysmal SAH.

Patients and methods We performed a retrospective observational study on ten patients admitted in our institution from January 2014 to September 2015 for SAH and monitored with PbrO2 (Lycox). PbrO2 data were recorded retrospectively. Cerebral vasospasm was assessed with angiography and treated with in situ vasodilators. PbrO2 before and after cerebral spasm angioplasty was defined “with” and “without” cerebral spasm values of PbrO2. Continuous variable PbrO2 was presented mean (SE) and compared with paired t test.

Results Ten comatose SAH patients (WNFS IV) were included, and 48 angiographies were performed for cerebral vasospasm. PbrO2 probe was inserted in all patients 7 (3–14) days after SAH without any complication. The mean duration of PbrO2 monitoring was 8 (4–16) days. Two vasospasm episodes were excluded because PbrO2 probe was located in the opposite side of the vasospasm area. In four cases of vasospasm, the angioplasty failed and it was impossible to obtain the “without spasm” value of PbrO2.

In cerebral vasospasm, mean value of PbrO2 was 19(9) mmHg but with a large spread of values. After angiographic treatment of vasospasm, mean value of PbrO2 was significantly higher and reached 27(9) mmHg (p < 0.0001) (Fig. 20). Despite a wide overlap between the two groups of PbrO2 values, we could observe that a PbrO2 value lower than a threshold value of 22 mmHg predicted vasospasm with both sensibility and specificity of 75 %, and when PbrO2 value was under 11 mmHg, all patients had a cerebral vasospasm.

Fig. 20
figure 20

See text for description

Conclusion Continuous monitoring of PbrO2 could have an interest to predict the cerebral vasospasm and to avoid delayed brain ischemia in comatose patients after aneurismal subarachnoid hemorrhage.

Competing interests None.


1. Symptomatic vasospasm diagnosis after subarachnoid hemorrhage: evaluation of transcranial Doppler ultrasound and cerebral angiography as related to compromised vascular distribution. Suarez JI, Qureshi AI, Yahia AB, Parekh PD, Tamargo RJ, Williams MA, Ulatowski JA, Hanley DF, Razumovsky AY. Crit Care Med 2002;30(6):138–55.

2. Brain oxygen monitoring: in vitro accuracy, long-term drift and response-time of Licox and Neurotrend sensors. Hoelper BM, Alessandri B, Heimann A, Behr R, Kempski O. Acta Neurochir (Wien). 2005 Jul;147(7):767–74.

P8 Noninvasive ventilation: an alternative to the intubation during myasthenic crisis

Correspondence: Mathilde Perrin -

Annals of Intensive Care 2016, 6(Suppl 1):P8

Introduction Myasthenic crisis is defined by acquired myasthenic gravis severe enough to require intubation. The actual recommendations are to intubate patients who have a vital capacity <15 mL/kg, or 30 % (1) while waiting for treatment by intravenous immunoglobulins to be effective.

The use of noninvasive ventilation (NIV) to prevent intubation was described in only two retrospective studies (2, 3). NIV was successful for 20–25 % of patients showing myasthenic crisis. We wanted to describe our experience of NIV in myasthenic crisis.

Patients and methods This is an observational prospective study from 01.11.2013 to 01.03.2015. Patients included were assisted with NIV when they showed a single myasthenic crisis, an Osserman score <70 and a vital capacity (VC) <50 %. Durations of NIV were adapted to the patient’s comfort.

The NIV failure was: inhalation, intubation, hypoxemia, major asthenia.

Every day, the demographics and treatment modalities, pulmonary complications, intubation outcomes, duration of NIV and duration of stay in intensive care were recorded.

Results Eight patients were included. The median age was 53 (16–81) years. At the admission in intensive care unit, the median Osserman’s score was 50 (23–68) and VC 38 % (22–43 %). Median duration of NIV was 4 days (2–10), and median ICU length of stay was 7 days (4–13). Nobody was intubated. Five patients with deglutition troubles did not show pulmonary complications.

Conclusion In this group, NIV was always successful. Systematic use of NIV seems to prevent muscular asthenia caused by gravis myasthenic crisis and avoids patient’s respiratory worsening. NIV in intensive care unit seems to be a good way to not intubate patients with gravis myasthenic crisis (Fig. 21).

Fig. 21
figure 21

See text for description

Competing interests None.


1. Reanimation 14 (2005) 118–125.

2. Neurocrit care 2009; 10(1):35–42.

3. Arch Neurol 2008 Jan; 65(1):54–8.

P9 Interest of the transcranial Doppler ultrasound in the management of severe traumatic brain injury

Correspondence: Soumia Benbernou -

Annals of Intensive Care 2016, 6(Suppl 1):P9

Introduction After a serious traumatic brain injury (TBI), the occurrence of intracranial hypertension (intracranial hypertension) is associated with an increased risk of morbidity and mortality. The appearance of intracranial hypertension at an early stage worsens the prognosis, and thus, through the clinic, brain CT, monitoring intracranial pressure and evaluation of cerebral circulation by transcranial Doppler ultrasound treatment can be given quickly. The objective of the study was to compare the advantage brought by one or other of the means used for the diagnosis and monitoring of intracranial hypertension.

Patients and methods The work was exhibited/performed at the ICU of the UMC CHUOran.

This is a prospective study over 2 years (2012 and 2013) of 59 patients who received simultaneously a monitoring intracranial pressure (ICP) and a transcranial Doppler monitoring by the TCDU. Simultaneous recording of intracranial pressure, cerebral perfusion pressure, the pulsatility index, systolic and diastolic velocity has been achieved.

Results The number of STBI compiled was 59. The study population was young, the median age was 29 years, and 80 % were trauma patients. No pupillary changes are observed for higher numbers of ICP >45 mmHg. There is a linear relationship between the ICP and the pulsatility index since the correlation coefficient was 0.43 and p = 0.019965 and between the CFC and the pulsatility index since the correlation coefficient: r2 = 0.60 and p = 0.000613. The literature is contradictory on this issue.

Discussion A study was conducted in 1994 by Boishardy and collaborators in France, out of a total of ten serious head trauma whose age was between 17 and 37 years and that benefited dice admission and then twice a day, of a simultaneous measurement SV, DV, PI and ICP by means of an intraparenchymal catheter to show a significant correlation between mean ICP and PI r = 0.837 and between the PI and CPP (r = 0.563; p < 0.001) [1].

By against the US study conducted by Thomas in California on a group of 99 severe TBI patients showed that the correlation between ICP and PI was classified as moderate to zero [2].

Conclusion All severe TBI monitoring means have their place, with superior non-invasive tests on simple clinical surveillance, and indicate the need for the installation of the monitoring of the ICP in due time, since many patients can develop at any time during their stay an ICHT.

Competing interests None


1. Boishardy, J.C. Granry, J.P. Jacob, N. Houi, D. Fournier, A. Delhumeau. Intérêt du doppler transcrânien dans la prise en charge du traumatisés crâniens graves. Annales Françaises d’ Anesthésie et de Réanimation, 13: 172–176, 1994.

2. Thomas C. Glenn, Arun K. Sherma, David L. McArthur, Xiao Hu, Christopher R. Hanuscin, Mehjabeen S. Furreedan, David A. Hovda, Paul M. Vespa, and Neil A. Martin The Linear Relationship Between Transcranial Doppler Pulsatility Indices and Intracranial Pressure Is Influenced by Traumatic Brain Injury and Vasospasm University of California Neurotrauma Initiative and NIH/NINDS 1P01NS058489-01.

P10 VV-ECMO on ARDS: toward a lower anticoagulation ratio?

Correspondence: Elie Zogheib -

Annals of Intensive Care 2016, 6(Suppl 1):P10

Introduction Veno-venous ECMO (VV-ECMO) is a rescue therapy in refractory ARDS. Despite technological progress and enhanced biocompatibility, anticoagulation with unfractionated heparin remains the most often used as recommended. Bleeding complications remain a main cause of increasing morbidity and mortality in patient treated with VV-ECMO. The aim of our study was to study the effectiveness of the dosage of unfractionated heparin (aPTT, anti-Xa activity) in patients on VV-ECMO for ARDS

Patients and methods We performed a retrospective observational study stretching from 2008 to the first quarter 2015. Patient’s baseline characteristics, the dose of UFH anticoagulation and biological data of anticoagulation (aPTT, anti-Xa activity, Hemochron ACT, ATIII, fibrinogen) were collected. We studied the incidence of hemorrhagic or thromboembolic events under UFH anticoagulation.

Results Forty-eight patients with ARDS requiring VV-ECMO were included, with a majority of men [35 (63.6 %) vs 20 (36.4 %) (p < 0.5)]. The mean age was 47 years (±15 years). In 35.4 % of cases, a double cannula Avalon common type used (mean diameter of 27 GB), while the others had a femoro-jugular cannulation with an average diameter of femoral vein of 25 GB. UFH anticoagulant was used in most patients. During installation of VV-ECMO, 33 % of patients did not have unfractionated heparin (UFH). The average value of aPTT was 35 ± 8 s and the anti-Xa was 0.14 ± 0.06. The average of the Hemochron was 171 ± 23 s. The values of UFH administered averaged 778 ± 601 IU/h IVSE. Bleeding complications were noted in 56.25 % of cases. None of our patients had ECMO thrombosis cannulas or oxygenator. Thirty-nine percent of patients had positive anti-PF4, and only 11 patients had a change in anticoagulation.

Conclusion Anticoagulation used in our study appears lower than recommendations without increasing thromboembolic complications, with persistence of a high rate of bleeding complications with or without a direct connection with ECMO. These results would reinforce the possibility of using ECMO with less or no anticoagulant?

Competing interests None.


1. Extracorporeal life support for patients with acute. Ann Intensive Care. 2014 May 24;4:15. doi: 10.1186/2110-5820-4-15 . eCollection 2014.

P11 Inter-hospital transportation of patients implanted with veno-venous ECMO: Experience of the Pitié-Salpêtrière mobile ECMO team

Correspondence: Nicolas Brechot -

Annals of Intensive Care 2016, 6(Suppl 1):P11

Introduction The way to organize veno-venous circulatory assistance in France is still a matter of debate. Particularly, data on the efficiency and safety of mobile ECMO teams are still scarce. The aim of this study was to evaluate the safety of the mobile ECMO team in our center.

Patients and methods In this retrospective monocenter study, the outcome of patients implanted with a veno-venous ECMO by the mobile ECMO team of la Pitié-Salpêtrière hospital from 2009 to 2012 was compared with patients who were implanted on site. Results are expressed as median (25th–75th IQR) or n (%).

Results A total of 118 patients were implanted by the mobile ECMO team and 39 on site. Both groups were quite comparable and consisted in very severe patients, reflected by a high SAPS-II score of 71 (61–81) and a PaO2/FiO2 of 58 (50–73). Major complications at ECMO implantations were identical between both groups [3 (2.5 %) in the mobile ECMO group vs. 2 (5.1 %) in the onsite group, p = 0.6]. All transportations were performed by the SAMU with ground facilities, in the presence of a specialized perfusionist. Median time between call and ECMO implantation was 192 min (154–225). Distances ranged from 0.5 to 241 km (median 15 (6–25)), for a median time of 35 min (25–35). No major incident was observed during transportation, except 1 (0.8 %) patient who died from hemodynamic deterioration. Eight (6.8 %) patients in the mobile group needed a subsequent switch to veno-arterial ECMO due to hemodynamic deterioration, compared with two (5.1 %) in the onsite group (p = 0.7). ICU mortality was comparable, 46.6 % in the mobile group versus 53.8 % in the onsite group, p = 0.5. In multivariable analysis, implantation by the mobile ECMO team was not associated with ICU mortality (OR 1.1 [0.4–1.7], p = 0.85). Severe ECMO-related complications were frequent but comparable between groups [63 (53.4 %) in the mobile group vs. 21 (53.8 %) in the onsite group, p = 1.0]. Particularly, the rate of subsequent recannulation, bleedings or infection at the insertion site and bacteremia did not differ between groups.

Conclusion Implantation of a veno-venous ECMO by a mobile ECMO team and subsequent transportation of the patient to an experienced ECMO center appears to be feasible and safe.

Competing interests None.

P12 Evolution of tidal volume and compliance under pressure-controlled mode during the first 24 hours of ECMO ARDS patients transported to a referral center

Correspondence: Hadrien Roze -

Annals of Intensive Care 2016, 6(Suppl 1):P12

Introduction Extracorporeal membrane oxygenation (ECMO) is proposed for patients with severe acute respiratory distress syndrome (ARDS), which may be defined as not being effectively and/or safely treated by current clinical types of mechanical ventilation. Once on ECMO, it is immediately possible to limit stress and strain with a volume- and pressure-limited protective ventilation strategy. This study focuses on the mechanical ventilation during the first 24 h of ARDS patients transferred with an ECMO mobile team (UMAC) to the referral center of Aquitaine region.

Patients and methods All data were prospectively recorded, 1 h before ECMO (H-1), at ECMO start before transportation (Humac), at 1 h (H1) and 24 h (H24) in the referral center. Patients were transported by road with SAMU 33 using a Cardiohelp (Maquet, Sweden) and a ventilator Elisé 350 (ResMed, Australia). During the first 24 h under ECMO, patients were paralyzed and ventilated with a pressure-controlled mode targeting a plateau pressure of 24 cmH2O, a respiratory frequency of 12 and a PEEP of 10 cmH2O. Results are expressed as median (interquartile range 25–75 %)

Results We analyzed 15 ARDS patients, between 2013 and 2014, requiring veno-venous ECMO. Median age was 35 (30–55) years; IGS2 was 50 (43–59). At H-1, PaO2/FiO2 was 49 (42–54), and thoracopulmonary compliance (Ctp) was 16 (13–25) ml/cmH2O. Between H-1 and Humac, protective ventilation decreased driving pressure and VT from 25 (16–27) to 12 (12–15) cmH2O and 6.4 (5.6–7.1) to 3.8 (2.5–5.8) ml kg−1 PBW, respectively, p < 0.0001, whereas PEEP did not change 8 (8–12) versus 12 (10–12) cmH2O, p = 0.12. At Humac, Ctp was unchanged with 15 (10–27) ml/cmH2O. At H24, VT and Ctp were significantly lower than Humac 2.6 (1.5–3.2) ml kg−1 PBW p = 0.01 and 11 (8–14) ml/cmH2O p = 0.03, respectively. At H1, eight patients (53 %) had a decrease of VT >15 % during transportation, median reduction was −44(19–73) %, and VT was 1.5 (1.3–3.7) ml kg−1 PBW. No patient responds to alveolar recruitment procedures at H24. Duration of ECMO was 14 (7–24) days, and three patients died in the referral center (20 %) (Fig. 22).

Fig. 22
figure 22

See text for description

Discussion ECMO allowed immediate reduction in driving pressure with a stable Ctp. We found a significant reduction in VT and Ctp during the first 24 h and for some patients during transportation. This might be explained by the worsening of ARDS, by alveolar derecruitment due to a pressure-controlled mode (1), by the use of a transport ventilator under extremely low Ctp and curare, by the disconnections of the patient from the ventilators at each steps of transportation or by the airway suctions.

Conclusion Limiting driving pressure of severe ARDS under ECMO is of high clinical interest. By nature, pressure-controlled ventilation permits to achieve this goal, but it may also promote derecruitment and affect compliance, especially during patient’s transportation.

Competing interests None.


1. Rittayamai et al.Chest 2015;148:340–355.

P13 Extracorporeal dioxide carbon removal (ECCO2R), a French national survey

Correspondence: Stéphane Gaudry -

Annals of Intensive Care 2016, 6(Suppl 1):P13

Introduction Extracorporeal dioxide carbon removal (ECCO2R) is an extracorporeal decarboxylation technology described in 1978. Physiological studies showed that 50 % of produced CO2 were eliminated. Potential indications are: ultra-protective mechanical ventilation for acute respiratory distress syndrome (ARDS) and hypercapnic patients at risk of non-invasive ventilation failure. Because of the lack of scientific evidence, ECCO2R is not available in the USA. We sought to assess the use of ECCO2R in France.

Materials and methods This retrospective, observational study was performed in French intensive care units (ICUs) from January 2010 to January 2015. A phone interview was conducted with French ICUs affiliated to national societies and public and private hospitals registries. Data recorded were the following: use and indications of ECCO2R, type of ECCO2R, number of patients treated during the study period, complications associated with the technique, satisfaction rates (in terms of efficacy, tolerance and global) based on a scale [0 (total unsatisfaction)–10 (total satisfaction)], and concomitant use of extracorporeal membrane oxygenation (ECMO) in the unit.

Results A total of 239 French ICUs were contacted (147 mixed, 55 medical, 25 surgical, four cardio-thoracic, six paediatric and two neurosurgical ICU). Only three refused to participate. Thirty-five (15 %) ICUs had used ECCO2R at least once in the past 5 years, in 303 patients. The most frequently used devices were: iLA® (Novalung) (63 %) and Hemolung® (Alung) (37 %). The median number of patients treated per ICU was 3 [1–8]. The most frequent indication was ultra-protective ventilation for ADRS (54 %). Other indications were: failure of non-invasive ventilation during chronic obstructive pulmonary disease exacerbation (30 %), weaning from invasive mechanical ventilation in chronic obstructive pulmonary disease patients (12 %) and miscellaneous (4 %).

Among ICUs using ECCO2R, 22 (63 %) reported at least one complication. The most frequent complications were bleeding (45 %) and membrane failure (18 %). Satisfaction rates were: in terms of decarboxylation 7.9 ± 2.4; tolerance 6.9 ± 2.6; and overall satisfaction 6.8 ± 2.2. Twenty-one (63 %) of the 35 ICUs using ECCO2R also used ECMO. The main reasons for not using ECCO2R were the lack of trained staff, unavailability of the device and the lack of scientific evidence (for, respectively, 53, 39 and 17 % of responders). Among 204 ICUs not using ECCO2R, 20 (10 %) had the project to start ECCO2R in the coming months.

Conclusion These results show that ECCO2R is not widely used in French ICUs. The lack of strong scientific data on outcome is probably the main reason behind the limited use of ECCO2R. French studies currently in progress will help define indications of ECCO2R and impact on outcome.

Competing interests None.

P14 Levels of microparticles in acute respiratory distress syndrome with veno-venous extra-corporeal membrane oxygenation

Correspondence: Fanny Klasen -

Annals of Intensive Care 2016, 6(Suppl 1):P14

Introduction Veno-venous extra-corporeal membrane oxygenation (vvECMO) allows blood oxygenation and CO2 removal for the most severe case of acute respiratory distress syndrome (ARDS). One of the major complications associated with this technique is the occurrence of microthrombi on the ECMO membrane oxygenator which can require the change in circuit. The microparticles (MPs) are proinflammatory and procoagulant nanovesicles derived from circulating cells, and their circulating levels may be elevated in many pathologies associated with inflammatory, thrombotic and infectious disorders. The main objective of our study is to quantify the presence and to characterize the cellular origin of MPs in patients treated by vvECMO for severe ARDS. Secondary objectives are to analyze the kinetic of MPs rates between ECMO application and removal and to establish a link between the need for a circuit change and MPs evolution.

Materials and methods This is a monocentric pilot study, conducted between January 2012 and June 2014, which included patients with severe ARDS treated with vvECMO. We measured total MPs, platelet-, erythrocyte-, endothelial- and leukocyte-derived MPs, coagulation and hemolysis parameters within 6 h after the ECMO application. The same parameters were measured at the time of the circuit removal (either during weaning or during a circuit change for biological complication or because of death).

Results Nineteen patients were included in the study analysis. Leukocyte-derived (p < 0.0001) and erythrocyte-derived (p = 0.04) MPs were higher at the admission in patients who died under ECMO compared with those who have been weaned alive. Leukocyte-derived (p = 0.01) and endothelial-derived (p = 0.001) MPs were higher in patients who needed a circuit change.

Discussion There are subpopulations of MPs associated with the prognosis of patients with severe ARDS treated with veno-venous ECMO.

Conclusion This pilot study suggests a potential role of these biomarkers as diagnostic or prognostic factor in the occurrence of complications related to the technique.

Competing interests None.

P15 In patients under extracorporeal CO2 removal therapy (ECCO2R) for ARDS can we do prone position? Efficiency, stability and safety of the maneuver

Correspondence: Jean Francois Georger -

Annals of Intensive Care 2016, 6(Suppl 1):P15

Introduction The mechanical ventilation of some patients with ARDS could be facilitated by ECCO2R allowing the reduction in blood acidosis and the reduction in tidal volume for the application of the protective ventilation. Prone position (PP) could be used for some patients with PaO2/FiO2 <150. We do not know whether we could associate PP and ECCO2R in ARDS patients

The aim of this study is to describe the feasibility of PP under ECCO2R, the stability of the parameters of the device and whether we have side effects of the PP under ECCO2R.

Materials and methods In our intensive care unit of 15 beds with a large experience of PP, we have retrospectively included all sessions of PP (at least 16 h of PP) performed on patients under ECCO2R therapy between August 2014 and March 2015. We used ILA ACTIVVE® device (NOVALUNG®) with MINILUNG® membrane and a double-line femoral catheter (NOVAPORT TWIN® 24F). The gas flow was 10 l/min. For each session, we compared PaO2/FiO2 and the PaCO2 before and after 1H of PP. The device records the flow and pressure parameters every 10 s, and we recovered these parameters after each prone position session. For each session, we did the mean of blood flow and drainage pressure (P1) during a length of 1 h: during the last hour before PP, the first hour after PP and the last hour before stopping PP. We compared with a Freidman’s test, the mean and the coefficient of variation of each parameter to evaluate the stability of the device. We noted all the side effects of the PP (bleeding, decannulation, etc.).

Results We performed nine PP sessions on five patients, one in three patients and three in two patients. The PaO2/FiO2 ratio was higher during PP [136 (78–250)] than before PP [126 (58–145)] (p < 0.05). There is no difference of PaCO2 before PP [55 mmHg (34–80)] than during PP [54 (34–70)]. Between before, the beginning and the end of PP, we did not find difference in blood flow, respectively, 1472 ml/min (1201–1971), 1403 ml/min (1216–1850) and 1447 ml/min (1231–2012), and in P1, respectively, −37 mmHg (−46 to −25), −41 mmHg (−50 to −28) and −41 mmHg (−47 to −29). The coefficient of variation of the blood flow was low, and we did not find variations of it between these three moments, respectively 0.9 % (0.7–2.8), 0.7 % (0.4–2.1) and 0.6 % (0.4–1.6). The coefficient of variation of P1 was low, and it was lower at PP than before PP (p < 0.05), respectively: 8.2 % (3.7–9.9), 5.6 % (2.8–6.8) and 4.2 % (2.9–5.8). We did not find side effects of the PP maneuver. There were no decannulation, bleeding and skin lesion on the tubing paths.

Conclusion Prone position under ECCO2R with a femoral catheter is possible. We found no side effects of this technique. No differences in the blood flow, in the drainage pressure and in the stability of the blood flow were found. The stability of the drainage pressure is better in PP. The PaO2/FiO2 ratio is better in PP.

Competing interests None.

P16 Feasibility of the implementation of a technique of extra-corporeal CO2 removal (ECCO2R) in an intensive care unit which does not use ECMO and its real utilization

Correspondence: Jean Francois Georger -

Annals of Intensive Care 2016, 6(Suppl 1):P16

Introduction The utility and feasibility of ECCO2R in an intensive care unit (ICU) which does not use ECMO has not been described. In our general ICU of 15 beds, we do not use ECMO, and we decided to introduce ECCO2R in COPD and ARDS patients.

The purpose is to describe the feasibility of this technique in our ICU and the actual uses of it.

Materials and methods We chose ILA ACTIVVE® device (NOVALUNG®) with MINILUNG® membrane and a double-line femoral catheter (NOVAPORT TWIN 24F). A training for each member of the team was conducted (physicians and nurses), and we wrote a detailed procedure between February 2014 and July 2014. We wrote a procedure for: the preparation of the device, laying the catheter, monitoring, removal of the catheter and the machine, any malfunctions, etc.). Between august 2014 and February 2015, data were collected for each patient receiving ECCO2R. Indication, mortality at J28, duration of ECCO2R, complications and success of the technique were evaluated.

Results During this period, ten of the 348 patients admitted to our ICU have received ECCO2R therapy. We included five patients with ARDS (we laid the device because tidal volume (Vt) was 6 ml/kg PBW with a PaO2/FiO2 ratio between 80 and 150, after at least one prone position session, and with a plateau pressure between 25 and 30 cm of water and high level of PaCO2). We included three COPD patients in failure of non-invasive ventilation (NIV) in a hypercapnic coma, one asthmatic patient in severe respiratory acidosis under invasive ventilation (IV) and one severe COPD patient unweaning of the IV. Patients with ARDS normalized arterial pH, and we decreased Vt (3–4 ml/kg) and respiratory rate. Three of them are alive at J28 and had received 5, 7 and 14 days of ECCO2R, and two of them died under ECCO2R after 16 and 9 days of ECCO2R. The reason of the death was nosocomial lung infection. None of the COPD patients with hypercapnic coma were intubated, and they received 5, 3 and 6 days of ECCO2R and were alive at J28. The asthmatic patient received 7 days of ECCO2R, was extubated under ECCO2R and was alive at J28. The unweaning COPD patient was extubated two times and reintubated two times due to severe neuropathy and tracheal inhalation. He was tracheotomized and weaned from the ECCO2R after 19 days, and he was alive at J28, but died at J76. There was no bleeding leading by catheter but one hemothorax requiring a blood transfusion during an overdose of heparin

Conclusion The ECCO2R in a general ICU could have multiple indications in COPD and ARDS patients. These indications remain to be validated, but ECCO2R could improve the management of those patients. The implementation of ECCO2R is possible but requires a great investment, a training of the medical and the paramedical team and a written detailed procedure. Nevertheless, the installation of the device for each patient must for now be discussed in a team taking into account the risk–benefit ratio of each procedure.

Competing interests None.

P17 Six-hour daily session of high-frequency oscillatory ventilation in moderate-to-severe ARDS: impact on the end-expiratory lung volume, compliance and oxygenation

Correspondence: Bérénice Puech -

Annals of Intensive Care 2016, 6(Suppl 1):P17

Introduction High-frequency oscillatory ventilation (HFOV) is usually considered as an alternative continuous technique to conventional mechanical ventilation (CMV) with the aim of improving the oxygenation. Our goal was to use HFOV not as a surrogate to CMV but as a recruitment process during only 6 h a day alternating with the CMV and to test whether this new sequential HFOV use could increase the end-expiratory lung volume (EELV) and the respiratory system compliance (used as recruitment markers) and thus improve the oxygenation.

Patients and methods Interventional prospective study, in single-center ICU, was performed in patients with moderate-to-severe ARDS (PaO2/FiO2 ratio <150 after a 6-h period of optimized CMV for maximal alveolar recruitment). Patients were placed under HFO sequentially (6 h a day) up to 4 days in a row, as long as PaO2/FiO2 ratio was <200. The primary endpoint was the before/after HFO session EELV difference measured by the nitrogen wash-in/wash-out technique. Secondary endpoints were the before/after HFO session PaO2/FiO2 ratio and respiratory system compliance differences.

Results Twenty-two patients (mean IGS 55 ± 14) were enrolled between November 2012 and May 2015 for a total of 29 sessions. Three of these (10 %) were stopped (two for hemodynamic failure and one for respiratory failure). After the 6-h HFOV session, the EELV increased significantly by more than 30 %, while the PaO2/FiO2 ratio and the respiratory system compliance increased significantly by more than 90 % and more than 20 %, respectively. By considering as responder to the HFOV session a patient with an EELV increase of more than 15 % after the HFOV, we obtained 19 responders (66 %) for 10 non-responders. By considering as responder to the HFOV session a patient with an EELV increase of more than 15 % and a PaO2/FiO2 ratio increase of more than 30 % after the HFOV, we obtained 16 responders (55 %) for 13 non-responders. No significant change in hemodynamic parameters was found, except the significant 10 ± 6 % cardiac index decrease after 1 h of HFOV for the 15 sessions monitored by transpulmonary thermodilution. No pneumothorax occurred during the protocol.

Conclusion In ARDS patients with a PaO2/FiO2 ratio <150, an early 6-h HFOV session provides a significant improvement in oxygenation and ventilatory mechanics by alveolar recruitment. HFOV could be used sequentially in addition to CMV as a recruiting technique and not as an alternative rescue method.

Competing interests None.

P18 Can we optimize prescription of laboratory tests in surgical intensive care unit (ICU)? Study of appropriateness of care

Correspondence: Marion Fresco -

Annals of Intensive Care 2016, 6(Suppl 1):P18

Introduction In the lack of guidelines, prescription of laboratory tests in ICU sometimes seems to pertain to individual or team habits. We aimed to assess the impact of clinical guidelines to improve the volume of laboratory tests performed.

Materials and methods It was a monocentric, comparative study, before and after guidelines implementation at patient bedside, in a surgical ICU of a French university hospital. All consecutive patients present between May 1 and October 31, 2013, and between May 1, and October 31, 2014, were enrolled. The prescription help-guide was displayed in each patient room from May 2, 2014. Physicians were encouraged to rely on this help-guide for laboratory tests prescription, once patients stabilized. The primary outcome measure was the variation of routine laboratory tests per patient ICU day, before compared with after guidelines. Secondary outcomes were the number of ICU days, the mortality, the variation of blood transfusion, the variation of nosocomial infections potentially related to blood samplings and the economic impact after guidelines implementation. We also assessed the variation of laboratory tests per patient ICU day, during the same periods, in the medical ICU of the university hospital, without any guidelines.

Results A total of 274 patients were admitted, staying for 3167 ICU days during Period-1, and 342 patients were admitted, for 2799 ICU days during Period-2. The patients’ general characteristics were similar, except for the ICU days significantly shorter for Period-2. A 27.21 % relative reduction in routine laboratory tests per patient ICU day was observed after guidelines implementation. A 27.45 % relative reduction in blood transfusion was also observed, without difference neither on mortality in ICU, nor on nosocomial infections potentially related to blood samplings. The mean decrease per laboratory test was significantly more important in the surgical ICU than in the medical ICU (27.21 vs. 15.13 %, p = 0.00837). There were a 124,000 € overall cost reduction on 6 months related to the guidelines implementation, and a transfusion economy of 53,000 €.

Conclusion This help-guide induced an important decrease in the volume of routine laboratory tests performed. We also demonstrated a major reduction in blood transfusion, without difference either on mortality, or on nosocomial infections potentially related to blood samplings. We observed a very important economic impact of our clinical guidelines.

Competing interests None.


1. Prat G, Lefèvre M, Nowak E, Tonnelier JM, Renault A, L’Her E, & al. Impact of clinical guidelines to improve appropriateness of laboratory tests and chest radiographs. Intensive Care Med. 2009 Jun; 35(6):1047–1053.

2. Lehot JJ, Heuclin C, Neidecker J, Cartier R, French PF, Reverdy ME et al. How can we optimize medical orderings in intensive care unit? Ann Fr Anesth Reanim. 2010 Oct; 29(10):682–686.

P19 Impact of the implementation of guidelines for laboratory testing in an intensive care unit

Correspondence: Pierre-Edouard Bollaert -

Annals of Intensive Care 2016, 6(Suppl 1):P19

Introduction The aim of this study was to evaluate the long-term effects of the implementation of “good laboratory test practices” procedure on blood test prescription, cost cutting laboratory test and patient outcome and red cells transfusion needs. In addition, we looked for factors associated with transfusion.

Patients and methods This is a retrospective, single-center study.

Considering that prescription of laboratory tests was excessive in our ICU, we developed in 2009 a procedure aiming at limiting assays and blood sampling. All adult patients over 18 years admitted in the ICU from January 1, 2008, to December 31, 2012, were consecutively included. For each patient, the following data were retrospectively obtained from medical records and institutional databases: demographic and severity of illness parameters, the amount and costs of laboratory tests, daily blood volume drawn, number of red cells transfusions and ICU and in-hospital outcomes.

Results A total of 3568 patients were included between January 1, 2008, and December 31, 2012. Patient’s characteristics were similar during the study with respect to overall severity and vital organ support, except the need for vasoactive drugs which increased across time. After implementation of guidelines, the total number of laboratory tests decreased from 78,406 in 2008 (18.1 tests/patient/day) to 27,514 in 2012 (6.4 tests/patient/day). 318,000€ were saved in 2012 on the five more expensive tests as compared to 2008. Average daily blood drawn volume decreased from 29.2 ± 13 ml/day in 2008 to 22.4 ± 10.1 ml/day (p < 0.001) in 2012 in the whole population. The decline was larger for transfused patients (33.8 ± 15.6 ml/day in 2008 to 23.6 ± 10.7 ml/day in 2012 p < 0.001). We observed a nonsignificant reduction in PRBC administered after the institution of guidelines. Hemoglobin threshold for blood transfusion remained unchanged over time. There was a significant relationship between daily blood volume loss and blood transfusion [OR 1.012 per ml volume drawn CI (1.009–1.014), p < 0.001], but mechanical ventilation, renal replacement, invasive monitoring and use of vasoactive drugs were more potent independent factors for transfusion. ICU and in-hospital mortality did not change across time (p = 0.367 and p = 0.361)

Conclusion Guidelines of laboratory tests prescription led to a sustained reduction in the number of laboratory tests and related costs, the average daily blood volume drawn, without changing the ICU and in-hospital mortality. Although there was a correlation between the amount of blood sampled and the transfusion of PRBC, the number of patients transfused and the amount of red blood cells transfused did not significantly decrease.

Competing interests None.

P20 Identifying useless precaution to reduce costs in ICU

Correspondence: Vincent Peigne -

Annals of Intensive Care 2016, 6(Suppl 1):P20

Introduction ICU caregivers can achieve useless precautions to limit stress and feeling of uncertainty about their patients.

We tried to identify and eradicate some behaviors of useless precaution inducing supplementary use of medical devices of drugs in our ICU. We made the hypothesis that suppressing these practices could reduce the variable costs of our 12-bed unit.

Patients and methods In a first step, two senior intensivists analyzed the expenditures of drugs and medical devices during 1 year to identify excessive consumptions related to useless precautions. Strategies were set on to reduce these behaviors. In a second step, the impact of these measures was analyzed 1 year after their implementation.

Results Three behaviors of useless precaution were identified: (1) the daily systematic preparation of two sets of intubation drugs (succinylcholine, ketamine, atropine, and epinephrine), (2) routine bed-side hematocrit measurement in non-bleeding patients and (3) use of carbapenem for community-acquired infections.

Corrective measures included: (1) intubation drugs were routinely gathered but were prepared only if an intubation was decided by the attending physician, (2) bed-side hematocrit was performed only after medical prescription and physicians were encouraged to limit its use to bleeding patients, and (3) recommendations encouraging use of ceftriaxone instead of ertapenem for community-acquired pneumonia and intra-abdominal infections were implemented in the unit.

One-year consumption of the drugs and devices related to these measures decreased dramatically after the intervention: 245 doses of succinylcholine versus 860 (1896€ vs 6656€), 448 doses of ketamine versus 925 (821€ vs 2081€), 750 doses of atropine versus 3234 (133€ vs 728€) and 1660 doses of epinephrine versus 2400 (1444€ vs 2087€). Only 4800 bedside measurements of the hematocrit were performed during the year after the intervention versus 13,200 before (5052€ vs 14,841€). The consumption of ertapenem decreased from 330 doses (12,359€) to 87 doses (3198€).

No adverse event was related to the new practices.

Conclusion Useless precautions were frequent reassurance behaviors in our unit. Identification of such practices and strategies to limit them allowed us to decrease the consumption of drugs and medical devices. The reduction in annual expenditures was superior to 26,000€.

Competing interests None.

P21 The ICU of the teaching hospital of Kinshasa

Correspondence: Joseph Nsiala Makunza -

Annals of Intensive Care 2016, 6(Suppl 1):P21

Introduction The objective of an ICU is to take care of critically ill patients who are presenting at least one potential or effective organ failure impairing directly with the vital prognosis and requiring the setting in emergency of replacement techniques.

According to the complexity of the care bring to these patients, a special attention is focused on the quality of the care in ICU.

Very few studies concerning the quality of the care in ICU are made in countries with low incomes.

The objective of this study was to have a global overview of our ICU and to evaluate our activities.

Patients and methods We made an evaluation of professional practices by following-up indicators of structure, of procedures of care and of results. All medical records of patients admitted in our ICU between 1 January and 30 June 2015 were analysed retrospectively. An investigation of satisfaction a day given was carried out, to collect the opinion of families of patients on the quality of the care given to their neighbours during their stay in our ICU.

Our service is a multipurpose ICU with a capacity of eight beds. It is held by a professor of university assisted by 14 specialists in resuscitation and 34 doctor-assistants (internal) in formation. Only the doctor-assistants take night duties our ICU.

Our equipment is composed by a life support machine and a multiparametric monitor multiparametric for two beds. We have no equipment for measuring gases of blood, neither bronchial fibroscope, nor cardiac echography device. No technique of renal replacement is available.

Results During 6 months, 140 patients (71 % men) were admitted in our service. The mean age was of 46 ± 22 years. The main reasons for admissions were related to medical pathologies (60 %), followed by surgical pathologies (35.6 %) and traumatic pathologies (4.4 %). No gravity score was recorded at the admission

During the stay in ICU, 39 % of patients were ventilated, 27 % received vasoactives amines, no patient received a parenteral nutrition, and only three patients were dialysed. Especially severe septic patients who represent the majority of our admissions, the global compliance in the recommendations of the Surviving Sepsis Campaign “SSC” passed from 50 % during the second half-year 2014 to 33.3 % in the first half-year 2015.

The rate of observance of hand hygiene practice was low (11 %) and the rate of nosocomial infections was high (31.1 %). The mean duration of stay was 8.2 ± 4.2 days, and the mortality rate was 86.6 %.

Surprisingly, the majority of the questioned people declared very satisfied with the care given to their relatives in our ICU.


  • Multipurpose ICU of young adults;

  • Numerous specialists in resuscitation (2md/bed vs 1MD/4beds);

  • No specialists on duty;

  • Lack of nurses (1 inf/4 beds);

  • Lack of equipment;

  • Most patients are ventilated and got nosocomial infections (NI 31.1 %, SAPI 5–40 %, France 5.4 %);

  • Lack of protocols;

  • Not use of gravity score;

  • High mortality.

Conclusion This study shows that the equipment of our intensive care unit has to be optimised. During the first half of 2015, the mortality rate was high and the compliance to recommendations, of SSC, was low. It will be advisable to develop and to promote protocols of care and to ensure the presence of a specialist in resuscitation in our ICU 24 h a day and 7j/7.

Competing interests None.


1. Martin W. Dünser, MD; Inipavudu Baelani, MA; Lundeg Ganbold, MD A review and analysis of intensive care medicine in the least developed countries. Crit Care Med. 2006 Apr;34(4):1234–42.

2. SRLF. Structures et organisation des unités de réanimation: 300 recommandations. Réanimation 12/2012; 21(3). doi: 10.1007/s13546-012-0510-9 .

P22 Security culture and risk management in intensive care units: risk cartography in a French intensive care unit

Correspondence: Amélie Mazaud -

Annals of Intensive Care 2016, 6(Suppl 1):P22

Introduction During their stay in hospital, patients might be confronted to many risks due to health care. In France, health authorities command to identify those risks: either in an a posteriori way for intensive care units (ICU) like morbidity–mortality meetings, or in an a priori way, that is to say before the occurrence of undesirable events, in such process that operating rooms. A priori risk identification is not recommended for ICU, yet ICU patients are very exposed to iatrogenic events, which can eventually conduct to permanent after effects, or even death. Our aim was to identify risks of patients during their stay in ICU, in an “a priori” way, by realising a risk cartography.

Materials and methods This study took place in a French ICU of a general hospital. The methodology used is the one described in French quality reference documents. A volunteer multi-professional group was created, including doctors, nurses, auxiliary nurses, head nurses, and quality engineer. Risks identification meetings took place from April to September 2014. Patient’s general process through ICU was firstly described, from the admission to the discharge of the unit. Risk quotation was made through scales about frequency, seriousness and control of the risks, concerning each risk of the patient’s process. From these quotations, gross criticality was calculated by multiplying frequency and seriousness, and residual criticality by multiplying gross criticality and control. Residual criticalities were classified into three groups: criticality bearable without any action (up to 6), criticality bearable, but actions will be needed later (up to 27), and unbearable criticality, needing immediate action (over 28). Cartography representation was made, as a Kiviat diagram, using the mean, minimal and maximal values of residual criticality for each step of the patient’s process.

Results A 188 risks have been identified during the different steps of the patient’s process in ICU, as follows: admission/welcoming, circulatory care, ventilatory care, blood purification, hygiene/isolation, and discharge. Cartographic representation shows that none of the mean residual criticalities of those previous steps has reached the threshold of unbearable residual criticality. Hygiene/isolation has the worst mean residual criticality (18). Inside every steps but one (blood purification) of the process, some individual risks reached the threshold of unbearable residual criticality, requiring correcting measures. Many actions, including additional lecture of previous procedures, creation of new procedures, creation of healthcare pathways, and training courses for nurses and physicians, were recorded in the 2015 action plan. A new assessment of those actions shall take place in 2016–2017 (Fig. 23).

Fig. 23
figure 23

See text for description

Conclusion Our study of a priori risks during the whole ICU stay of patients is completely unique to our knowledge. It has shown that many point of our practice can be improved rapidly, without waiting for the undesirable event to occur, avoiding potential iatrogenic complications for patients. Furthermore, it is a complete match with the current French trend, commanded by our healthcare authorities, of healthcare management through security culture and risk management. However, the results of this cartography cannot be applied directly to any ICU: The identified risks are specific only to the ICU where has taken place the study. Nevertheless, the methodology used in this study, and probably a good part of the patient’s process through ICU, can be reused in other ICU. That way our study could be an inspiration tool for other ICU eager to improve their risk management.

Competing interests None.


1. Moulaire M. La cartographie des risques, un outil de management des risques en établissement de santé. Risques Qual. 2007;IV(4):221–9.

2. Garrouste-Orgeas M, Timsit JF, Vesin A, et al. Selected medical errors in the intensive care unit: results of the IATROREF study: parts I and II. Am J Respir Crit Care Med. 2010 Jan 15;181(2):134–42.

P23 Acquiring messages of medical articles according to the reading style: the BIBLIOFLASH multicenter study

Correspondence: Romain Sonneville -

Annals of Intensive Care 2016, 6(Suppl 1):P23

Introduction Reading medical articles represents one of the main sources of information for medical doctors. However, the exponential availability of medical articles leads to difficulty in the selection of articles and reading the full version of an article is time-consuming. A reading of a shorter version of the article could represent an interesting and time-saving alternative to the full version.

Materials and methods We aimed to test the impact of an abbreviated reading of a medical article on the understanding and acquisition of medical messages by residents in intensive care medicine. We performed a multicenter prospective study in three medical intensive care units of the Paris area over a 6-month period. Residents in intensive care medicine read articles selected by senior intensivists. For each article, residents were randomly assigned to one of the three following versions: (1) full version of the article; (2) incomplete version, without discussion; and (3) flash version, restricted to abstract, tables and figures. The evaluation of the understanding of the article was evaluated by a senior physician according to a standardized grid, assessing six main points: (1) objectives, (2) study population, (3) design, (4) main outcomes, (5) main bias and (6) external validity. Residents were asked to give the main message of the study. All articles were then discussed with the senior intensivists during a journal club session. One month later, residents were asked to recall the main message of each article. Data are presented as percentages or median (interquartile range).

Results Twenty residents evaluated 25 different articles (randomized controlled studies n = 16, cohorts n = 7). A total of 131 articles were read in three different versions, as follows: full version (n = 50), incomplete version without discussion (n = 35) and flash version (n = 46). The characteristics of the three centers, of residents and of the studied articles were not different between groups. The median time spent on reading the article was significantly lower for the flash version group [15 (10–16) min] than for the complete version group [40 (30–50) min] and for the incomplete version group [30 (21–30) min] (p < 0.0005). Reading only the flash version of the article allowed residents to accurately identify the objective, the population, the design and the primary outcome of the study, with no significant difference compared with the other groups. However, accurate identification of potential bias and of external validity was significantly less frequently observed in the flash version group compared with the two other groups. Moreover, an accurate identification of the main message of the study was less frequently observed in the flash version group (52 %) than in the full version (78 %) and the incomplete version (71 %) groups (p = 0.02). At 1 month, the recalling of the main message from the initial reading was significantly reduced in the full version group (52 vs. 78 %, p = 0.006) and in the incomplete version group (46 vs. 71 %, p = 0.03), while there was no change in the flash version group (61 vs. 52 %, p = 0.4).

Conclusion The reading of a shorter version of a medical article by residents in training accurately identifies the objective, the population, the design and the main outcome of a study. However, it is associated with a decrease in the understanding of important methodological issues which may lead to inappropriate translation of the results into clinical practice. An accurate recalling of the main message of the article at 1 month was observed in approximately 50 % of cases, irrespective of the reading style.

Competing interests None.

P24 ICU portrayal in social media channels

Correspondence: Pauline Perez -

Annals of Intensive Care 2016, 6(Suppl 1):P24

Introduction Intensivists communicate daily with family members about the main diagnosis (‘cardiac arrest’ and ‘acute respiratory distress syndrome’) and various medical complications (‘tracheal intubation’, ‘artificial ventilation’ and ‘septic shock’). Frequently family members go on home and make internet searches concerning the terms they have heard.

Concordance between information obtained on the internet, medical publications and social media channels have rarely been studied.

Materials and methods We used the Google Trend application to retrieve search volumes about the aforementioned diagnosis and complications. We compared search volumes on Google to the number of publications on PubMed from 2004 to 2015. In addition to that, a search through Ritetag (a site for trends in tweets) was conducted.

Results ‘Cardiac arrest’ and ‘septic shock’ searches on Google had a constant distribution over the years. Meanwhile, searches conducted for ‘artificial ventilation’, ‘acute respiratory distress syndrome’ and ‘tracheal intubation’ had a burst starting in 2008. Occurrences in PubMed followed a similar distribution with the exception of ‘cardiac arrest’ that showed two downfalls in 2008 and in 2010. The trends over the time period were, however, different between Google and PubMed. For Tweeter, we have found a median of four tweets per hour for the term ‘cardiac arrest’, whereas for the other terms it was difficult to retrieve any viable information (Fig. 24).

Fig. 24
figure 24

See text for description

Conclusion This preliminary study suggests that searches for medical terms related to intensive care unit show similar results in Google and PubMed. This could suggest that intensivists and family members tend to have access to similar information. Meanwhile, the social media channels have yet to catch up in this domain.

Competing interests None.

P25 The burnout syndrome in a Tunisian intensive care unit: what is the truth?

Correspondence: Hassen Ben Ghezala -

Annals of Intensive Care 2016, 6(Suppl 1):P25

Introduction Burnout syndrome results from a strong emotional interpersonal implication in help professions. It is composed of three dimensions: emotional distress, depersonalization and professional fulfillment. It is frequent in intensive care units all over the world. The aim of our study was to identify the prevalence and to describe the epidemiological profile of burnout syndrome in a Tunisian medical intensive care unit. We tried also to compare the prevalence of the burnout syndrome between caregivers working in the intensive care unit and between the departments of the hospital.

Materials and methods We performed a prospective observational study in the teaching department of emergency and medical intensive care medicine (EMICU) in Zaghouan Hospital in Tunisia. All caregivers who gave their approval to participate were enrolled in the study. Each caregiver answered to a questionnaire about socio-demographic characteristics (age, sex, marital status) and the work conditions (intensive care unit conditions, number of hours of work per day, the number of on-call duty hours at the hospital). The Malasch Burn Inventory score was used to assess the burnout syndrome in our sample.

Results One hundred questionnaires were distributed in the departments of EMICU, surgery and pediatrics. Only 79 were collected, registered and then analyzed.

The participation rate was about 79 %. We found that 58 % of all caregivers working in EMICU had a burnout syndrome and 26 % of them were seriously affected. Forty-two percent of the affected caregivers had a high level of emotional distress, 43 % had a high level of depersonalization, and 39 % had a low level of professional fulfillment.

The mean average age of our sample was 33 years with a standard deviation of 9. We have a female predominance with a sex ratio of 0.46. Fifty-six percent of the caregivers enrolled in the study were married.

From all the persons who answered to the questionnaire, 47 % chose their department of work. More than half of our population (72 %) works in the hospital from < 10 years. The average number of hours of work per week was 41 h. Twenty-seven percent of all caregivers had to work at least two on-call duties per week.

Analytical study showed that among the socio-demographic characteristics, the female sex was significantly associated with the lowest score of professional fulfillment (p = 0.03). Seventy-two percent of caregivers affected by burnout were male with 40 % who were seriously affected. Sixty-one percent of our affected population was married with a severe score in 22.7 % of cases.

General surgery department had the highest level of emotional distress and depersonalization (73 and 64 %) followed by our EMICU department (56 and 60 %), while the level of professional fulfillment was the lowest in pediatric department (14.3 %) followed by our EMICU department (23.5 %).

Conclusion As expected, the prevalence of burnout syndrome is particularly high in our Tunisian emergency and intensive care unit. We found also that it was correlated with socio-demographic characteristics. Our department had one of the highest levels of burnout syndrome in the hospital. We are trying to identify the causes of this high level. Preventive and interventive measures against burnout syndrome should be started. This work is the first step of a multicenter Tunisian study.

Competing interests None.

P26 Resistance of endotracheal tubes measured after extubation in ICU patients

Correspondence: Claude Guérin -

Annals of Intensive Care 2016, 6(Suppl 1):P26

Introduction Increase in resistance of endotracheal tube (RETT) during mechanical ventilation in ICU should reflect reduction in internal diameter due to accumulation of secretions. The aim of this study was to measure RETT after extubation in ICU patients. Our hypothesis was that RETT increased with the length of use of invasive mechanical ventilation.

Patients and methods The study was performed over patients intubated for at least 1 day in our ICU. Once the patient was extubated, the tube was immediately stored in a plastic bag at room temperature and kept in a safe place until bench assessment. This was performed maximal 24 h after extubation as follows. The endotracheal tube was attached to a filter (Hygrobac), and both were set to ASL 5000 active servo lung (IngMar Medical). The lung model was set in passive condition in order to deliver two consecutive breaths at constant flow from 2 to −2 L/s. The filter was tested first, and then, the filter and the endotracheal tube were run. The relationship of pressure (P) to flow was fitted to the following equation P = K1 flow + K2 flow2, where K1 and K2 are constants. P pertaining to endotracheal tube was obtained by subtracting P from filter to P from filter and endotracheal tube. Dividing P by flow led to RETT = K1 + K2 flow. RETT at 1 L/s (cm H2O) was equal to K1 + K2. The relationships of K1, K2 or RETT to length of intubation was analysed by linear mixed model where tube brand and size were factors with random effects.

Results We included 52 patients (34 male) of median (first–third quartiles) age 68 (61–78) years. The median duration of intubation was 5.5 (2–9) days (min 1–max 19 days). Endotracheal tubes were from Mallinckrodt (n = 45), TaperGuard (n = 4) or Rush (n = 3) brands and internal diameter 7.0 (n = 10), 7.5 (n = 39), 8.0 (n = 3) mm. The relationships of K1, K2 or RETT to length of intubation were not significant taking into account both brand and size of endotracheal tubes (Fig. 25).

Fig. 25
figure 25

See text for description

Conclusion Increase in resistance of endotracheal tube used in the ICU is not related to the length of tracheal intubation.

Competing interests None.

P27 Length of endotracheal tube, humidification system and airway resistance: an experimental bench study

Correspondence: Frédéric Duprez -

Annals of Intensive Care 2016, 6(Suppl 1):P27

Introduction After intubation, the resistance of the inspiratory line is usually increased. This rise in airway resistance is associated with the presence of the endotracheal tube, a heat and moisture exchanger (HME) and/or any another tubing in the breathing circuit. During assisted ventilation or in spontaneously breathing patient, these high resistances will increase the work of breathing and negatively impact the chance of weaning. To limit this elevation in resistance, the endotracheal tube is sometime shortened and a heated humidifier used instead of a HME. The aim of this study was to evaluate the actual effect of these strategies on the airway resistance.

Materials and methods A two-compartment model of adult lung (DTL: TTL 1600 Dual Test Lung, Michigan Instrument) was connected to a Dragër Evita 4® Ventilator. The ventilator was set in volume-controlled mode (Vt: 0.5 L; respiratory frequency: 20 bpm; inspiratory flow: 70 L/min). To simulate normal and obstructive clinical conditions, two different resistances were placed at the entry of the airline connected to DTL by using Pneuflo® (parabolic resistor, Michigan Instrument; mean ± SD: 2.5 ± 0.06 and 17.04 ± 0.03 cmH2O/L s−1). Compliance of the artificial lung was set to 70 ml/cmH2O.

The airline was compounded of an endotracheal tube and a humidification system. Three different inside diameters (ID) (Portex® endotracheal tube of 7, 8 and 9 mm) and two humidification systems (HME (Gibbeck Humid Vent® Adult) or Fisher and Paykel MR850 Heated Humidifier®) were evaluated before and after shortening of the tube at 10 cm length. The change in pressure was measured by an analog iWorx station/digital IWx/214. Resistance value was calculated from the following equation:

$$ \begin{aligned}&{\text{Inspiratory}}\,{\text{Resistance}} \\ &\quad = ({\text{Peak}}\,{\text{pressure}} -{\text{Plateau}}\,{\text{pressure}}/{\text{Inspiratory}}\,{\text{flow}}) \end{aligned}$$

Statistical test (Student’s test) was performed. Values are presented as mean ± standard deviation.


Conclusion The use of heated humidifier instead of HME and shortening of endotracheal tube (10 cm of length) allow significant decrease in airway resistance. Both strategies may be helpful for the weaning of obstructive patients. However, the shortening of the tube (relative impact) has a decreasing impact when the tube diameter is growing and even for smaller tube the impact of humidifier is more important (Table 10).

Table 10 See text for description

Competing interests None.


1. Morgan NJ. Air flow resistance of three heat and moisture exchanging filter designs under wet conditions: implications for patient safety. Br J Anaesth 2001; 87: 289–91.

P28 Diagnosis contribution and safety of bronchoalveolar lavage in intensive care unit

Correspondence: Amira Jamoussi -

Annals of Intensive Care 2016, 6(Suppl 1):P28

Introduction Bronchoalveolar lavage (BAL) is a diagnostic tool that explores the deep lung; it can provide useful histological and microbiological information. However, endobronchial injection of important volumes of saline serum may cause hypoxemia. This risk is particularly important to consider in ICU patients. The aim of this study was to determine the diagnostic value and to identify incidents attributable to BAL in intensive care unit.

Patients and methods This was a retrospective study conducted from January 2011 to December 2014 at the respiratory ICU of the Abderrahmen Mami Hospital in Ariana (Tunisia). Were included all patients who underwent BAL in intensive care unit. We recorded demographic, clinical and paraclinical characteristics of patients, results and incidents of BAL.

Results During the 4 years of the study, 100 BALs were performed in 100 patients (55 men and 45 women) with a mean of age of 44 years [15 years—87 ans].

Respiratory history was present in 22 patients and systemic disease in 31 patients, and 42 patients were smokers.

The main cause of admission was acute respiratory failure (97 patients).

On chest X ray, alveolar and/or interstitial damage was found in 93 patients; 78 were bilateral.

BAL was performed under invasive ventilation in 20 patients, under NIV in 20, on oxygen in 54 and for six patients in ambient air.

BAL was contributory to diagnosis in 77 patients: 28 cases of intra-alveolar hemorrhage, 26 cases of bacterial pneumonia (ten community-acquired pneumonia and 16 nosocomial), eight cases of pulmonary pneumocystosis, six cases of active pulmonary tuberculosis, six cases of malignant pulmonary infiltrate, two cases of eosinophilic pneumonia and one case of histiocytosis X.

The major occurring incident was severe hypoxemia in four patients among which three had required endotracheal intubation. In the 96 remaining patients, BAL was safe and harmless.

Conclusion In ICU, BAL is a good and safe tool; it contributes to diagnosis in 77 % of cases. However, the risk of worsening breathing after BAL should be evaluated, especially in critically ill patients.

Competing interests None.

P29 Impact of fiberoptic bronchoscopy performed under noninvasive ventilation on the outcome of critically ill patients: a cohort study over 11 years

Correspondence: Antoine Marchalot -

Annals of Intensive Care 2016, 6(Suppl 1):P29

Introduction Fiberoptic bronchoscopy (FOB) is frequently performed in intensive care unit (ICU) for diagnostic and/or therapeutic procedures. The main complication of FOB is hypoxemia, which can lead to intubation and mechanical ventilation in patients with hypoxemic acute respiratory failure (ARF), and heavily strike the prognosis, particularly in cases of underlying immunosuppression. Noninvasive ventilation (NIV) has been proposed to improve oxygenation during FOB. The purpose of this study was to evaluate the impact of FOB performed under NIV on the outcome of patients with hypoxemic ARF.

Materials and methods We conducted a retrospective cohort study, in the medical ICU of the Rouen University Hospital from January 2003 to December 2013 and compared two different strategies of FOB, under either NIV or spontaneous breathing with standard oxygen therapy by mask (O2). The main evaluation criterion was the incidence of intubation after FOB within 48 h. Secondary criteria were ICU mortality, predictive factors for intubation and ICU mortality, as well as diagnostic and therapeutic yield of FOB.

Results Ninety-two patients were included. In the FOB-NIV, 14 of 39 patients were intubated (36 %) versus 8 of 53 patients in the FOB-O2 group (17 %) (HR 2.58; p = 0.03). Mortality in the FOB-NIV group was 46 versus 12 % in the FOB-O2 group (HR 1.66; p = 0.29). Patients in the FOB-NIV group were older (64 vs 60, p = 0.028), more severe (SAPS2, SOFA, PaO2/FiO2 ratio) on admission or the day of FOB. In multivariate analysis, predictive factors for intubation were higher heart rate (HR per heart beat: 1.03, p = 0.02), SpO2 <95 % (HR 5.07; p = 0.001) and SAPS2 score the day of FOB (HR per unit: 1.06; p < 0.0001). The type of ventilatory support during FOB was not found as an independent predictive factor for intubation (p > 0.05).

Conclusion Performing FOB under NIV in hypoxemic ARF patients seems to be associated with an increased risk of intubation as compared to FOB with O2, but without increase in their ICU mortality. As the ventilatory support used during the FOB does not appear as a predictive factor for intubation within the 48 h after the FOB, this risk appears closely related to the underlying severity of patients. Finally, the choice to perform FOB under NIV by the clinician appears to be guided by this severity.

Competing interests None.


1. Cracco C, Fartoukh M, Prodanovic H, Azoulay E, Chenivesse C, Lorut C, et al. Safety of performing fiberoptic bronchoscopy in critically ill hypoxemic patients with acute respiratory failure. Intensive Care Med. 2013;39(1):45–52.

P30 New single-use flexible bronchoscope dedicated for the invasive mechanical ventilation: a proof of concept

Correspondence: Mai-Anh Nay -

Annals of Intensive Care 2016, 6(Suppl 1):P30

Introduction Bronchoscopy can be particularly challenging in patients with acute respiratory distress (ARDS) who required lung-protective ventilation strategies. In a previous study, we showed that a bronchoscope with an external diameter of 4 mm (or less) would allow safer bronchoscopic interventions. Diameter reduction cannot be at the expense of the suction channel which needs to be sufficient to easily removed thick secretion. From the single-use bronchoscope technology, we proposed an innovation of flexible bronchoscope with a variable outer diameter as follows: a proximal 4 mm diameter for the part inserted in the endotracheal tube, distal head camera with classical diameter (5.3 mm), and 2 mm diameter for the suction channel. The aim of our study was to demonstrate that our prototype of flexible bronchoscope: (1) does not disrupt the protective ventilation, while it is in place in the endotracheal tube (ETT); and (2) enables an efficient suctioning capacity.

Materials and methodsIn vitro We simulated ARDS lungs mechanically ventilated according to actual guidelines. We used four models of flexible bronchoscope (FB): reusable adult, single-use adult, reusable paediatric, and prototype. All combinations between FB and ETTs (7.0–8.0 mm of internal diameters) were evaluated in pressure-controlled ventilation mode (inspiratory pressure of 22 cmH2O for a target tidal volume of 380 mL, respiratory rate at 27/min, PEEP at 10 cmH2O, inspiratory flow at 60 L/min, compliance at 35 mL/cmH2O, inspiratory/expiratory 1:2). We measured the percentage of minute ventilation (MV) delivered compared with baseline and the increase in total positive end-expiratory pressure (ΔtotalPEEP). To test the capacity of the suction channel, we measured the suction flow (water and viscous fluid) with two vacuum levels.

In vivo We described and compared the consequences of bronchoscopy during protective ventilation on two ventilated pigs with a single-use FB and our prototype of FB.

ResultsIn vitro Inserting adult FB in ETTs severely altered ventilation: 19–61 % of the baseline MV was delivered (7.0–8.0 ETTs). On the contrary, the modifications on the ventilation were reduced with the prototype FB (72–87 % of the baseline MV). When the inspiratory pressure was set to guaranty the targeted tidal volume, the ΔtotalPEEP was +13 and +11 cmH2O, respectively, for the reusable and for the single-use adult FB (ETT 7.0), whereas totalPEEP was unmodified for the prototype (as for the paediatric FB). At −150 mmHg vacuum level, prototype FB suction flow was identical (247 ml ± 12 mL/min for water and 176 mL ± 4 mL/min for viscous fluid) to the adult FB and was clearly superior to the suction flow provided by paediatric FB (147 ± 10 mL/min for water and 100 ± 6 mL/min for viscous fluid). Similar observations were made for the 300 mmHg vacuum level.

In vivo

When the adult FB was inserted in the 7.0 ETT of the ventilated pigs, it was not anymore possible to ventilate them with the initial settings (inspiratory pressure of 15 cmH2O, respiratory rate at 27/min, PEEP at 10 cmH2O). Indeed, the inspiratory pressure was increased up to 52 cmH2O to deliver the targeted tidal volume, leading to a total PEEP of 16 cmH2O. On the contrary, no significant difficulties were observed with the prototype of FB.

Conclusion Our prototype of single-use FB did not disrupt the lung-protective ventilation strategies and had still efficient capacity of suction.

Competing interests None.

P31 Incidence of and risk factors for difficult intubation in ICU

Correspondence: Sophie Marque - Icu

Annals of Intensive Care 2016, 6(Suppl 1):P31

Introduction Difficult intubation (DI) remains an issue of concern in critically ill patients. Though its incidence and risk factors are well described in anesthesiology, few authors studied difficult airway management in ICU patients (1). The aim of our study was to describe the incidence and risks factors for DI in our ICU.

Patients and methods We conducted a monocentric, prospective observational study between January 2014 and May 2015. All patients admitted to our ICU and requiring intubation were included in the study. Exclusion criteria comprised age under 18 and refusal from the patients or their relatives to participate.

An intubation procedure was defined as difficult whenever it required more than two laryngoscopic attempts by a senior intensivist or the use of a different technique (stylet, gum bougies, laryngeal masks, fiberoptic intubation, cricothyroidotomy or tracheostomy). Patients’ comorbidities, demographic data and classical risk factors for difficult intubation in anesthesiology, as well as the MACOCHA score, were collected on admission. For each intubation, modalities of the procedure were recorded. A multivariate logistic regression was used to identify risk factors for DI. Variables were selected after univariate analysis with a significant association (p < 0.20).

Results During the study period, 263 intubation procedures were studied. Patients were mostly men (n = 173, 65.3 %), with a mean IGS II of 54 (52–56). The cause of ICU admission was medical for the great majority of them (n = 239, 90 %). A total of 33 (12.6 % [8.8–17.2]) procedures out of 263 turned out to be difficult.

When the MACOCHA score was recorded (n = 153, 58 %), it was significantly higher in the DI group (median = 2 [1–7] vs 1 [1–2]). In this subgroup of 153 patients, the MACOCHA score appeared to be the only predictive factor for DI (OR per point = 1.47, 95 % CI [1.23–1.75]).

In the global population (n = 173), patients with a DI had a significantly higher Mallampati score, had a more frequent history of head and neck disease (n = 7 [21 %] vs n = 4 [1, 72 %], p < 0.0001) and presented more often with a low thyro-mental distance (n = 13 [39 %] vs n = 34 [15 %]) or a reduced mobility of cervical spine (n = 5 [15 %] vs n = 7 [3 %]). After adjustment, a Mallampati score higher than 2 (OR 37.8 [10.6–135.1]) and a limited mouth opening (OR 5.38, 95 % CI 1.1–26.4) were both associated with DI.

Discussion In our study, the incidence of DI was comparable to previous studies. Predictive independent factors for DI (Mallampati score higher than 2 and limited mouth opening) differed from those which recently led to the validation of the MACOCHA score in a medico-surgical ICU population 1. In our medical ICU population, the MACOCHA score often showed to be difficult to record. The Mallampati classification, which represents one of its fundamental items, was not hardly ever feasible in emergency in medical patients, nor was it predetermined by the anesthesiologist (as in scheduled surgery patients).

Conclusion Two clinical items predictive of difficult intubation were identified in our study. The MACOCHA score seemed to be difficult to record in a medical ICU population, but confirmed to be a helpful score when available.

Competing interests None.


1. De Jong A et al., Early Identification of Patients at Risk for Difficult Intubation in the Intensive Care Unit. Am J Respir Crit Care Med 2013.

P32 Endotracheal self-extubation in the intensive care unit: a two-year observational study

Correspondence: Michel Djibre -

Annals of Intensive Care 2016, 6(Suppl 1):P32

Introduction Endotracheal self-extubation (ESE) is a common adverse event in the intensive care unit (ICU). It may carry significant safety risks for the patients and exceptionally cause death. We aimed to investigate the incidence and consequences of ESE in ICU patients and to assess the risk factors for complicated ESE

Materials and methods This is a single-center prospective observational study conducted in a medical and surgical ICU and affiliated intermediate care of 26 beds. From January 2013 to December 2014, all ESEs were prospectively recorded including the circumstances of the event, the characteristics of the patients and the outcome. Quantitative and qualitative variables are presented as mean (standard deviation), median [interquartile range] or number (%), as appropriate.

Results During the study period, 42 episodes of ESE occurred in 41 patients, among 690 intubated patients (6 %). The patients (31 males) were aged of 56.5 [47–72] years and had a median Charlson comorbidity index of 5 [4–7] points. The main reasons for ICU admission were respiratory failure (54 %), postoperative monitoring (14 %), and severe sepsis or septic shock (10 %). The SAPSII score was 42 [33–58]. ICU length of stay (LOS) was 16 [9–26] days, and ICU mortality rate was 19.5 % [95 % CI 0.07–0.32].

At the time of ESE, 35 patients (83 %) were on a respiratory weaning program (PSV), with a pressure support of 12 [8–13] cmH2O and a FiO2 of 35 % [30–40]; 18 patients (43 %) were still sedated, 15 (36 %) had agitation, and 15 (36 %) had physical restraints. ESE occurred 4 [3–9] days after intubation and 64.5 [20–102] h after PSV onset. Nine patients (21 %) had a spontaneous breathing trial before ESE. Altogether, 66 % of ESE occurred around 7:00 a.m. (n = 7), 13:00 p.m. (n = 15) and 7:00 p.m. (n = 6) and 31 % (n = 13) occurred during the weekend. ESE resulted in a complicated outcome (CO) in 24 patients (57 %), including non-invasive ventilation (NIV) (n = 11; 26 %), reintubation (RI) (n = 16; 38 %), bacterial pneumonia (n = 7; 17 %) and death (n = 8; 19 %). On univariate analysis, the need for RI was associated with a prolonged ICU LOS (p = 0.003), an increased ICU mortality (p = 0.03) and a trend for an increased pulmonary infection (p = 0.08). Analysis of specific circumstances before ESE highlights a shorter time from the beginning of weaning (p = 0.03) and a higher pressure support level (p = 0.05) in case of CO.

Discussion Most of ESE occurred during the weaning period and around some critical hours during the day, particularly during the work shifts in the morning and the evening, and at noon. Risk factors associated with a CO, like a shorter time from the beginning of weaning and a higher pressure support level, might reflect insufficient preparation for extubation.

Conclusion ESE is a common and serious adverse event in the ICU, and the need for RI may potentially be associated with higher morbidity and mortality. A prevention policy of ESE is warranted to minimize ESE, particularly in patients at risk of CO.

Competing interests None.

P33 Criteria to assess extubation readiness and prediction of successful weaning

Correspondence: Jean-Daniel Chiche -

Annals of Intensive Care 2016, 6(Suppl 1):P33

Introduction Daily evaluation of patients for weaning from mechanical ventilation (MV) by nurses has been shown to decrease duration of MV. Evaluation is based on the search of weaning criteria allowing the start of a T-piece trial and the assessment of patient tolerance during spontaneous breathing. Nonetheless, reintubation occurs in 5–10 % of cases [1], increases the duration of MV and ICU stay and is associated with significant morbidity and mortality. Failure to wean patients may result from respiratory failure (high respiratory workload, poor gas exchange) or extubation failure (inability to clear secretions, laryngospasm). Weaning criteria commonly used are poor predictors of extubation failure. We have tested the interest of adding to our weaning test-specific criteria to assess muscle strength, cough effectiveness, airway patency and patient’s consciousness (Table 11).

Table 11 See text for description

Patients and methods We have retrospectively reviewed charts of all patients mechanically ventilated in the 24-bed MICU of Cochin University Hospital from 01/2007 to 12/2013. These patients were assessed twice a day by nurses who look for the presence of weaning criteria. Patients who met those criteria underwent a spontaneous breathing trial on T-piece during which tolerance was assessed. All patients data were continuously recorded in a clinical information management system (Clinisoft®, GE Healthcare), including data related to weaning success/failure. In 08/2008, we progressively introduced a five-item new component to our weaning protocol to improve our ability to test muscle strength, cough effectiveness, airway patency and patient’s consciousness. To evaluate the effectiveness of this strategy, we assessed the incidence of weaning failure (defined as respiratory or extubation failure requiring reintubation within 48 h), the incidence of self-extubation, the duration of MV and ICU stay, and survival before (T1), during (T2) and after (T3) implementation of the new component. Informed consent was waived by our IRB for this study. Data are reported as median and interquartile ranges. A P < 0.05 was considered significant for all statistical tests performed.

Results A total of 4488 treated with invasive MV have been included in the study from 01/2007 to 08/2008 (T1, n = 982), 08/2008 to 08/2009 (T2, n = 642) and 08/2009 to 12/2009 (T3, n = 2864). Main results are summarized below.

Conclusion Addition of a new component to an existing nurse-driven weaning protocol to evaluate readiness for extubation was not associated with improved ability to predict successful extubation. These results mandate a thorough analysis of causes of reintubation.

Competing interests None.


1. Thille AW et al. AJRCCM 2013,187;12: 1294–1302.

P34 Domestic accidents during school holidays in children

Correspondence: Nabil Tabet Aoul -

Annals of Intensive Care 2016, 6(Suppl 1):P34

Introduction The accident represents the first cause of death, handicap and hospitalization in children. Most of the accidents arise further to a relaxation of the surveillance and the awareness of the parents. The risk is everywhere, in the house and in the street. Some situations favor the relaxation of awareness as it is the case during the school holidays.

Our aim is to describe the epidemiological profile of the accidents during the school holidays

Materials and methods It is a descriptive observational cohort study during the school holidays from December 2014 to January 2015, at the pediatrics hospital in Canastel in Oran. All children under age of 15 years who are victims of accident were included.

Results One hundred and fifteen children were enrolled. The mean of age was 5.2 years (±4.1) with sex ratio of 1.78. The most frequent consultation in the pediatric emergency was the traumatology (62.6 %) of the cases followed by the burn (32.2 %) of the cases. The majority of the children are from Oran (63.5 %). Eight couples were divorced. The child was alone at the time of the accident in 80.3 % of the cases. 61.3 % of the children live with the big family. The age of the father was between 30 and 40 years in 46.2 % and more than 40 years in 41.5 %, while the age of the mother was between 20 and 30 years in 49.2 % of the cases and between 30 and 40 years in 43.1 %. No attempts of suicide or voluntary accident were noted. The children required a hospitalization in 24.3 % and stay in pediatric intensive care in 4.3 % of the cases. A 9-month-old child died following an inhaled foreign body.

Conclusion The most of the accidents could be avoided. The implementation of a preventive strategy to emphasize the information and the education of the family and on the necessity of creating a safer environment for children seems to be necessary in our city.

Competing interests None.

P35 Pediatric ARDS: interest of prone positioning and NAVA ventilation during veno-venous ECMO

Correspondence: Sylvie Soulier -

Annals of Intensive Care 2016, 6(Suppl 1):P35

Introduction The mortality of ARDS is related to several factors: etiologies, severity of hypoxemia (PaO2/FiO2 ratio or oxygenation index value) and prolonged mechanical ventilation. Sometimes, VV ECMO has been used to rescue patients suffering from pulmonary failure unresponsive to conventional therapies (2). Due to high risk of thrombotic and bleeding complications, its duration and effectiveness may be limited. Prone position (PrP) and lung-protective and ultra-protective mechanical ventilation (tidal volume <2 ml/kg with Peep >10 cmH2O, ±HFOV) are recognized pulmonary rehabilitation strategies. The increase in oxygenation has been reported, associated with VILI reduction and improvement in alveolar recruitment (1). The main objective of the study was to reduce the duration of VV ECMO and ARDS morbi-mortality, by using these respiratory strategies in pediatric ARDS, requiring VV support.

Patients and methods Following the recommendations of Pediatric Acute Lung Injury Consensus conference, we used double-lumen venous cannula VV support, in right internal jugular vein, with echographic control (3). During ECMO, the respiratory strategies were: ultra-protective controlled ventilation (UCV), PrP for 12 h daily, early reversal neuromuscular blockade (NMB) and spontaneous neurally adjusted ventilatory assist (NAVA) ventilation. Durations of ECMO, UCV and NMB were collected, as well as the time of introduction of NAVA during ECMO and extubation after the weaning off ECMO.

Results Six patients suffering from ARDS were admitted in our ICU and included in this study. The age ranged from 21 days to 4 years. The diverse causes of ARDS were VRS bronchiolitis, Enterobacter cloacae pneumonia, systemic inflammatory reaction syndrome (SIRS) and stage 4 drowning and macrophage activation syndrome. No increased cannulation site bleeding and flow issues, notably during the PrP procedure, were observed. The reversal NMB was at 3.7 days (±2.3) during the ECMO. NAVA ventilation was initiated also during the ECMO at 7 days (±2.6), and the duration of VV support was 7.3 days (±3). The patients were extubated 3.9 days (±2.6) after decannulation. All children survived.

Discussion The durations of ECMO and invasive ventilation observed in our study were dramatically short regarding the severity of ARDS (2). Our study showed the successful of respiratory rehabilitation strategies during VV support that included rapid improvement in oxygenation, early NMB weaning, using NAVA ventilation before decannulation and then decrease ECMO duration.

Conclusion VV ECMO, with double-lumen venous cannula, in severe ARDS in pediatric patients, can be optimized by respiratory rehabilitation strategies such as prolonged PrP, UCV and early respiratory weaning with NAVA ventilation. All these strategies seem to improve outcomes of these patients. However, further studies still required to confirm our encouraging results (Table 12).

Table 12 See text for description

Competing interests None.


1. Fineman LD et al. Pediatric Crit Care 2006;7(5):413–422.

2. Flamant C et al. Eur J Pediatrci 2005;164:93–98.

P36 Esogastric pressure measurement as a guide for noninvasive ventilation in unexplained hypercapnic respiratory failure in infants

Correspondence: Guillaume Mortamet -

Annals of Intensive Care 2016, 6(Suppl 1):P36

Introduction In critically ill patients, noninvasive ventilation is classically set on a variable association of clinical parameters. However, this “clinical” setting provides no accurate data regarding the improvement in the respiratory effort and the synchronization of the patient with the ventilator. Esophageal pressure and gastric pressure have been shown to be useful in the ventilatory management of adult patients with acute respiratory failure. The aim of the present study is to describe the value of esophageal pressure and gastric pressure measurements for determining the usefulness and optimal setting of noninvasive ventilation in children with respiratory failure admitted to the pediatric intensive care unit (PICU).

Patients and methods We conducted a one single-center retrospective descriptive study. After the insertion of the esogastric catheter, the study started with a period of spontaneous breathing. After this period, a series of different noninvasive ventilation settings were performed with the first aim to normalize or to maximally reduce the patient’s respiratory effort, reflected by the normalization of the esophageal pressure and transdiaphragmatic pressure swings. The second aim was to obtain the best synchrony between the patient and ventilator.

Results All the six patients had a severe underlying condition with an associated lung disease. Four patients were naive to noninvasive ventilation. The median age of the patients was 7.7 ± 2.8 months. All the patients had severe hypercapnic respiratory failure with a mean PvCO2 of 71.1 ± 8.9 mmHg and an increase in respiratory rate (mean 61.8 ± 20.4 breaths/min). Four patients were assisted by CPAP ventilation and two with BiPAP ventilation.

NPPV indication was proved by Pes measurement in four patients, as assessed on a reduction in the PTPes/min and PTPdi/min from 303.0 ± 112.0 and 329.0 ± 141.0 to 169.0 ± 115.0 and 194.0 ± 106.0 cmH2O s min−1 during NPPV, respectively (p < 0.05). In these four patients, two died and one was still on noninvasive pressure ventilation (NPPV) at PICU discharge. This decrease in respiratory effort was associated with a decrease in RR and PvCO2.

NPPV non-indication was proved by esophageal pressure measurement in two patients. In these two patients, despite the physiological findings, NPPV was continued in one patient after esophageal pressure and gastric pressure measurements because of an important clinical improvement. One of these two patients died.

Conclusion This descriptive and retrospective study shows that a physiological approach, based on the esophageal and gastric pressures measurement, may be gainful to manage noninvasive ventilation, to indicate or non-indicate NPPV and to optimize its settings, especially in a heterogeneous population of critically ill children.

Competing interests None.

P37 Experience Of Antibacterial Treatment Of Bone And Joint Sepsis Of Young Children

Correspondence: Gennadiy Khanes -

Annals of Intensive Care 2016, 6(Suppl 1):P37

Introduction An adequate choice of antibiotics and antifungal drugs with the help of microbiological studies decides the fate of infants with bone and joint sepsis.

Materials and methods Over the past 10 years, microbiological studies were carried out in accordance with the standards and CLSI and EUROCAST. To establish the etiological diagnosis of the disease, we investigated punctate synovial fluid (mycological and bacteriological analyzes) and venous blood (serological and PCR investigations). Simultaneously, we examined the activity of the inflammatory process by determining blood levels of CRP and PCT.

Results Isolated from the synovial fluid of 16 cultures of microorganisms: nine staphylococcus (three S. aureus, three S. epidermidis and three S. heamolyticus); six Enterococcus (four E. faecalis and two T. faecium); and one Bacillus spp.

CRP was 12–96 mmol/l and PCT was 0.5–2.3 ng/LB depending on the timing of onset of the disease.

With active septic process (data CRP and PCT)—staphylococci isolated antibody titer of 1: 1280–2560: and PCR in different staphylococci (Figs. 26, 27).

Fig. 26
figure 26

See text for description

Fig. 27
figure 27

See text for description

Table sensitivity to the antibiotics and the schedule of the selected microorganisms.

Conclusion Studies have demonstrated the practical value of microbiological research in the treatment of bone and joint sepsis.

Competing interests None.

P38 Comparison of toxic and septic shocks in children

Correspondence: Solenn Remy -

Annals of Intensive Care 2016, 6(Suppl 1):P38

Introduction Toxic shocks (TS) and septic shocks (SS) share many common clinical signs of an exacerbated inflammatory response but with different pathophysiological mechanisms. The objectives were to compare their characteristics of shock and their cellular immune response.

Patients and methods We did a monocentric retrospective study of children <18 years old admitted to PICU for TS or SS between January 2011 and September 2014. Immunocompromised and precociously died patients were excluded (death within the first 24 h).

Comparisons of the time course of leukocytes, organ dysfunction (PELOD2 score) and nosocomial infections were made between TS and SS groups. We used time course of lymphocytes, precisely persistent lymphopenia at day 3, like a marker of immunoparalysis.

Results Seventy-nine patients were included, 42 of SS and 37 of TS. Most of them were lymphopenic at day 1 (83.3 vs 97.3 % in SS and TS groups, respectively), and, respectively, 45.2 and 35.1 % were still lymphopenic at day 3.

Significantly, children with TS were more initially lymphopenic and corrected faster than septic patients (p = 2.52 × 10−5).

In both groups, the PELOD2 score was higher at day 1 in patients who are more lymphopenic.

The rate of nosocomial infection was similar between groups and was higher in children with persistent lymphopenia at day 3. In SS, 26.3 % patients with nosocomial infections were still lymphopenic at day 3 and 17.6 % were not, while In TS, 23.0 versus 11.8 %. But the differences were not statistically significant: in SS OR 1.64, 95 % CI [0.3–12.7], and in TS OR 2.2, 95 % CI [0.2–30.7].

TS were less lymphopenic at day 3 and had less secondary nosocomial infections than SS.

Discussion Whether children who had at day 3 lymphopenia are more prone to develop nosocomial infection, as demonstrated in adults, may require a larger cohort.

Conclusion The difference in time course of lymphocytes between TS and SS groups may be explained by the superantigenic activation of lymphocytes in TS group. Our hypothesis is that superantigenic activation leads to a deeper and more precocious immunosuppression, but the recovery is faster than in patients with septic shock.

Immunomodulation therapy perspectives need better knowledge of immunosuppression induced by different severe infectious shocks.

Competing interests None.


1. Decreased T-Cell Repertoire Diversity in Sepsis: a preliminary study. F. Venet, G. Monneret and all. Critical Care Medicine 2013.

2. Hall MW, Knatz NL, Vetterly C, Tomarello S, Wewers MD, Volk HD, et al. Immunoparalysis and nosocomial infection in children with multiple organ dysfunction syndrome. Intensive Care Med. 2011 Mar;37(3):525–32.

P39 Pediatric carbon monoxide poisoning in the Ile-de-France Region, Link between HbCO level and clinical presentation

Correspondence: Jean Bergounioux -

Annals of Intensive Care 2016, 6(Suppl 1):P39

Introduction Carbon monoxide poisoning remains a threat for children. Clinical severity and presentation of carbon monoxide (CO) poisoning are dependent of COHb level. Our aim was to describe the children hospitalized in our department for CO poisoning and to compare HbCO level to clinically reported symptoms.

Materials and methods We are a secondary care center hospital with hyperbaric oxygen therapy facilities and represent the reference center for Ile-de-France region for CO poisoning. The medical records of patients aged between 0 and 16 years with a confirmed diagnosis of CO poisoning, defined as the presence of a COHb level of more than 5 %, were evaluated. Relevant information such as age, sex, source of CO, coaffected family members, month of presentation, time of presentation and presenting symptoms, duration of oxygen treatment in the emergency department, need for admission to an inpatient ward or intensive care unit, and Glasgow coma scale scores was recorded for each patient on preprepared forms.

Results The records of 50 patients were included. Of them, 26 (57.1 %) were female and 24 (42.9 %) were male, with a median age of 7.0 years (range 1 month–16 years) and a mean COHb level of 7 %. Thirty-two patients (64 %) had a COHb between 1.5 and 10 % on presentation, whereas the remaining 18 patients (36 %) had a presenting COHb of >10 %. Neurologic symptoms such as headache, syncope, seizures, and confusion were encountered more frequently in the COHb >10 % group compared with the group with 1.5–10 % COHb levels.

Conclusion In this study, we managed to demonstrate the presence of more frequent symptoms in patients with a COHb level of 10 % or greater. Further analysis revealed that severe symptoms were more pronounced in adolescents and that the severity of symptoms increased with age.

Competing interests None.

P40 Refractory septic shock in children: an ESPNIC definition

Correspondence: Luc Morin -

Annals of Intensive Care 2016, 6(Suppl 1):P40

Introduction Septic shock is a constantly evolving challenge for pediatric intensive care units (PICU) physicians. Despite a decrease in the trend of mortality, part of these patients dies in the first hours of care from refractory septic shock (RSS). A definition for RSS is needed to help its earlier diagnostic, have a better understanding of the disease and evaluate specific treatments. We aimed to define RSS with a two-round Delphi study among the European Society of Pediatric and Neonatal Intensive Care (ESPNIC), to create septic shock prognostic scores on a development cohort and to validate these tools as diagnostic of RSS on a multicentric cohort.

Patients and methods A total of 114 PICU physicians from 27 countries answered a clinical case-based two-round Delphi study. Criteria most associated with diagnosis of RSS in clinical cases were selected, then validated in the second round of the study and compiled in a draft definition. During a 4-year period (2010–2013), all patients admitted for septic shock without limitation of active treatment were screened. Seventy patients from one PICU (21 deaths, 30 %) were included in the development population and 324 patients from four PICUs (39 deaths, 12 %) in the validation population. Two septic shock scores were created after combination of the survey results and the multivariate analysis on the development population.

Results The resulting draft definition for pediatrics constitutes myocardial dysfunction with high blood lactate levels despite adequate vasopressor treatment. Analyzed in the development population, each item of the draft definition was independently associated with need for ECMO or death. Two septic shock scores were created. A simplified septic shock score (4S) ≥2/5 had a mortality positive predictive value of 60.4 and 84.6 % if ≥4/5. The complete septic shock score (cSSS) had a discriminative power measured with the area under the responder-operating curve of 95 % (95 % CI 92–98 %).

Discussion There is no existing score to assess risk of death in children with septic shock. Blood lactates, VIS and septic cardiomyopathy are known factors associated with mortality in septic shock. The combination of the three criteria is a powerful and sensitive way to diagnose patients in RSS. Cutoff proposed for this diagnosis is a 4S ≥2/5 which is associated with a hazard ratio of death of 68.69 (95 % CI 30.47–154.9).

Conclusion Refractory septic shock in pediatrics defined with the ESPNIC selects patients with high mortality rate in PICU. Septic shock scores are useful tools to assess risk of death or need for ECMO at the bedside of the patient (4S) or to compare populations of septic shock patients (cSSS).

Competing interests None.

P41 Impact of the Leonetti law on end-of-life practices in a French NICU and PICU, from January 2011 to December 2014

Correspondence: Stephanie Litzler-Renault -

Annals of Intensive Care 2016, 6(Suppl 1):P41

Introduction End-of-life is a highly debated subject. In France, the Leonetti law was promulgated in 2005 to clarify medical practices regarding this topic. It draws attention to the aspect of unreasonable obstinacy and authorizes withholding or withdrawal of treatments such as mechanical ventilation, vital drugs or artificial nutrition when they appear “useless or disproportionate.” It also emphasizes the necessity to alleviate pain, even when doing so could result in shortening life. Ten years after its promulgation, this law still seems not fully implemented in French Neonatal Intensive Care Unit (NICU) and Pediatric Intensive Care Unit (PICU).

The aim of this study is to assess the implementation of the Leonetti law in our NICU and PICU and to study the evolutions of this implementation over time.

Materials and methods We performed a retrospective observational monocentric study using data collected from medical and nursing charts. For each case, four criteria to be respected regarding the Leonetti law were sought from the charts: collegiate decision regarding the implementation of end-of-life care, parental involvement in the decision, traceability of the decision within the patient medical file, and the way dignity and quality of life that were maintained throughout end of life. This classification has been done by a physician outside our unit, unaware of the child’s situation, in order to lead to a consensual classification. The interpretation of the data was based on univariate analysis, descriptive at first. We then compared those data year by year.

Results Over this period, 94 cases out of 119 were included. The four criteria were met in 40 % of the cases, and this value significantly improved over time [8–52 % between 2011 and 2014] (p < 0.01). Three criteria out of four were met in on average 76 % of the cases. The legal obligation of collegiate decision was respected in 73 % of the cases [58–76 %] (p 0.3), parental involvement in 85 % [58–92 %] (p < 0.01) and traceability in 99 %. Dignity and quality of life were maintained suitably in 53 % of the cases [21–72 %] (p < 0.01). While three criteria out of four tended to improve over time, the traceability rate was stable around 99 %.

Discussion Despite the biases caused by its retrospective nature, our study allowed us to confirm the improvement in our practices in children end of life. However, there is still room for improvement, particularly regarding collegiate decision (e.g., systematic presence of an external consultant) and assessment of pain and discomfort in end-of-life situations.

Conclusion Laws and medical recommendations cannot bring all answers to end-of-life situations. However, many elements regarding the implementation of the Leonetti law within our units still need to be improved. Following this study, a series of reflections on ethical issues aiming at fulfilling the predefined legal criteria will be set up within our units.

Competing interests None.


1. Loi n°2005-370 du 22 avril 2005 relative aux droits des malades en fin de vie. J Off Repub Fr Ed Lois Decrets 2005; 95: 7089. Disponible sur: .

2. Limitations et arrêts des traitements en réanimation pédiatrique de la décision à son application. Hubert P., Centre Laennec 2009; 57:22–34.

P42 Mortality risk factors for anticoagulant-related bleeding

Correspondence: Ines Sedghiani -

Annals of Intensive Care 2016, 6(Suppl 1):P42

Introduction Vitamin K antagonists (VKA) have served as the cornerstone of long-term anticoagulant therapy. Out of the respect of the recommendations of prescription and biological monitoring, this treatment can be complicated by life-threatening bleeding events.

Patients and methods We carried out a retrospective study of 58 patients taking oral anticoagulants and who were hospitalized for hemorrhagic events from January 2013 to June 2015.

The aim of this study is to describe the characteristics of patients treated with VKA presented a bleeding event, to study the hemorrhagic event and its management and to search for prognostic factors.

Results Among the 58 patients (24 men and 34 women, aged 69 ± 11 years) with anticoagulant-associated bleeding event, 41 patients survived and 17 patients died. All patients were already taking acenocoumarol for 65 ± 61 months. VKA were used for the prevention of stroke due to atrial fibrillation in 41 (70 %) cases, prosthetic cardiac valve replacement in 13 (22 %) cases and venous thromboembolism in four (7 %) cases.

Conclusion The results of this study showed that IGSII ≥37, intracranial and gastrointestinal bleeding and elevated AST level are useful prognostic factors for predicting anticoagulant-associated bleeding event mortality (Table 13).

Table 13 See text for description

Competing interests None.

P43 Management of the polytrauma in the emergency department of CHUOran

Correspondence: Soumia Benbernou -

Annals of Intensive Care 2016, 6(Suppl 1):P43

Introduction The polytrauma is a major public health problem, and assessment of severity is an important part of the initial management which determines prehospital means necessary.

The host structure must take into account the particularities of these patients.

The need for a complete and rapid lesion balance makes desirable the management of trauma patients in centers with a complete technical platform.

Some studies have shown that up to 30 % of deaths in trauma patients could have been avoided by better management.


The goal of this study was to define the epidemiological profile of polytrauma at the service of medical and surgical emergencies CHUOran.

Patients and methods Retrospective study of trauma patients admitted to the intensive care unit of the UMC from January 1, 2009, to December 31, 2011.

Data collection was carried out through a pre-established record where data are noted. The coding data entry and data analysis are performed using the EpiInfo Version 6 software.

The main items of the plug are age, sex, mechanism, collection, time of admission hemodynamic, respiratory and neurological constants, initial lesion balance, surgery, secondary lesion balance, ICU hospitalization and evolution of the patient.

Results A total of 217 severe trauma patients were collected during the period from January 2009 to December 2011.

We note a clear predominance of the male sex with a sex ratio of 8.77.

Most traumatized patients were young with an average age of 34.217 years.

The accident in the way was the origin of the trauma in 77.4 % followed by the fall with a frequency of 22.6 %.

The collection was medicalized in 39.6 % of cases. 11.5 % were transferred from other cities.

20.7 % had systolic blood pressure ≤90 mmHg and 23 % had SpO2 ≤90 %.

11.1 % had a Glasgow score ≤3/15. For 4.6 % of them, it was of 4/15, 6 % had a score of 5/15, 8.3 % had a score of 6/15, 13.8 % had a score of 7/15, 11.2 % had a score of 8/15, and 45 % had a score ≥9/15 on the Glasgow coma scale. 18 % had anisocoria on admission and 1 % was in mydriasis.

The revisited trauma score was ≤5 in 19.4 % of patients.

43.78 % had an orthopedic injury. 62.7 % were intubated dice admission.

Chest radiography was performed in 92.6 % of patients, and that of the basin and abdominal ultrasound was made in 94 % of them.The computed tomography was performed in 95.4 % of traumatized patients, and that of the thorax and abdomen was with a respective proportion of 76 and 39.6 %. 85.3 % of our patients were polytraumatized, and trauma of the head was the most frequent lesion with a proportion of 85.3 % followed by the thorax with a frequency of 62.8 % and finally the abdomen with a rate of 21 %. The average ICU stay was 19.19 days. The outcome was unfavorable in 30.9 % of patients

Discussion In our study, we found that the polytrauma mainly touched young males under 40 years old. This constitution is similar to that reported by David J. Thurman and collaborators in the published report on the brain injury in the USA. In the Annual Report of 2013 of the Trauma Data Bank and by the state of 805 health centers in the USA, the fall was the leading cause of head injury with a frequency of 40.66 % followed by accident thoroughfare of 28.1 %.

At a time where the prehospital medicalization is the subject of numerous publications and a wide debate between two different management systems:

  • In the USA, initial support is provided by paramedics, who after treatment of respiratory and hemodynamic distress use the shortest time possible to transfer the patient to the nearest trauma center, and this concept is called “scoop and run.”

  • In France and in some European countries, the management is performed by physicians who take all the time necessary for the treatment of different afflictions before transferring the patient to the hospital, and it is called “stay and play.”

Our study has found that the initial management of patients was not medicalized in 60 % of patients. First, the multicenter study was able to show that prehospital medicalization of patients was associated with a significant reduction in risk of death in 30th post-traumatic day.

Conclusion The care of the injured polytrauma is complex and must not be improvised in any case.

This management must be rational and constantly evaluated in order to improve the quality and results. It concerns civil society, hospital practitioners and public authorities.

Competing interests None.


1. David J. Thurman, Clinton Alverson, Doug Browne Traumatic Brain Injury in the United States: A Report to Congress 1999.

2. J.-M. Yeguiayan, D. Garrigue, C. Binquet, C. Jacquot, J. Duranteau, C. Martin, F. Rayeh, B. Riou, C. Bonithon-Kopp, M. Freysz*. Prise en charge actuelle du traumatisé grave en France : premier bilan de l’étude FIRST (French Intensive care Recorded in Severe Trauma) Ann. Fr. Med. Urgence (2012) 2:156–163. doi: 10.1007/s13341-012-0181-1 .

P44 Realization of a feedback system of respiratory parameters during CPR

Correspondence: Lhoucine Ben Taleb -

Annals of Intensive Care 2016, 6(Suppl 1):P44

Introduction The objective of this work is to achieve a system for monitoring parameters of respiratory tract and ensure feedback on the quality of chest compressions during cardiopulmonary resuscitation (CPR).

Materials and methods This prototype is designed by flow and pressure sensors that plug at the respiratory tract, as well as efficient data acquisition card which processes and monitors signals. Communication with the user is done by visual interfaces (graphical and numerical) via an efficient development tool LABVIEW 2014 (National Instrument).

The settings for the compression quality, frequency, flow, volume and respiratory tract pressure are recorded and displayed during CPR.

Results The figure presents a preliminary test and calibration of the system’s sensors on a graphical interface using the measurement system (Fig. 28).

Fig. 28
figure 28

See text for description

Conclusion The realized system will give a feedback on respiratory parameters generated by chest compressions. Additional tests are expected to validate this prototype in more realistic situations.

Competing interests None.

P45 Burns from French military operations: a 14-year retrospective observational analysis

Correspondence: Clément Hoffmann -

Annals of Intensive Care 2016, 6(Suppl 1):P45

Introduction Burns in combat usually represent 5–20 % of injuries [1]. The purpose of the study was to describe epidemiological characteristics and the management of burns during French military operations.

Patients and methods A retrospective observational study was performed of French casualties with burns, transported to Percy Burn Center (R4 Medical Treatment Facility) between January 1, 2001, and December 31, 2014.

Results Forty-six patients were included. Seventy-six percent of burns analyzed were non-battle injuries, mainly due to mishaps occurring with the burning of waste and the use of fuel. Combat burns (n = 11/46) were all explosions related (improvised explosive devices). The median total body surface area burned was 15 % [9; 27]. Major anatomical regions affected were upper limbs (91 %), head/neck (78 %) and hands (70 %). Battle-related burns resulted in higher median Injury Severity Score, more patients under mechanical ventilation, more blood transfusions and surgeries, respectively, 16 [6; 23] versus 4 [1; 8], 64 versus 17 %, 8 [0; 17] versus 0 and 3 [0; 4] versus 0 (p < 0.05). The case fatality rate was 2.2 % (n = 1/46).

Discussion Majority of burns are non-battle injuries, small in size and accessible to prevention [2]. For battle-related burn injuries, the anatomical topography can be explained by the personal protective equipment and the higher severity score due to the associated trauma and mechanism. This study is only the visible tip of the iceberg.

Conclusion All military care providers should be familiar with the assessment and treatment of burns in military settings.

Competing interests None.


1. Cancio LC et al. J Burn Care Rehabil 2005;26:151–161.

2. Kauvar DS et al. J Burn Care Res 2009;30:700–704.

P46 Drowning in Martinique—characteristics and infectious complications

Correspondence: Dabor Resiere -

Annals of Intensive Care 2016, 6(Suppl 1):P46

Introduction Drowning is a public health problem, with a high incidence of 500,000 people worldwide, and 150,000 deaths each year in mainland France and the overseas territories. This is one of the leading causes of accidental death in children and adolescents. Immersion exposes them to multiple complications that are often life threatening. Infectious complications are, however, poorly understood and not well described. The objective of this study was to define the frequency and nature of lung infections in victims of drowning admitted to the emergency and ICU departments at the University Hospital of Martinique.

Patients and methods We conducted a retrospective study from victims of drowning cases admitted to the ER and ICU CHU Martinique over a period of 3 years (2012–2015), supplemented by data from SAMU 972 records, including presentation of descriptive data and univariate predictors of death.

Results One hundred and forty-four patients were included [102 men/42 women, mean age 36.5 years (range 1–83 years)]. Drownings occurred in seawater (N = 120, 83 %); swimming (N = 15, 10 %); freshwater (N = 5, 3 %) or polluted water (N = 1, 1 %). Seventy-eight patients (54 %) had comorbidities including hypertension (N = 14, 10 %); diabetes (N = 10, 7 %); epilepsy (N = 12, 8 %); heart disease rhythm and ischemic (N = 10, 7 %); chronic alcoholism (N = 8, 6 %); and obesity (N = 6, 4 %). The majority of victims (31 %) were in deep coma (score Glasgow: 3) and severe hypoxia; 32 victims (22 %) suffered cardiac arrest. Complications were as follows: multiple organ failure (97 %), lactic acidosis (75 %, mean arterial pH 7.20), acute respiratory failure (49 %) and ARDS (16 %, 2 refractory ARDS patients required veno-venous ECMO). Bacterial pneumonia (N = 22, 15 %) was the main cause of infectious complications, early onset (within 24 h). Antibiotic therapy was started probabilistically on admission (N = 85, 59 %), with amoxicillin/clavulanic acid (92 %), cefotaxime (6 %) or cefotaxime metronidazole+ (1 %). Only 17 % of victims had blood cultures and 52 (36 %) lung sampling. Antibiotic therapy was administered before performing lung samples (82 %). The most frequently encountered germs were Aeromonas hydrophilia, Enterobacter aerogenes, Enterobacter cloacae, Morganella and Morgani. The average length of ICU stay was 7 days. Eighteen of these patients (13 %) died in the ICU. Predictors of mortality endpoints were: the occurrence of a pre-hospital cardiac arrest (p < 0.0001), ARDS (p < 0.001), the immersion time (p < 0.05), metabolic acidosis (p < 0.05), the Glasgow coma scale score <5 (p < 0.01), the resuscitation period (p < 0.01) and hypothermia (p < 0.05).

Conclusion The prognosis remains poor after immersion in the absence of a rapid response and appropriate treatment, especially in cases of occurrence of out-of-hospital cardiac arrest (OHCA).

Competing interests None.


1. Ouanes-Besbes L, F Dachraoui, Ouanes I Abroug F. Drowning pathophysiological and therapeutic. Réanimation 2009.

P47 Prevalence, characteristics, and associated factors to physician’s conflicts in Emergency Department: A multicentric Moroccan study

Correspondence: Jihane Belayachi -

Annals of Intensive Care 2016, 6(Suppl 1):P47

Introduction Despite evidence that conflicts are common and harmful in the Emergency Department (ED), no study has been described in the literature. This study aimed to evaluate the prevalence and characteristics of conflict in ED physicians and factors associated with still ongoing conflict.

Materials and methods This was an observational cross-sectional survey conducted among physicians in 12 Moroccan ED, from June 2012 to January 2013. Socio-demographics, work characteristics, conflict-related variables, burnout level and self-estimation measures were included. Conflict-related question was formulated in the following way: In the past years, have you been drawn into any kind of conflict in ED? Response alternatives were (yes or no). To those who answered in the affirmative, the followed characteristics of conflict were collected; numbers of episodes of conflicts (<5 or ≥5). (a) The source of the conflict (Control, Prevision, Resources, Organisation, Coordination); (b) the form of the conflict; (Conflict of content, Conflicts of persons, Conflicts of procedure); (c) the degrees of the conflict: (Confrontation of viewpoint, Destructive personal attacks); (d) the conflict is important or unimportant; (e) the result of the conflict is good or poor; (f) The conflict is terminated or still ongoing; (g) the response to the conflict: (Avoidance, Accommodation, Competition, Collaboration). Comparison between “terminated conflict” emergency physicians group and “still ongoing conflict” group was made. Multiple logistic regression was used to evaluate factors associated with still ongoing conflict.

Results A total of 101 emergency physicians were included with a response rate of 81 %. The mean age of ED physicians was 40 ± 8 years, and 70.3 % among them are men. Conflicts were reported in 97 % of emergency physicians, 49.5 % reported more than five episodes of conflicts, and 41 % of conflict is still ongoing. The principal sources of conflict were: organisation in 82.2 % of case, resources 87.1 % of case and coordination 81.2 % of case. Comparison between the two groups showed that ED physicians with still ongoing conflict have a longer years of practices (11.6 ± 5 vs 9.1 ± 6.8, p = 0.02). Concerning burnout, ED physicians with still ongoing conflict are more emotionally exhausted (37.3 ± 9.7 vs 32.6 ± 12.8, p = 0.04) and have lesser personal accomplishment (27.5 ± 3.6 vs 32.6 ± 9.4, p = 0.01). Concerning self-rated scales, ED physicians with still ongoing conflict perceived a worst mental health (6.6 ± 1.3 vs 7.3 ± 1.3, p = 0.01) are less joyful (6 ± 1.2 vs 6.7 ± 1.5, p = 0.02).

Multiple logistic regression showed that longer years of practice (OR1.2, 95 % CI 1.02–1.31; p = 0.02) and emotional exhaustion (OR 1.2; 95 % CI 1.004–1.11; p = 0.03) were the factors associated with still ongoing conflict.

Conclusion The prevalence of conflict is very high in ED. Longer years of practice and emotional exhaustion in physicians was associated with still ongoing conflict. This high prevalence of conflict suggests that conflict in the ED may be a universal phenomenon that should be addressed.

Competing interests None.

P48 Serious peripartum complications needing admission in obstetrical ICU: retrospective study about 127 cases

Correspondence: Khalil Tarmiz -

Annals of Intensive Care 2016, 6(Suppl 1):P48

Introduction Peripartum morbidities present a clinical challenge due to maternal physiological adaptations related to pregnancy. Pregnant women may be afflicted with surgical condition and/or medical requiring admission to intensive care unit. The aim of our study is to identify serious complications occurred during pregnancy and childbirth in our health structure and analyze the epidemiological aspects, the management and the development of these complications.

Patients and methods We conducted a retrospective descriptive study of 127 patients admitted in obstetric ICU for severe peripartum complications, during the period from January 2011 to December 2013.

Results The average age of patients was 31 ± 5 years, ranging from 19 to 45 years. Eighty-six women were classified as ASA 1 (67 %), and 41 had a medical history including cardiomyopathy (29 %). The average gestity in these patients was 2 ± 1 [1–8], the parity was 1 [0–6], and the medium-term pregnancy at the time of hospitalization was 34 weeks [28–42]. The main reasons for admission were the management of postpartum hemorrhage in 36 % of cases followed by severe preeclampsia in 31 % of cases. The admission to gravity scores were 10 ± 7 for APACHE 2 [0–35], 17 ± 13 for the IGS2 and 3 ± 3 for the SOFA score. Artificial ventilation was necessary in 42 % of cases. The duration of ventilation was 2 ± 2 days [0–16]. The main therapeutic implemented during the ICU stay has been in order of frequency: the antibiotic therapy in 43 % of patients, anticonvulsants (magnesium sulfate) and antihypertensives (40 %), transfusion of blood products (27 %) and use of vasopressors (15 % of cases). Fifty-six patients had at least one complication during hospitalization. The most frequent one was acute kidney failure, infectious and respiratory complications. The average hospital stay was 5 days [1–85]. Five patients died.

Conclusion Our study confirms that direct obstetric complications, mainly hypertensive and hemorrhagic, predominate. Severity scores are lower than for the general population resuscitation, the observed mortality was lower than the predicted mortality, and longer durations of hospitalization and ventilation are shorter. A structure of obstetric ICU could promise earlier management of maternal conditions at high risk of serious complications.

Competing interests None.

P49 Epidemiology of suicide in emergency department

Correspondence: Neila Maaroufi -

Annals of Intensive Care 2016, 6(Suppl 1):P49

Introduction This study aimed to illustrate the characteristics of suicide attempters treated in the Emergency Departments of a general hospital in Tunisia and to provide relevant data for early psychological treatment.

Patients and methods Between January 2009 and December 2014, 571 suicide attempters were treated in the Emergency Departments. Data were collected using a semi-structured questionnaire. Descriptive statistics, Chi square tests, and multivariate analyses were used to identify the factors associated with suicidal behaviors

Results Females outnumbered males at a ratio of 3.5–1. The greatest proportion of cases was in the age group of 15–30 years (62.7 %). Patients who finished middle school or high school accounted for most of the suicide attempters. The most common method used for attempted suicide was drug ingestion (85.5 %). The majority of cases attempted suicide at home. Marriage frustration, work and study problems, family fanaticism and conflict, somatic disease, and history of mental disorders were all significantly associated with suicide attempts. The ratio of patients to be discharged or to die was similar in occupation, marital status, and the place of suicide attempt; however, the results were different in gender, age, educational level, methods used for suicide, time of day, and reason.

Discussion and conclusion Suicide is an important public health problem. A previous suicide attempt is a risk factor for suicide; thus, assessing the characteristics of suicide attempts or instrumental suicide-related behaviors with/without injuries is necessary to prevent these attempts.

Competing interests None.

P50 Beneficial cardiovascular effects of O-GlcNAc stimulation in early phase of septic shock

Correspondence: Julien Cadiet -

Annals of Intensive Care 2016, 6(Suppl 1):P50

Introduction Septic shock is a worldwide burden with over 750,000 cases per year in the US only. This pathology, characterized by a systemic inflammation, is associated with a severe cardiovascular dysfunction. Currently, there is no specific treatment for the septic shock beside antibiotherapy, fluid resuscitation and vasopressor amine. Finding new therapeutic leads in septic shock is important because this disease is a major public health problem. Recently, Not et al. showed, during hemorrhagic shock in rats, a beneficial effect of the O-GlcNAcylation stimulation on cardiovascular parameters. O-GlcNAcylation, a posttranslational modification, is the end product of the hexosamine biosynthetic pathway (HBP). We postulated that increase in total protein O-GlcNAcylation at the early phase of septic shock could improve cardiovascular function and reduce mortality.

Materials and methods To induce an endotoxemic shock, rats (n = 6–8) received iv either lipopolysaccharide (LPS, 5 mg/kg) or saline (CTRL). After 1 h, fluid resuscitation (FR, 15 ml/kg of colloid, iv) was associated or not with HBP substrate: glucosamine (GlcN, 180 mg/kg) or an O-GlcNAcase inhibitor (NButGT, 10 mg/kg). Two hours later, echography and mean arterial pressure (MAP) evaluation were performed; blood samples and heart were then collected to evaluate biological parameters (lactate, troponin T, creatinine), inflammation, autophagy and total O-GlcNAcylation by western blot. Survival analysis was performed in parallel.

Results In vivo studies showed an hypotension in LPS restored by FR and treatments (MAP: CTRL 87 ± 1, LPS 73 ± 5*, LPS-FR 93 ± 6#, NButGT 83 ± 5, GlcN 92 ± 4 mmHg, *: p < 0.01 vs CTRL, #: p < 0.01 vs LPS) and a systolic dysfunction in LPS with a trend toward improvement by NButGT and GlcN (ejection fraction: CTRL 81 ± 2, LPS 65 ± 3*, LPS-FR 69 ± 2, NButGT 75 ± 2#, GlcN 74 ± 2 %, *: p < 0.01 vs CTRL, #: p < 0.05 vs LPS). NButGT and GlcN efficiently increased total O-GlcNAc (200 and 300 %, respectively, vs LPS-FR). This effect was associated with a reduced cardiomyocyte insult, tissue dysoxia and renal dysfunction (creatinine: CTRL 25.4 ± 1.2, LPS 51.4 ± 5.8*, LPS-FR 77.5 ± 6.8*, NButGT 33.7 ± 4.7£, GlcN 38.7 ± 8.2 μmol/l£, *: p < 0.01 vs CTRL, £: p < 0.05 vs LPS-FR). Moreover, while all LPS + R rats did not survive after 10 h, 20 % of treated rats survived after 18 h. Interestingly, NButGT and GlcN effects were not associated with modification of autophagic pathway (Beclin-1, LC3, Bcl-2) or inflammation (TNF-α, IL-6).

Conclusion At the early phase of septic shock, NButGT and GlcN induce an increase in cardiac O-GlcNAcylation leading to improve tissue oxygenation and renal function and to reduce cardiac injury. These treatments improved cardiovascular function. The next step is to determine the mechanisms involved in these beneficial effects.

Competing interests None.


1. PMID: 20016375.

2. PMID: 17909453.

P51 MiR-21 deficiency increases mortality during experimental septic shock

Correspondence: Hafid Ait-Oufella -

Annals of Intensive Care 2016, 6(Suppl 1):P51

Introduction MicroRNAs (miRs) are noncoding RNAs found in all eukaryotic cells that regulate the human genome, primarily through translational repression. MiR-21 is expressed by inflammatory and endothelial cells in response to bacterial LPS, TNF-α or apoptotic debris. Mir-21 limits the production of cytokines and the expression of endothelial adhesion molecules and modulates cell survival. Most of the miR-21 targets being involved in the pathophysiology of septic shock, we hypothesized that miR-21 could play a role in tissue injury following bacteria invasion.

Patients and methods We demonstrated genetic invalidation (miR-21−/−) or pharmacological inhibition of miR-21 (antagomiR-21). We reviewed mouse models of severe infections, endotoxinemia (LPS 15 mg/kg) and non-resuscitated peritonitis.

Results In vitro, LPS-stimulated miR-21−/− splenocytes produced more IL-12p70 than miR-21+/+ splenocytes (p < 0.05). In addition, miR-21 −/− bone marrow-derived macrophages stimulated by LPS produced more IL-12p70, Il-1b, TNF-a and less Il-10 than control miR-21+/+ macrophages (P < 0.05). In a mouse model of endotoxinemia, genetic invalidation (miR-21−/−) of miR-21 significantly increased mortality (P < 0.01). Pharmacological inhibition of miR-21 using antagomiR-21 significantly increased mortality in two models of lethal infections (P < 0.01, endotoxinemia and peritonitis). MiR-21+/+ and miR-21−/− mice were irradiated and retransplanted with MiR-21+/+ and miR-21−/− bone marrow. Four groups of chimeric mice were generated and challenged intraperitoneally with LPS. Interestingly, both miR-21 deficiency in leucocytes and in endothelial cells increased mortality in a non-redundant manner.

Conclusion MiR-21 deficiency induced a deviation of the immune response towards a pro-inflammatory profile and increased mortality during experimental shock septic.

Competing interests None.


1. Sheedy et al. Frontiers in Immunology 2015.

P52 Isolated heart rate reduction by ivabradine does not improve cardiac and vascular function in experimental septic shock

Correspondence: Antoine Kimmoun -

Annals of Intensive Care 2016, 6(Suppl 1):P52

Introduction In recent studies, β1-adrenergic blocker demonstrated cardiovascular protection in sepsis. Experimental research showed that these benefits are associated with a down-regulation of inflammatory pathways. However, the specific impact of heart rate reduction on the inflammatory pathways remains unknown. Ivabradine is a pure heart rate-lowering drug that acts specifically on the sinoatrial node by selectively inhibiting the If current. The present study was designed to assess the effects of heart rate reduction by ivabradine on cardiovascular function and on the inflammatory pathways involved in peritonitis-induced septic rats.

Materials and methods Sixteen Wistar male rats were treated with cecal ligation and perforation (CLP) to evoke peritonitis. Four hours after CLP, rats were randomly allocated to CLP (n = 8) and ivabradine (n = 8, administrated per os at H4) groups. Another eight Wistar male rats underwent sham operation. All rats received a continuous infusion of saline [10 mL/(kg h)], analgesic [nalbuphin: 0.2 mg/(kg h)] and antibiotic (imipenem and cilastatin sodium: 10 mg/kg) 4 h after the surgery. Assessment at 18 h included in vivo cardiac function by echocardiography and ex vivo vasoreactivity by myography. Cardiac and vascular protein expressions of NF-κB and eNOS/Akt/iNOS pathways were also assessed by western blotting.

Results Compared with Sham animals, CLP induced tachycardia (376 ± 26 vs. 431 ± 22 bpm, p < 0.05), hypotension (111 ± 7 vs. 80 ± 7 mmHg, p < 0.05), decrease in cardiac output (95 ± 13 vs. 56 ± 15 mL/min, p < 0.05), hyperlactatemia (1.2 ± 0.3 vs. 2.5 ± 0.5 mmol/L, p < 0.05) and vascular hypo-responsiveness to vasopressors. Adjunction of ivabradine decreased heart rate (329 ± 38 vs. 431 ± 22 bpm, p < 0.05) without any impact on blood pressure (73 ± 10 vs. 80 ± 7 mmHg, p > 0.05), cardiac output (37 ± 18 vs. 56 ± 15 mL/min, p > 0.05), lactatemia (2.6 ± 0.6 vs. 2.5 ± 0.5 mmol/L, p > 0.05) and vascular responsiveness to vasopressors, compared with CLP group. Results of western blot will be provided later.

Conclusion Isolated reduction in heart rate by ivabradine in an experimental model of septic shock does not result in any effect on cardiovascular function.

Competing interests None.


1. Morelli A, Ertmer C, Westphal M, et al. Effect of heart rate control with esmolol on hemodynamic and clinical outcomes in patients with septic shock: a randomized clinical trial. JAMA. 2013 Oct 23;310(16):1683–91.

2. Katsuya Mori, Hiroshi Morisaki, Satoshi Yajima, et al. Beta-1 blocker improves survival of septic rats through preservation of gut barrier function. Intensive Care Med (2011) 37:1849–1856.

3. Gareth L. Ackland, Song T. Yao, Alain Rudiger, et al. Cardioprotection, attenuated systemic inflammation, and survival benefit of b1-adrenoceptor blockade in severe sepsis in rats. Crit Care Med 2010 Vol. 38, No. 2.

4. Jerome Aboab, Veronique Sebille, Mercé Jourdain, et al. Effects of esmolol on systemic and pulmonary hemodynamics and on oxygenation in pigs with hypodynamic endotoxin shock. Intensive Care Med (2011) 37:1344–1351.

5. Takeshi Suzuki, Hiroshi Morisaki, Ryohei Serita, et al. Infusion of the b-adrenergic blocker esmolol attenuates myocardial dysfunction in septic rats. Crit Care Med 2005 Vol. 33, No. 10.

6. Satoshi Hagiwara, Hideo Iwasaka, Hayato Maeda, et al. Landiolol, an ultrashort-acting beta1-adrenoceptor antagonist, has protective effects in an LPS-induced systemic inflammation model. SHOCK, Vol. 31, No. 5, pp. 515–520, 2009.

7. Antoine Kimmoun, Huguette Louis, Narimane AlKattani, et al. β1-Adrenergic Inhibition Improves Cardiac and Vascular Function in Experimental Septic Shock. Crit Care Med. 2015 Sep;43(9):e332–40.

8. Otilia Postea and Martin Biel. Exploring HCN channels as novel drug targets. Nat Rev Drug Discov. 2011 Nov 18;10(12):903–14.

9.Otilia Postea and Martin Biel. Exploring HCN channels as novel drug targets. Nat Rev Drug Discov. 2011 Nov 18;10(12):903–14.

P53 Exploration and modulation of the EGF-R during experimental septic shock

Correspondence: Hafid Ait-Oufella -

Annals of Intensive Care 2016, 6(Suppl 1):P53

Introduction The receptor of epidermal growth factor (EGF-R) is expressed by mesenchymal and immune cells. EGF-R is involved in survival, migration and proliferation. Pharmacological EGF-R inhibiting therapy using either neutralizing mAbs or small-molecule tyrosine kinase inhibitors (TKI) has been successfully used in cancer treatment. Few evidences suggested that EGF-R could modulate directly the immune system, but the consequences in response to bacteria injury remain unknown. We aimed to investigate the effects of EGF-R blocking in an experimental model of septic shock.

Patients and methods We demonstrated the specific invalidation of Egf-r in myeloid cells (LysMCre Egf-rlox/lox) or pharmacological inhibition of EGF-R using TKI in vitro (AG-1478) and in vivo (Erlotinib, Tarceva®). We reviewed the model of endotoxinemia by intraperitoneal injection of LPS (15 mg/kg). Cell culture of splenocytes, purified bone marrow neutrophils (isolation kit Miltenyi) and bone marrow-derived macrophages (BMDMs) were tested.

Results In vitro, LPS-stimulated LysMCre Egf-rlox/lox splenocytes produced more IL-6 (+300 %, P < 0.05) and more IL-10 (+70 %, P < 0.05) than control LysMCre Egf-r+/+ but no difference regarding TNF-α. LPS-stimulated LysMCre Egf-rlox/lox BMDMs produced significantly more IL-6 (+70 %) and more IL-10 (+180 %) than control LysMCre Egf-r+/+ cells. In addition, LysMCre Egf-rlox/lox neutrophils stimulated by LPS produced more TNF-α (+48 %) and have a higher MPO activity (+50 %, P < 0.05) than control cells. Genetic invalidation and pharmacological inhibition of EGF-R significantly decreased the recruitment of immune cells within the peritoneal cavity following LPS injection (respectively, −65 and −42 %, P < 0.05). However, genetic invalidation or pharmacological inhibition of Egf-r did not impair mouse survival in a model of endotoxinemia.

Conclusion EGF-R deficiency/inhibition modulated innate immune response but did not impair survival during experimental sepsis.

Competing interests None.

P54 An intravenous omega-3 bolus at reperfusion time improves shock and vascular dysfunction in a myocardial ischemia–reperfusion rat model

Correspondence: Julie Boisramé-Helms -

Annals of Intensive Care 2016, 6(Suppl 1):P54

Introduction Omega-3 supplementation for several weeks has been shown to improve myocardial resistance to ischemia–reperfusion in rats. In a rat model of myocardial ischemia–reperfusion, we investigated the effect of an intravenous bolus of omega-3, associated or not with iodinated contrast media (used for reperfusion process during coronarography), at reperfusion time, so as to mimic emergency clinical management of myocardial infarction and reperfusion-induced shock.

Materials and methods A myocardial ischemia–reperfusion model was performed by left anterior descending coronary artery ligation in rats. After a 30-min ischemia, but before reperfusion, an intravenous bolus of EPA and DHA (6:1), associated or not with iodinated contrast media, was administered. After a 4-h reperfusion, circulating procoagulant microparticles were quantified phenotyped. Vascular and heart inflammation, oxidative and nitrosative stresses were assessed. Ex vivo vascular reactivity was performed with a pharmacological selective inhibitor of iNOS. Infarct size was assessed by triphenyltetrazolium chloride staining.

Results Treating rats with an EPA/DHA bolus before reperfusion significantly improved the ischemia–reperfusion syndrome, increasing mean arterial pressure (151 ± 13 vs 122 ± 17 mmHg, p < 0.05), coronary and carotid blood flow, and decreasing infarct size (39.9 ± 2.3 versus 26.8 ± 5.7 % of left ventricle, p < 0.05). Moreover, ex vivo mesenteric resistance artery sensitivity to phenylephrine was improved. Finally, endothelial CD54+ microparticle release was decreased (9.1 ± 2.5 vs. 4.8 ± 2.0 nM EqPhtdSer, p < 0.05), as well as vascular inflammation and oxidative stress, reflecting a reduced vascular dysfunction.

Discussion In our model, the omega-3 vascular beneficial effects may partly result from endothelial protection. Indeed, ischemia–reperfusion-induced endothelial dysfunction results in phenotypic and physical changes in the endothelium, with a deregulated release of potent vasodilators nitric oxide and prostacyclin, reduced vascular reactivity to vasoconstrictors, associated with leukocyte and platelet aggregation and deregulation of nitric oxide signaling. Rat hemodynamic parameters were subsequently improved. Associating omega-3 to iodinated contrast media in the reperfusion process therefore deserves further investigating and might be a promising line of development in humans.

Conclusion In this rat model, an intravenous omega-3 bolus before reperfusion significantly improved cardiovascular resistance to ischemia–reperfusion-induced cardiovascular failure and shock.

Competing interests None.

P55 Impact of urantide antagonist of urotensinergic system on myocardial function during a murine model of septic shock

Correspondence: Emmanuel Besnier -

Annals of Intensive Care 2016, 6(Suppl 1):P55

Introduction New therapies to improve prognosis of patients with septic shock are a major area of research. Urotensinergic system plays a role in both modulation of inflammation and cardiovascular system and thus could be a therapeutic target. Unpublished work previously done highlighted an improvement in inflammatory parameters by administration of the antagonist urantide in an endotoxinic model. The aim of our study was to evaluate survival and myocardial function after administration of urantide during sepsis.

Materials and methods This animal study was subdivided into three experimental process involving different mice (C57Bl/6): a survival analysis (n = 10/group), an echocardiographic analysis (n = 5/group) and a study of tissular inflammation markers (n = 5/group). Three groups of mice were made, a sham group without sepsis, a control group (LPS-NaCl) and an urantide-treated group (LPS-urantide). Intraperitoneal injection of 30 mg/kg of lipopolysaccharide (LPS) from Escherichia coli was administered for septic groups. Echocardiographic analysis was performed every 3 h during 9 h for left ventricular ejection fraction (LEVF), cardiac output, aortic flow, heart rate and the E/A ratio obtained from mitral flow. Injection of either NaCl or urantide (10−4 M) at hour 3 and 6 was administered. Analysis of survival was performed according to the same experimental process, and tissular expression of hepatic, renal and myocardial NF-κB was performed at hour 9.

Results Survival was significantly improved from 30 % up to 88.9 % by urantide. Administration of LPS induced impairment of LVEF 9 h after the onset of sepsis. Urantide induced a significant difference in the LVEF and cardiac output at hour 9 with an increase up to 46 % (p < 0.05) and 51 % (p < 0.01), respectively, versus the control group. No difference was reported for heart rate or E/A ratio. Renal and hepatic expression of NF-κB was reduced down to 60 and 65 %, respectively, by urantide administration. Myocardial expression was not performed because of technical issues.

Discussion Our results highlighted a key role of urotensinergic system in cardiac performance during septic shock and a potential beneficial effect of its antagonizing. Nevertheless, our study was not able to determine the preload status of mice. Thus, it was not possible to discriminate a beneficial effect due to vascular or inotropic modulation.

Conclusion Urantide administration induced an improvement in survival and systolic myocardial function with increase in LVEF and cardiac output in an endotoxinic model of sepsis. Improvement in tissular inflammation may be involved. Further study is necessary in order to explore the effect on myocardial performance independently from vascular function.

Competing interests None.

P56 Association of dexamethasone and antibodies to poly-β(1,6)-n-acetyl-glucosamine in prevention and treatment of neonatal bacterial meningitis: an experimental study

Correspondence: Stéphanie Pons -

Annals of Intensive Care 2016, 6(Suppl 1):P56

Introduction Bacterial meningitis is still a severe infectious disease worldwide, with high rates of morbidity and mortality in children and adults. Dexamethasone (DXM) is recommended in association with antibiotics in some specific cases to decrease the risk of sequelae and possibly the mortality rate. The poly-β-(1,6)-N-acetyl-glucosamine (PNAG) is a surface polysaccharide expressed by numerous pathogenic bacteria, including species responsible for meningitis as Streptococcus pneumoniae, Neisseria meningitidis, Haemophilus influenzae, or Escherichia coli K1. Passive immunotherapy by monoclonal (MAb) to PNAG and polyclonal antibodies (PAb) to deacetylated PNAG (dPNAG) can prevent and treat various infections. The aim of this study was to analyze the association of DXM and antibodies to PNAG in neonatal bacterial meningitis prevention and treatment, using an E. coli K1 neonatal meningitis murine model.

Materials and methods The presence of PNAG on E. coli K1 surface in vitro and in vivo was assessed by confocal microscopy using fully human immunoglobulin G1 MAb with V-regions encoding specificity for either P. aeruginosa alginate, MAb F429 (control), or PNAG, MAb F598. In vitro experiments of bacterial adhesion on blood brain barrier (BBB) human cells were conducted using hCMEC/D3 cell line. Confluent cells were infected with 3.5 × 10e8 of the well-described E. coli K1 S88, alone or with DXM, goat PAb to dPNAG or normal goat serum (NGS). Experiment in the neonatal meningitis model was performed in 3-day-old CD1 mice by intraperitoneal (IP) administration of E. coli S88 K1 (10−6/mouse). The IP administration of PBS or goat PAb to dPNAG (50 μL/mouse) was performed 24 h before and 2 h after challenge. DXM (0.7 mg/kg) was also injected 2 h after infection. The newborn mice were killed 24 h after challenge, the brain harvested and the colony-forming unit (CFU) count of E. coli determined.

Results In vitro, bacterial growth of E. coli K1 was not affected by the presence of different concentrations of DXM, antibodies to PNAG or both. PNAG expression was conserved in vitro on the bacterial surface of E. coli K1 in the presence of different concentrations of DXM, meaning that antibodies to PNAG could be associated with DXM. PNAG was also found on E. coli K1 surface in positive brains’ homogenates of newborn mice, confirming that PNAG could effectively serve as a therapeutic target in vivo. hCMEC/D3 experiments showed a 50 % decrease of bacterial adhesion in the presence of PAb to dPNAG compared to E. coli K1 alone (p = 0.001) or NGS (p = 0.007). DXM was responsible for a significant increase of E. coli adhesion in a dose-dependent manner (DXM 100 μM per well 2 × 10−7 vs 8.2 × 10−6, p = 0.001). The addition of different concentrations of DXM to PAb or to NGS did not change E. coli adhesion on hCMEC/D3 compared, respectively, to PAb alone (p = 0.98) or NGS alone (p = 0.48). In vivo, E. coli K1 meningitis rate was 100 % after IP administration. Compared with PBS, the administration of PAb decreased significantly the CFU per brain (4.1 × 10−5 vs 4 × 10−2, p < 0.001). DXM administration after bacterial challenge did not change the CFU in the brain of neonates compared with bacteria alone (1.7 × 10−5 vs 1.3 × 10−5, p = 0.43). The association of PAb to DXM showed a significant decrease in CFU in the brain compared with PBS (4.1 × 10−5 vs 10 × 10−3, p = 0.007) but no difference with PAb alone (103 vs 4 × 10−2, p = 0.33).

Conclusion PAb to dPNAG decrease E. coli K1 adhesion on the BBB in vitro and E. coli CFU per brain in vivo. DXM in association with PAb to dPNAG does not change the effect of PAb alone in vitro neither in vivo. Further experiments using the E. coli K1 neonatal model as well as adult animal models of meningitis when testing S. pneumoniae or N. meningitidis will better determine the possible use of DXM in association with PAb or MAb F 598 in bacterial meningitis treatment as well as the general and local levels of inflammation with both treatments.

Competing interests None.


1. Brouwer MC, et al. Corticosteroids for acute bacterial meningitis. Cochrane Database Syst Rev. 2013 Jun 4;6:CD004405.

2. Wang X, et al. The pgaABCD locus of Escherichia coli promotes the synthesis of a polysaccharide adhesin required for biofilm formation. J Bacteriol. 2004 May;186(9):2724–34.

P57 Hyperoxemia as a risk factor for ventilator-associated pneumonia

Correspondence: Saad Nseir -

Annals of Intensive Care 2016, 6(Suppl 1):P57

Introduction Several studies demonstrated an increased mortality rate in critically ill patients with hyperoxemia. Reactive oxygen species production, atelectasis, alveolar macrophage dysfunction, and acute lung injury are well-known consequences of hyperoxemia. In addition, these factors have been reported to increase the risk of pneumonia in intubated patients. To our knowledge, no study has previously evaluated the relationship between hyperoxemia and ventilator-associated pneumonia (VAP).

Patients and methods This retrospective observational study was performed during an 18-month period in a 30-bed medical and surgical ICU. All intubated patients requiring mechanical ventilation for more than 48 h were eligible. The primary objective was to determine the impact of arterial hyperoxemia (defined as PaO2 >120 mmHg) on the risk of VAP occurrence. VAP definition included clinical, radiological, and quantitative microbiological criteria. Patient characteristics and information on VAP occurrence were prospectively collected. Only data on arterial blood gases were collected retrospectively. Each day with at least one PaO2 >120 mmHg accounted for 24 h with hyperoxemia. Risk factors for VAP were determined using univariate analysis and logistic regression multivariate analysis.

Results VAP was diagnosed in 141 (28 %) out of the 503 included patients. The incidence rate of VAP was 14.7 per 1000 ventilator-days. Pseudomonas aeruginosa (34 %) and Staphylococcus aureus (11.3 %) were the most frequently isolated bacteria. Hyperoxemia at ICU admission (67 vs. 53 %, OR [95 % CI] 1.8 [1.2–29], p = 0.004) and number of days spent with hyperoxemia (5 vs 3 days, p < 0.001) were significantly higher in patients with VAP, compared with those without VAP. Univariate analysis also identified (p < 0.05) age, SAPS II, LOD score, shock at ICU admission, proton-pump inhibitor (PPI) use, red blood cell transfusion, sedation, and duration of mechanical ventilation as risk factors for VAP.

Multivariate analysis identified the number of days spent with hyperoxemia (OR [95 % CI] 1.1 [1.04–1.2] per day, p < 0.05), SAPS II (1.01 [1.002–1.024] per point, p < 0.05), red blood cell transfusion (1.8 [1.1–2.7], p = 0.01) and PPI use (1.9 [1.03–1.2], p < 0.05) as independent risk factors for VAP. Two other logistic regression models were performed to determine the relationship between the percentage of days spent with hyperoxemia, or hyperoxemia at ICU admission and VAP. In these secondary analyses, the percentage of days spent with hyperoxemia, and hyperoxemia at ICU admission were also independently associated with VAP occurrence.

Conclusion Hyperoxemia is an independent risk factor for VAP. Further prospective large multicentre studies are required to confirm our results.

Competing interests None.Footnote 2

P59 Significance of prior digestive colonisation with extended-spectrum beta-lactamase producing Enterobacteriaceae in patients with ventilator-associated pneumonia

Correspondence: Rémi Bruyère -

Annals of Intensive Care 2016, 6(Suppl 1):P59

Introduction Ventilator-associated pneumonia (VAP) is frequent in intensive care units (ICU). Extended-spectrum beta-lactamase-producing Enterobacteriaceae (ESBL-EB) are difficult-to-treat pathogens. We sought to assess the interest of screening for ESBL-Eb rectal carriage as a way to predict their involvement in VAP.

Patients and methods A retrospective cohort study of patients with suspected VAP in a medical ICU was conducted. Every patient admitted between January, 2006, and August, 2013, was eligible if subjected to mechanical ventilation for more than 48 h. Each patient with suspected VAP was included in the cohort. Active surveillance culture for ESBL-EB detection was routinely performed in all patients on admission and then weekly throughout the study period. ESBL colonisation was defined by the isolation of at least on ESBL-EB from rectal swab culture.

Results Among 587 patients with suspected VAP, 40 (6.8 %) were colonised with ESBL-EB prior to the development of pneumonia. Over the study period, 20 patients (3.4 %) had VAP caused by ESBL-EB, of whom 17 were previously detected as being colonised with ESBL-EB. Sensitivity and specificity of prior ESBL-EB colonisation as a predictor of ESBL-EB involvement in VAP were 85.0 and 95.7 %, respectively. The positive and negative predictive values were 41.5 and 99.4 %, respectively. The positive likelihood ratio was 19.8.

Conclusion Screening for ESBL-EB digestive colonisation by weekly active surveillance cultures could reliably exclude the risk of the involvement of such pathogens in VAP in low prevalence area.

Competing interests None.


1. Razazi K, Derde LP, Verachten M, et al. Clinical impact and risk factors for colonization with extended-spectrum beta-lactamase-producing bacteria in the intensive care unit. Intensive Care Med 2012.

2. Blot S, Depuydt P, Vogelaers D. Maximizing rates of empiric appropriate antibiotic therapy with minimized use of broad-spectrum agents: are surveillance cultures the key? Intensive Care Med 2008.

P60 Lower respiratory tract colonization in mechanically ventilated patients: Incidence, risk factors and impact on prognosis

Correspondence: Souheil Elatrous -

Annals of Intensive Care 2016, 6(Suppl 1):P60

Introduction Ventilator-associated pneumonia (VAP) complicates the course of 8–28 % of patients receiving mechanical ventilation (MV). Its mortality rate can reach 76 % when lung infection is caused by high-risk pathogens as Acinetobacter baumannii. Bacterial colonization of the trachea and bronchi preceding VAP is well recognized as part of the pathogenic mechanisms. The objectives of this study were to determine the incidence, the associated risk factors and the impact on prognosis of lower respiratory tract colonization in patients under MV.

Patients and methods It is a monocentric prospective study performed within 3 years. All patients admitted to our intensive care unit requiring invasive mechanical ventilation for more than 48 h were included. Eligible patients had a bacteriological follow-up that starts at the third day of hospitalization by realizing tracheal aspirate (TA) which is repeated every 3 days during MV length. Lower respiratory tract colonization is defined by a positive TA culture in the absence of clinical suspicion of VAP.

Results Two hundred and sixty-six patients on MV for more than 48 h were prospectively followed in this study. The airway colonization occurrence was 38 % with a maximum between day 3 and day 6 of MV. The two most frequently isolated germs are Acinetobacter baumannii and Pseudomonas aeruginosa with a significant increase in their occurrence in time.

In multivariate analysis, the use of proton pump inhibitors, MV’s duration and length of stay are identified as associated factors to the onset of airway colonization.

The incidence of VAP was significantly higher in colonized versus non-colonized patients (32 vs 18 %; p = 0.006). The onset of airway colonization was significantly associated with prolonged duration of MV and an increased length of stay [19 ± 17 vs 9 ± 7 days (p < 0.0001) and 25 ± 23 vs 12 ± 9 days (p < 0.0001), respectively)] The mortality rate was comparable between colonized and non-colonized patients.

Conclusion The lower respiratory tract colonization is frequent in patients under MV. The use of proton pump inhibitors, MV’s duration and length of stay are identified as associated factors to the onset of airway colonization. Colonized patients had a greater risk of developing VAP. The onset of airway colonization was significantly associated with prolonged duration of MV and an increased duration of stay.

Competing interests None.

P61 Epidemiology of ventilator-associated pneumonia in the ICU among patients already hospitalized: early- versus late-onset pneumonia

Correspondence: Gaël Piton -

Annals of Intensive Care 2016, 6(Suppl 1):P61

Introduction Initiation of antibiotherapy is of major importance in the treatment of nosocomial pneumonia in the ICU. Antibiotherapy must be early, appropriated to the presumed bacteria, and must take into account epidemiology of the ICU and of the hospital. The aim of this study was to describe epidemiology of ventilator-associated pneumonia (VAP) in the ICU for patients already in hospital before ICU admission (HCAP) and to identify risk factors for multidrug-resistant pathogens.

Patients and methods This was a monocentric retrospective study of patients admitted to a single ICU in a large regional university hospital, with a diagnosis of ventilator-associated pneumonia, originating from another unit in the same hospital. Patients coming from home and from another hospital, and immunocompromised patients were excluded. Ventilator-associated pneumonias were divided into early onset (<5 days) and late onset (≥5 days). Duration of hospitalization before ICU admission, and the main risk factors for multidrug-resistant pathogens were collected.

Results Between January 2014 and May 2015, 114 patients developed ventilator-acquired pneumonia in the ICU. There were 69 (61 %) early-onset pneumonia and 45 (39 %) late-onset pneumonia. Bacterial diagnoses were made on tracheal aspirations (99.1 %) and broncho alveolar lavages (35.9). There were no significant differences in the bacteriology of early- and late-onset VAP. Gram negative bacteria were predominant (30.4 vs 31.1 %, p = 0.94). There were 18.4 % of multidrug-resistant pathogens in the overall population. Multidrug-resistant pathogens were mainly gram-negative bacteria (four BLSE, 11 cephalosporinase, four multiresistant pseudomonas aeruginosa, one stenotrophomonas maltophilia, and 1 SARM). There was a trend toward a higher prevalence of multidrug-resistant pathogen among patients with late-onset VAP (13 vs 27 %, p = 0.07). In univariate analysis, multidrug-resistant pathogens were associated with duration of hospitalization before ICU admission (5.9 ± 8.5 vs 24.8 ± 74.5, p < 0.05), duration stay in the ICU (8.8 ± 6.8 vs 27.2 ± 25.9, p = 0.004), and duration of mechanical ventilation (7.1 ± 5.9 vs 20.6 ± 25.3, p = 0.04).

Conclusion This work did not find significant differences in the epidemiology of early-onset versus late-onset VAP for patients already in hospital before ICU admission. However, there was a trend toward a higher rate of multidrug-resistant pathogens in late-onset VAP. Multidrug-resistant pathogens were mainly gram-negative bacteria, whereas prevalence of SARM was anecdotal. Duration of hospital stay before ICU admission, duration of ICU-stay, and duration of mechanical ventilation were risk factors for multidrug-resistant pathogens.

Competing interests None.

P62 Systematic review of inclusion/exclusion criteria, judgment criteria and statistical hypothesis in randomized controlled trials assessing the efficacy of antimicrobials for hospital-acquired and ventilator-associated pneumonia treatment

Correspondence: Emmanuel Weiss -

Annals of Intensive Care 2016, 6(Suppl 1):P62

Introduction Hospital-acquired pneumonia (HAP) is the second most common nosocomial infection in hospitalized adults and is a leading cause of death among critical infections. To face the dramatically increasing prevalence of multidrug-resistant pathogen-related HAP, new therapeutic options are currently in development. The evaluation of these new treatments will require properly designed studies with appropriate inclusion criteria and endpoints. During the past 10 years, the regulatory agencies repeatedly updated guidance for the design of randomized controlled trials (RCT) aiming to evaluate the efficacy of antimicrobials for treatment of hospital-acquired and ventilator-associated pneumonia (HAP/VAP). However, our hypothesis is that, to date, HAP/VAP RCTs greatly differ in population enrolled, in HAP/VAP definition criteria used and in primary and secondary outcomes assessed. Such differences between studies may be of importance because they may impact results of studies. Thus, we performed a systematic review of characteristics of enrolled populations, inclusion/exclusion criteria and endpoints in RCT addressing the efficacy of antimicrobials for HAP/VAP treatment.

Patients and methods A double data search (PubMed–EMBASE, ICAAC, ESCMID) was performed of all RCT of the past 20 years evaluating antimicrobials for treatment of intensive care unit (ICU)-HAP/VAP.

Results The search retrieved 997 abstracts, of which 93 studies were reviewed. After exclusion of duplicates, post hoc analyses, studies without precise inclusion/exclusion criteria, studies without VAP and studies performed only outside the ICU, 25 studies were analyzed. One-third of studies included only VAP, while the other two-third included both HAP and VAP. The definition of HAP/VAP used various clinical radiological and biological criteria. Variable mandatory duration of prior hospital stay was used and length of mechanical ventilation was included in inclusion criteria in 36 % of VAP studies only. Microbiological criteria were required in only 48 % (with a microbiological threshold in 36 % of studies). Severe patients (APACHE score >25) were excluded in 24 % of studies. Most studies tested a hypothesis of non-inferiority (36 %) or equivalence (24 %), while 16 % of studies aimed to demonstrate superiority. Surprisingly, statistical hypothesis was not clearly defined and sample size calculation not mentioned in 28 and 64 % of studies, respectively. Clinical cure was considered as a primary endpoint in 92 % of studies. However, the primary endpoint was dual and even triple in, respectively, 40 and 12 % of studies because it was studied in different analysis population (28 % of studies) or it was associated with microbiological cure (16 %) and/or safety. Assumed clinical cure rate was anticipated in only one-third of manuscripts. Mortality was used as a primary endpoint only in two studies and safety in one. Definitions of clinical cure consisted in complete or partial remission of clinical signs and symptoms associated with pneumonia in 100 % of studies, but improvement in chest radiography was associated in only two-third of the studies. Interestingly, the lack of additional antibiotic requirement during follow-up was included in clinical cure definition in 29 % of studies. The test-of-cure visit evaluating the clinical cure ranged from the end of treatment to 28 days later.

Conclusion Our study provides a description of populations and endpoints of RCTs evaluating antimicrobials for treatment of ICU-HAP/VAP. We show a significant heterogeneity in enrollment criteria, endpoints and statistical design that may influence the ability of studies to demonstrate differences between studied drugs.

COMBACTE is supported by IMI/EU and EFPIA.

Competing interests None.

P63 Gram stain identification resulting from blindly inserted telescoping plugged catheter material. Does it contribute to diagnosis and treatment of ventilator-associated pneumonia?

Correspondence: Fabrice Cook -

Annals of Intensive Care 2016, 6(Suppl 1):P63

Introduction Ventilator-associated pneumonia (VAP) is a frequent infection that complicates the course of many critical care patients. Early antimicrobial therapy, initiated before the results of pulmonary specimen cultures, seems to improve outcome (1). However, empiric treatment strategies expose the risk of inappropriate treatment, or unnecessary broad-spectrum antibiotic. Best bacteriological technique for the diagnosis of VAP is still debated. Blindly inserted telescoping plugged catheter (TPC) sampling seems reliable and is recommended methods for the bacteriological diagnosis of VAP. The threshold of 103 colony-forming units (cfu) mL−1 provides a good sensitivity and specificity. Because 24- to 48-h delay is needed for quantitative cultures, early Gram’s stain identification performed upon TCP samples could guide empirical antibiotic therapy. However, accuracy of such strategy is controversial because true sensitivity and specificity of Gram’s stain for diagnosis of VAP remain questionable. In the present study, we evaluate whether the Gram stain upon initial TPC samples could predict the results of quantitative culture and identification of pathogens in patients suspected for VAP. We analyze associated factors that may influence accuracy of Gram’s stain identification and the resulting empiric antibiotic treatment adequacy.

Patients and methods We retrospectively analyzed 122 TPC samples from 77 patients between January and December 2012, corresponding to all TPC samples retrieved for VAP diagnosis in a 20-bed surgical intensive care unit of a University Hospital. Positive Gram’s stains and culture for any microorganisms, the presence of Gram-positive cocci (GPc) and/or Gram-negative rods (GNr) on Gram’s stain and culture were collected. Cultures were considered positive if they yielded organisms on final culture with a threshold value ≥103 cfu mL−1. The presence of leukocytes, moment of the day sampling being perform, prior to sampling antibiotic therapies, and duration of mechanical ventilation before sampling were analyzed as possible associated factors that may influence results of Gram’s stain. Fisher’s exact test was used for comparison of sensitivity and specificity between groups. Data are expressed as number (95 % confidence interval); p < 0.05 was considered to be statistically significant.

Results A total of 122 TPC samples were analyzed. Details of diagnosis performance (sensitivity, specificity, predictive value and likelihood ratio) of Gram staining for positive quantitative culture of any microorganism, GPc and GNr are summarized in Table 14.

Table 14 See text for description

Among associated factors, the presence of leukocytes on Gram’s stain was associated with an improvement in sensitivity versus the absence of leukocytes for any microorganism (0.87, p < 0.001), for GPc (0.83, p = 0.04) and for GNr (0.63, p = 0.036). Other associated factors failed to influence the results of Gram’s stains. False negative Gram’s stain for any microorganism was significantly associated with non-adapted empirical antibiotic treatment (p = 0.008).

Conclusion Gram’s stain identification upon TPC samples has acceptable accuracy for predicting positive culture for the diagnosis of VAP. Prediction of GPc or GNr bacteria involved in VAP has poor accuracy excepted for GNr specificity. The presence of leukocytes on TPC results of Gram’s stain improves sensitivity for VAP-positive culture. Taking Gram’s stain information into account for establishing empiric antibiotic treatment may generate non adapted therapy. Results must be confirmed by prospective and multicenter studies.

Competing interests None.


1. Irequi M, Ward S, Sherman G, Fraser VJ, Kollef MH. Clinical importance of delays in the initiation of appropriate antibiotic treatment for ventilator-associated pneumonia. Chest. 2002 Jul;122(1):262–268.

P64 Colistin in nosocomial pneumonia in intensive care

Correspondence: Hanane Ezzouine -

Annals of Intensive Care 2016, 6(Suppl 1):P64

Introduction The increase in the incidence of ventilator-acquired pneumonia (VAP) Acinetobacter baumannii to multiresistant (ABMR) is responsible for a high morbidity and mortality in intensive care environment worldwide. The gold standard treatment of these serious infections remains carbapenems, including imipenem and meropenem. Sulbactam is a validated therapeutic alternative. Few studies have proved the effectiveness of VAP colimycin onto ABMR.

Patients and methods This is a prospective observational study in the medical ICU of CHU IBN Rushd of Casablanca for 24 months (October 2012–October 2014). Inclusion criteria were patients hospitalized in intensive care and who developed nosocomial pneumonia after 48 h of mechanical ventilation, with isolation of Acinetobacter Baumanii only sensitive to colimycin, using the protected distal sampling. After the confirmation of the presence of a susceptibility to ABMR, antibiotic therapy was started.

Results In our study, 120 cases of ventilator-acquired pneumonia were collected. The average age of the patients was 40 ± 15 years, and the sex ratio was 1.5 with a male predominance. The reason for hospitalization in the ICU was dominated by diabetics ketoacidosis and serious head injuries. The simple gravity index (SAPS II) of the patients was 35 ± 12. The mean time to onset of VAP was 6 ± 1 days. Treatment with intravenous colimycine was considered effective in 114 patients (95 % of cases). The overall mortality rate was 25 % (30 cas) in our series, and only six cases (5 %) were attributable to VAP AB complicated multi-resistant refractory septic shock. There were no patients with impaired kidney function or clinical neurological impairment.

Conclusion Our study suggests that colimycin could be a first-line therapy for VAP in ABMR in our Third World countries due to its efficacy, safety and low cost.

Competing interests None.

P65 Bedside tissue perfusion evaluation predicts intradialytic hemodynamic instability in critically ill patients

Correspondence: Naïke Bigé -

Annals of Intensive Care 2016, 6(Suppl 1):P65

Introduction Intermittent hemodialysis is a key support therapy in ICU. Despite protocol-based optimization, arterial hypotension during intermittent hemodialysis remains a frequent issue ranging from 10 to 60 % in critically ill patients. Our objective was to test whether tissue perfusion parameters assessed at the bedside (mottling, index capillary refill time (CRT), and lactate) predict intradialytic hemodynamic instability (HI).

Patients and methods We conducted a prospective observational study in a 18-bed medical ICU in a tertiary university hospital including hemodialysis sessions performed for acute kidney injury. Exclusion criteria were the following: patients with dark skin and dialysis performed in extreme emergency. Mean arterial pressure (MAP), cardiac index, mottling, index CRT and lactate level were recorded just before starting hemodialysis. HI requiring fluid resuscitation or vasopressors introduction/increase was recorded 60, 120, and 240 min after hemodialysis starting.

Results One hundred and six hemodialysis sessions performed in 45 patients were recorded. Patients received vasopressors in 43 (41 %) sessions. HI occurred in 22 (21 %) of all the sessions and was more frequent among patients receiving vasopressors (42 vs 6 %, P < 0.0001). Mottling was more frequent (77 vs 40 %, P < 0.0001), index CRT was higher (3.3 [1.6–4.6] vs 1.1 [0.8–1.9] s, P < 0.0001), and lactate level was higher (2.8 [1.4–6.9] vs 1.1 [0.8–1.5] mmol/L, P < 0.0001) before sessions with HI in comparison with sessions without. The incidence of HI increases with the number of tissue perfusion alterations (2, 14, 31, and 65 % for 0, 1, 2, and 3 alterations, respectively, P < 0.0001), independently of MAP (P < 0.0001). A tissue perfusion score, defined as “index CRT(s) + lactate level (mmol/L) + 0/1 if mottling were absent/present”, was predictive of HI independently of MAP (OR 1.18 [1.05–1.32], P < 0.0001). According to ROC curves analysis, a threshold of 4.2 predicted HI with a sensitivity of 82 [60–95] % and a specificity of 82 [72–90] %.

Conclusion The incidence of HI increases with the number of tissue perfusion alterations independently of MAP. At the bedside, a combined tissue perfusion score including mottling, index CRT, and lactate level helps to identify patients at risk.

Competing interests None.

P66 Evaluation of electrolytes and acid–base disorders in patients undergoing citrate-based anticoagulation for continuous renal replacement therapy in intensive care unit

Correspondence: Benjamin Delmas -

Annals of Intensive Care 2016, 6(Suppl 1):P66

Introduction Specific electrolytes and acid–base disorders may occur using citrate-based anticoagulation (CBA) for continuous renal replacement therapy (CRRT). The aim of this study is to evaluate and to compare the impact of two CBA-CRRT protocols on electrolytes and acid–base status: one continuous veno-venous hemodialysis (CVVHD) protocol using 4 % citrate (group A) and one continuous veno-venous hemofiltration (CVVH) protocol using 18/0 citrate (group B), both of which use phosphate-containing solutions.

Patients and methods We conducted a retrospective study from January 2013 to March 2014. All patients treated by CBA-CRRT admitted in three medico-surgical ICUs have been included. We collected demographic, clinical and biological data and performed Wilcoxon’s matched-pairs signed-ranks test to compare for quantitative values. Data are expressed as median, first and third quartile.

Results Forty-seven patients were included, 30 in group A and 17 in group B. No severe metabolic alkalosis occurred. One severe metabolic acidosis occurred in group A as a complication of citrate accumulation syndrome (pH 6.99, base excess −19 mmol/L and bicarbonate 12 mmol/L), diagnosed as a total calcium over ionized calcium ratio of 2.54. This was also the only case of severe hypocalcemia (ionized calcium 0.89 mmol/L). Seven patients (15 %) presented mild hypophosphoremia, but no severe hypophosphoremia occurred. We did not notice any hyperphosphoremia in our patients. Median pH was 7.41 (7.38–7.46) on day 3 and 7.34 (7.28–7.39) on day 1 (p < 0001). Median natremia on day 3 was 141 (139–141) and 135 (134–135) mmol/L in group A and B, respectively (p = 0.002). Median chloremia was 108 (106–109) and 95 (95–96) mmol/L in group A and B, respectively (p = 0.0002). In group A, one severe hypernatremia occurred on day 1, corrected on day 2, and two moderate hypernatremia occurred on day 1, corrected on day 2. No hypernatremia occurred in group B (p = 0.29). Hypomagnesemia occurrence constantly increased from day 1 to day 3, from 0 to 71 %.

Discussion This study pinpoints three interesting findings. First, it is well known that hypernatremic metabolic alkalosis may occur under CBA-CRRT in case of using trisodium citrate, which contains a high sodium load (408 mmol/L) as compared to 18/0 citrate (140 mmol/L of sodium), but the mild risk of alkalosis due to hypochloremia using dilute citrate (86 mmol/L in 18/0) is mostly unrecognized. However, no patient presented severe metabolic alkalosis, but natremia was significantly higher in group A and chloremia lower in group B. Second, hypophosphoremia rate, which has been diagnosed in as many as 65 % of the patients undergoing CRRT when using dialysis or replacement solutions without phosphate, is effectively decreased by using phosphate containing solutions. Finally, attention must be drawn on the high proportion of hypomagnesaemia, even if dialysis and replacement solutions contain magnesium, as citrate creates complex with magnesium as well as calcium.

Conclusion Both protocols are effective for correcting acid–base disorders, but in different ways, one by increasing natremia and the other by reducing chloremia. Phosphate-containing solutes are effective to reduce hypophosphoremia occurrence, but not sufficient to avoid hypomagnesaemia.

Competing interests None.

P67 Continuous venovenous hemofiltration in critically ill patients with acute kidney injury: a cost study impact of replacement fluids consumptions

Correspondence: Elodie Jean-Bart -

Annals of Intensive Care 2016, 6(Suppl 1):P67

Introduction For patients with acute kidney injury (AKI), in our intensive care unit (ICU), continuous venovenous hemofiltration (CVVH) with citrate has been implemented since 2013. This has been conducted to change replacement fluids (RF), needing inside addition of ionic (phosphore and/or calcium), and had a cost impact.

The aim of the study was to assess the cost impact of citrate CVVH versus non-citrate CVVH and to estimate the cost impact of different RF use.

Patients and methods We performed a retrospective study about all ICU patients in AKI and requiring CVVH in 2014. Data collected were about: patients characteristics (age, sex ratio, BMI, IGS2); CVVH (indications, effective duration, filters, RF, calcium and phosphore consumption). For citrate CVVH, RF used were citrate anticoagulation solution and bicarbonate-buffered solution and for heparin or no anticoagulation, only bicarbonate-buffered solution. To estimate the economic impact of other RF, in citrate group, cost of restitution solution with integrated phosphore was used. In non-citrate group, cost of bicarbonate-buffered solution with calcium integrated was used. Results about costs were expressed for 24 h of effective CVVH. Differences between citrate and non-citrate group were assessed with Student’s test with p < 0.05. The cost impact simulation of different RF was assessed with appariate Student’s test with p < 0.05.

Results In 2014, 66 patients received CVVH in ICU. We included 64 patients (two excluded because of missing data). They had a mean age of 68.1 ± 16.6 years old, a mean IGS2 of 58.2 ± 20.5, a mean stay in ICU 9.0 ± 9.6 days and a mortality rate of 28.1 %. There was no difference between citrate and non-citrate group. Mean effective duration of CVVH was 52.1 h. Citrate anticoagulation was used for 40.0 % of them. About hemofiltration, there was no statistical difference between mean cost/24 h in citrate and non-citrate group. Cost simulations with bicarbonate-buffered solution with integrated ionic were significantly less expensive with a mean economy of 48.3€/24 h, so an annual total economy of 5726.3€ (Table 15).

Table 15 CVVH characteristics and cost impact simulations

Conclusion This study highlighted an interesting assessment of CVVH practices. There was no statistically difference between citrate and non-citrate group about the mean cost/24 h. But cost simulations with ionic integrated RF seem to be less expensive, especially as these results are not taking into account human costs. Most CVVH was shorter than 24 h, and reflection about the intermittent hemofiltration could be needed. Simulations showed that 5726.3 € could be spare. With this methodology, only evaluation of cost impact of fluids and materials consumptions in ICU is able to help us to identify where some interesting economies could be made.

Competing interests None.

P68 Acute kidney injury in patients receiving liposomal amphotericin B in an intensive care unit: register of 104 patients from 2008 to 2014

Correspondence: Marion Venot -

Annals of Intensive Care 2016, 6(Suppl 1):P68

Introduction Fungal infections represent an increasing and challenging issue in intensive care unit patients. Antifungal drugs have considerably evolved, with new molecules, whose choice depends on spectrum and adverse effects. Amphotericin B is an old drug, with a wide spectrum and highly efficient, but associated with renal toxicity. Liposomal amphotericin B is supposed to be less nephrotoxic, but other molecules may be preferred in case of renal dysfunction. We aimed to assess the impact of liposomal amphotericin B administration on renal function in critically ill patients.

Patients and methods A retrospective single-center cohort study was performed, including consecutive patients admitted to our intensive care unit between 31/01/2008 and 12/07/2014, and receiving liposomal amphotericin B for at least 2 days. Acute kidney injury was defined using the KDIGO criteria. Patients were divided into four groups: those who did not develop any acute kidney injury, those who developed acute kidney injury 1 day or more after liposomal amphotericin B initiation (de novo), those who had acute kidney injury at initiation and worsened their renal function 1 day or more after initiation and those who had acute kidney injury at initiation and did not worsen their acute kidney injury. Patients with worsening renal function after liposomal amphotericin B (de novo or on previous acute kidney injury) were compared to patients without worsening renal function (no acute kidney injury or not worsening previous acute kidney injury). The analysis of factors associated with liposomal amphotericin B-related acute kidney injury was performed in a competing risks framework (death and discharge without acute kidney injury considered as competing events).

Results One hundred and four patients were included. Median age was 51 [interquartile ranges 31; 58], 60 (58 %) were male, and median SAPS 2 was 50 [38; 68]. Main comorbidities were: hematologic malignancy in 88 (85 %), cardiovascular diseases in 12 (12 %), diabetes mellitus in seven (7 %) and chronic kidney disease in eight (8 %). During the week before intensive care unit admission, 36 (35 %) received contrast agents, 37 (36 %) vancomycin and 35 (34 %) aminoglycosides. At admission, 15 (14 %) were already treated by liposomal amphotericin B. Reasons for intensive care unit admission were hemodynamic failure in 48 (46 %), respiratory distress in 40 (38 %), neurologic disorder in seven (7 %) and close monitoring in nine (9 %). During intensive care unit stay, 71 patients (68 %) underwent invasive mechanical ventilation, 65 (62 %) needed vasopressors, 29 (28 %) received contrast agents, 68 (65 %) vancomycin and 77 (74 %) aminoglycosides. Intensive care unit, hospital and 3-month survivals were, respectively, 56.5, 52.5 and 35.5 %. Patients received liposomal amphotericin B during 14 days (from 2 to 64 days). 29 (28 %) patients worsened their renal function after liposomal amphotericin B initiation: 19 (18 %) de novo and 10 (10 %) on previous acute kidney injury. There was no renal degradation in the other 75 (72 %) patients: no acute kidney injury in 40 (38 %) and not worsening acute kidney injury in 35 (34 %). The only factor associated with the development of liposomal amphotericin B-related acute kidney injury was initiation of liposomal amphotericin B before intensive care unit admission (hazard ratio 3.75, 95 % confidence interval [1.70–8.26], p value = 0.001). Among the seven patients who worsened their renal function with liposomal amphotericin B and were alive 3 months after intensive care unit discharge, six (86 %) recovered their previous renal function. Among the 10 patients alive at month 3 who had an acute kidney injury at liposomal amphotericin B initiation and did not worsen their renal function, nine (90 %) recovered their previous renal function.

Discussion While liposomal amphotericin B has been reported to be less nephrotoxic than conventional amphotericin B, intensivists are often reluctant to use this drug in critically ill patients because they are at high risk of acute kidney injury. In our experience, liposomal amphotericin B can be used in critically ill patients with a good safety profile. Less than 30 % of the patients experienced acute kidney injury related to the drug and most of them recovered. Renal function should be carefully monitored in patients admitted to the intensive care unit and already treated with liposomal amphotericin B.

Conclusion In this retrospective single-center study, more than 70 % of the patients treated with liposomal amphotericin B did not worsen their renal function. In case of worsening, about 90 % recovered their previous renal function.

Competing interests None.

P69 Regional citrate-based anticoagulation for dummies in continuous venovenous hemofiltration: does it work?

Correspondence: Julien Jabot -

Annals of Intensive Care 2016, 6(Suppl 1):P69

Introduction Regional citrate-based anticoagulation (RCA) has proven effective to increase the filter lifespan during continuous venovenous hemofiltration (CVVH) [1] whenever performed by experienced practitioners in highly specialized teams. Moreover, from the perspective of beginners, such technique may seem to require more nursing time give to the numerous interventions to modify citrate doses and/or calcium intake. In our 23-bed ICU (University Teaching Hospital, Saint-Denis, Reunion island, France), at the decision to initiate RCA for CVVH, we had no medical expertise in dealing with the protocol, technical issues and training nurses. Therefore, we looked at reaching the same results as an experienced team with regard to the filter lifespan and the nursing workload.

Patients and methods A homemade protocol primarily tested on five patients was used in all patients meeting the criteria for CVVH with no contraindication for RCA. This protocol contained a target-oriented algorithm for post-filter ionized calcium concentration (between 0.15 and 0.4 mmoL L−1) and arterial ionized calcium concentration (between 1.00 and 1.15 mmoL L−1). Primary endpoints were filter lifespan and number of handlings for RCA therapy’s adaptation (changes in citrate anticoagulation and calcium compensation). Secondary endpoints were safety, changes in biological parameters and reasons to terminate the technique.

Results Thirty patients (20 men and 10 women, age 57 ± 12 years, SAPS2 53 ± 11) were included from November 2013 to April 2014. The protocol was set to deliver a renal replacement therapy dose of 30 mL kg−1 h−1. The median filter lifespan was 51 [37–65] h, and the median number of handlings was 2 [1–4]. Only six patients had to receive low dose of heparin in order to prevent thromboembolism. The 24 other patients received no anticoagulation. Reasons for terminating the therapy were 13 filter clottings, 12 interruption for CT scan, operating room or discharge, two catheters dysfunctions, two technical problems and one elective filter change (>72 h). We did not record any complication. Electrolytes and acid–base disturbances were uncommon and transient in patients treated with RCA: median initial pH was 7.39 [7.33–7.41], 7.44 [7.42–7.47] 24 h later and 7.40 [7.38–7.45] 48 h later; median initial serum magnesium level was 0.85 [0.74–0.94] mmol L−1, 0.69 [0.62–0.77] mmol L−1 at 24 h and 0.68 [0.60–0.81] mmol L−1 at 48 h.

Discussion Our “RCA for dummies” protocol fulfilled its objectives far beyond our expectations. The median filter lifespan (51 h) was surprisingly as high as that of reported by experts (49 h) [1]. The number of RCA therapy’s adaptation was low (one intervention every 20 h) with no significant nursing overwork.

These first results are promising ones for a non-expert team, and we wonder whether we were not too cautious than needed with this new technique. As RCA is a routine technique on our hands, we are conducting a second survey in order to rule out a “starter effect.”

Conclusion We confirm that RCA is both efficient and safe since its early use in ICUs by inexperienced practitioners, with fairly acceptable achievements in terms of filter lifespan, number of handlings and nursing overwork.

Competing interests None.


1. Efficacy and safety of citrate-based anticoagulation compared to heparin in patients with acute kidney injury requiring continuous renal replacement therapy: a randomized controlled trial. Stucker F, Ponte B, Tataw J, Martin PY, Wozniak H, Pugin J, Saudan P. Crit Care. 2015 Mar 18;19:91. doi: 10.1186/s13054-015-0822-z .

P70 Incidence of and risk factors for acute kidney injury in early postoperative liver transplantation according to AKIN classification

Correspondence: Jean-Charles Cartier -

Annals of Intensive Care 2016, 6(Suppl 1):P70

Introduction Liver transplantation (LT) is the reference treatment for hepatocellular terminal chronic diseases. Early postoperative acute kidney injury (AKI), which incidence varies between 29 and 60 % depending on the definitions and cohorts, is a factor of poor prognosis, especially when renal replacement therapy (RRT) is necessary. The objective of this study is to assess the incidence of early AKI (in the first 48 h) of a LT considering AKIN classification with consideration of diuresis and to identify associated risk factors.

Patients and methods Patients who received a first non-emergency LT orthotopic cadaveric graft between January 2011 and December 2014 at Grenoble University Hospital were included in this retrospective study. Eleven patients were excluded because of the need for RRT before the LT or related kidney transplant. Demographic and clinical and biological parameters pre-, intra- and postoperative, including comorbidities, etiology and staging of liver disease, anesthetic management, graft function, use of renal replacement therapy, diuresis by 6 h and evolution, were collected. Serum creatinine at 24 h was considered to reflect kidney function between H0 and H24 and 1–48 h to reflect that between H24 and H48. Logistic regression was used for multivariate analysis (alpha risk 0.05).

Results Of the 155 patients in the study, 129 (83.2 %) were male, median MELD and Child–Pugh scores were, respectively, 14.5 (11–23) and 9 (7–11). Seventy-three percent of patents had alcoholic cirrhosis, and median creatinine clearance (according aMDRD) was 95 ml/min with 26 patients (16.8 %) below 60 ml/min. 128 (82.6 %) had an AKI according to the AKIN criteria, and 31 (20 %) required at least one RRT in ICU. The inclusion of diuresis in the definition of the AKI increases the number of patients with AKI in 32 (20 %) with peak incidence between H24 and H30. During first 48 h, incidence of IRA varies from 54.4 to 78.1 % depending on the time slot. The median plasma volume was 3 liters of crystalloid intraoperatively and 4 liters during the first 24 h without significant differences by AKI as albumin dose. Patients who develop early AKI were more transfused in labile blood products. Incidence of nephrotoxics as iodine injection and aminoglycosides during first 48 h were, respectively, 20.6 and 6.5 % without significant difference. Only the IGS2 and monitoring by Swan–Ganz are significant independent predictive factors for AKI. Diabetics, admission norepinephrine concentration and high first residual tacrolimus dosage are significant independent predictive factors for RRT (respective OR of 3.75 (1.15 to −12.12), 2.04 (1.13 to −3.69) and 1.07 (1.05 to −1.09) per point of tacrolimus). No patient without early AKI according to these criteria has used an RRT during the ICU stay.

Conclusion The incidence of early post-LT AKI with the AKIN criteria is high. The use of AKIN classification shows dynamic nature of AKI in this context. These criteria appear prognostic for the use of renal replacement therapy in the ICU. Diabetic patients and high norepinephrine dose in immediate postoperative time seem to be related to AKI requiring RRT and should have special attention. These results need to be validated in prospective studies to clarify the impact of graft quality in the complex determinism of post-transplant AKI.

Competing interests None.

P71 Metabolic acidosis uncompensated by kidney in septic patients: a predictive factor for acute kidney injury?

Correspondence: Aline Pourcelet -

Annals of Intensive Care 2016, 6(Suppl 1):P71

Introduction Metabolic acidosis is a frequent acid–base disturbance observed in septic patients. Normally, the physiological adaptive response of the kidney is an increased urinary excretion of NH4Cl inducing a negative urinary anion gap (UAG = [Na+]u + [K+]u − [Cl−]u).

We investigated, in septic patients, the evolution of the urinary anion gap and the correlation with acute kidney injury according to kidney disease improving global outcomes (KIDGO) criteria.

Materials and methods This prospective study included patients with severe sepsis or septic shock admitted in a 24-bed medico-surgical intensive care unit of the CHU-Charleroi Marie Curie (Belgium). Blood and urinary chemistry were performed every 12 h after the admission during a maximum of 96th hour of intensive care unit length of stay.

Metabolic acidosis was defined according to Stewart’s method (SIDa <39 mmol/L) over a period of 24 h. Acute kidney injury was defined according kidney disease improving global outcomes (increased creatinine concentration ≥0.3 mg/dL unless 48 h or decreased diuresis <0.5 mL/kg/h on 6–12 h). We also defined renal tubular acidosis by a serum anion gap <16 mmol/L and urinary anion gap >0 mmol/L.

Data were presented in median values with [25th–75th] percentiles and compared by the Mann–Whitney test. A value of p < 0.05 was considered as significant.

Results Among the 40 patients included, 23 (58 %) have a metabolic acidosis at admission and 20 (50 %) developed an acute kidney injury. At the intensive care unit admission, no significant differences were observed between patients with and without metabolic acidosis according to the APACHE 2 score, creatinine (1.12 [0.78–1.85] vs 1.0 [0.62–2.04] mg/dL) and lactate concentrations (1.6 [1.0–4.2] vs 2.2 [1.5–3.2] mmol/L), quantity and the type of liquid received within 24 h (4800 [3438–6528] vs 4475 [2610–4830] mL).

On the 23 patients with metabolic acidosis, 17 (74 %) presented a renal tubular acidosis. All, except one, did not have kidney adaptation to metabolic acidosis. This inadequate response persisted during the 96 h of this study despite normalization of the acid–base status. Despite inadequate kidney response to metabolic acidosis, there was no difference in the onset of acute kidney injury between the two groups (10 acute kidney injury—group 1 in majority—in each group).

Conclusion Inadequate kidney response to metabolic acidosis was frequent in septic patients and persists despite normalization of the acid base status. Nevertheless, this tubular dysfunction does not correlate with the onset of acute kidney injury.

Competing interests None.


1. De Mendonça A, Vincent JL, Suter PM, Moreno R, Dearden NM, Antonelli M, et al. Acute renal failure in the ICU: risk factors and outcome evaluated by the SOFA score. Intensive Care Med. 2000;26(7):915–21.

2. Kellum J a. Determinants of blood pH in health and disease. Crit Care. 2000;4(1):6–14.

P72 Blood volume monitoring is useful to predict intradialytic hypotension during intermittent hemodialysis

Correspondence: Louis De Laforcade -

Annals of Intensive Care 2016, 6(Suppl 1):P72

Introduction During intermittent hemodialysis (HD), intradialytic hypotension (IDH) is a frequent complication, associated with worse outcomes. Blood volume monitoring (BV) by dialysis generator has shown its interest in chronic dialysis to predict IDH due to hypovolemia, but its utility is unclear in intensive care units.

Patients and methods We conducted a monocentric, prospective, observational study. The primary outcome was to assess whether IDH (defined by mean arterial pressure lower than 65 mmHg) was associated with a decrease in the ratio between BV and ultrafiltration (UF). Secondary outcome was to assess the association between dialysis parameters and IDH. Every dialysis session in our unit has been screened for analysis. When BV assessment was not reliable or when no UF was realized, the session was excluded of the analysis. Dialysis generators, paramedical monitoring and dialysis consumables were not modified compared with usual standard of care in our unit. Dialysis settings were left to the choice of the medical team.

Results A total of 204 sessions were screened, and 112 (corresponding to 42 different patients) were included. Median dialysate sodium was 140, and median dialysate temperature was 36 °C. Under pressor amines, 18.8 % of the sessions were made. IDH occurred in 40 % of the sessions.

BV/UF ratio was lower in session with IDH (−7.7 vs −10.5 %/L, p = 0.027). ROC curve showed an AUC of 0.624, but a cutoff value of −8.9 %/L was associated with a sensibility of 80 % to predict IDH.

Others parameters associated with IDH in univariate analysis were a higher natremia, a higher bicarbonate dialysate, a lower systolic, diastolic or mean arterial pressure at the beginning of the session, the use of pressor amines and sedation and a low difference between sodium dialysate and natremia. In multivariate analysis, the use of pressor amines and a low mean arterial pressure were associated with IDH.


Our study is the first to show an association with BV/UF and IDH for adults admitted in intensive care units. Good sensitivity of this test may be useful to detect IDH.

BV is widely used since 20 years in chronic hemodialysis and is now recommended by international guidelines. In intensive care unit, two studies have shown no interest in adults, and another did not find a decrease in IDH using a BV feedback protocol. In children, BV using was associated with a higher total UF without increasing IDH.

Other guidelines has been proposed in order to reduce IDH: high sodium dialysate concentration (145 mmol/l), or moderately cool dialysate temperature (0.5 °C beyond body temperature). Our study did not find any association between these settings and IDH.

Conclusion Decrease in BV/UF ratio is associated with IDH in our study. A 8.9 %/L threshold may predict IDH with a sensitivity of 80 % (Fig. 29).

Fig. 29
figure 29

See text for description

Competing interests None.


1. De Vries JP. Kidney International (1993) 44, 851–854.

2. Tonelli M. Kidney International (2002) 62, 1075–1080.

P73 Risk factors for occurrence of acute kidney injury in diabetic ketoacidosis

Correspondence: Ines Sedghiani -

Annals of Intensive Care 2016, 6(Suppl 1):P73

Introduction Acute kidney injury (AKI) is a common complication of non-equilibrated diabetes, and it may occur in patients with ketoacidosis (DKA) either in admission or during hospitalization

Patients and methods It is a retrospective study including all cases of diabetic ketoacidosis between January 2014 and August 2015. Patients with chronic renal failure were not included.

The diagnosis of diabetic ketoacidosis was defined by the association of blood glucose >11 mmol/L, ketonuria and metabolic acidosis with a pH <7.30 and/or plasma bicarbonate <15 mmol/L.

On admission, acute kidney injury was defined by a reversible increase in serum creatinine >110 μmol/L or in blood urea >9 mmol/l or a reversible decrease in the urinary output <0.5 ml/kg/day.

Results In this study, 53 patients with a mean age 41 ± 21 years were included. They have been diagnosed with type 1 diabetes in 60 % of the cases and with type 2 diabetes in 40 % of the cases, on average 5 years before admission. The majority of the patients were female (56 %). Sixteen of the patients have had a previous episode of DKA. The origin of DKA was attributed to treatment omission in 21 (39 %) patients, infections in 16 (30 %) patients and diet errors in six (11 %) patients.

On admission, they have a mean serum glucose 32 ± 12 mmol/L, mean pH 7.2 ± 0.08 and a bicarbonate level of 9.2 ± 4.2 mmol/L.

On admission, patients were dichotomized in with AKI and without AKI. Clinical and biological parameters were compared in these two groups. Twenty-six patients (49 %) have acute renal failure on admission, with a mean serum creatinine of 136 ± 41 μmol/L and a mean blood urea of 10 mmol/L compared with 72 ± 32 μmol/L and 6.6 ± 4 mmol/L at discharge (Table 16).

Table 16 See text for description

Conclusion Acute kidney injury complicates frequently diabetic ketoacidosis on admission in ICU. Age, blood glucose, high osmolarity and occurrence in type 2 diabetes are their risk factors.

Competing interests None.

P74 Face-to-face tracheal intubation with the video laryngoscope Airtraq in the sitting patients: a report of 3-year experience in patients after failed conventional tracheal intubation technique

Correspondence: Gilles Dhonneur -

Annals of Intensive Care 2016, 6(Suppl 1):P74

Introduction The video laryngoscope Airtraq (VAQ) is now placed in the difficult airway management algorithm as second-step device to intubate the tracheal of patients when conventional tracheal intubation (TI) technique with Macintosh laryngoscope (ML) has failed (1). VAQ was also shown in simulation conditions to promote simple face-to-face (F2F) tracheal intubation in the patients placed in sitting position (2). With significant clinical experience acquisition, we have decided to propose F2F-VAQ-TI as an alternative to awake fiberscope-assisted nasotracheal intubation (AANI) in the patients after failed ML.

Patients and methods After informed consent of the patient initially scheduled for elective AANI was obtained, those with previous failed ML, showing inter-incisor distance of >20 mm, with no expansive tumoral process in the upper airway and those with easy identification of anatomical cricothyroid membrane landmarks were proposed F2F-VAQ-TI in sitting position. An Eschmann stylet (ES) was systematically armed in the tracheal tube inserted in the VAQ channel. Preoxygenation and anesthesia induction techniques were standardized. Filmed TI maneuver was started 30 s after succinylcholine IV bolus (1 mg kg−1) was injected. Insertion of an LMA Fastrach was considered as a Plan B in case of difficulty to maintain SaO2 > 90 %. The fiberscope used with VAQ or LMA Fastrach was recommended as impossible tracheal access Plan B. In the PACU, the patients were asked to evaluate the level of comfort of the proposed airway management strategy using a visual analogue scale (VAS 0–100). For those patients that had previously been proposed AANI, they were asked to quote their comfort during both previous AANI and F2F-VAQ-TI. Moreover, they were asked to choose for the next surgery step the airway management technique among the two proposed options.

Results Over 3 years, 106 F2F-VAQ-TI were performed. The surgical indications requesting TI were as follows: maxillofacial reconstruction (41 %), orthopedic (32 %), neurological and visceral. The reported reasons for previous ML failure were as follows: restricted mouth aperture of below 25 mm, major abnormality of anterior neck structure, severe cervical spine ankylosis, previous causes combined with post-surgery/radiation sequels or unknown non-tumoral causes. Mean age of the patients was 49 years. Among them, 3 patients benefited of 3 F2F-VAQ-TI and 5 of 2. TI succeeded in all patients. TI duration was <15 s, 15 to <30 s, 30 s to <1 min, 1–2 and >2 min: 24, 25, 42, 6, and 3 %, respectively. ES was used in 61 % of cases. Plan B was used twice. One tracheal intubation was performed using the fiberscope inserted in the tracheal tube armed in the VAQ. Unrecognized laryngeal cyst prevented both tracheal tube and ES tracheal penetration. The second Plan B concerned a 39-year-old woman (1.29 m height; 35 kg weight) that was oxygenated and intubated through the LMA Fastrach (size 3). SaO2 < 90 % episodes occurred in 2 patients (nadir 86 and 78 %). All the patients (n = 36) that had received AANI for previous surgery would ask F2F-VAQ-TI for the next surgery. The mean (SD) comfort of F2F-VAQ-TI was remarkably high: 90 (6).

Conclusion We have shown that with respect to our selection process and airway management procedures F2F-VAQ-TI in the sitting patient was a safe alternative to AANI. This new airway management option seems simple for trained skilled operators and very comfortable for the patient. Comparison of the learning process of AANI and F2F-VAQ-TI is now made in our laboratory and department.

Competing interests Consulting activity for PRODOL Company.


1. Anesthesiology, 2011; 114:25–33.

2. Br J Anaesth, 2012; 108: 140–50.

P75 Ultrasound-guided central venous access technique among French intensivists

Correspondence: Julien Maizel -

Annals of Intensive Care 2016, 6(Suppl 1):P75

Introduction For many years, the use of ultrasound has been recommended to secure central venous catheter placement in the intensive care units. Many studies and meta-analysis have demonstrated that ultrasound-guided technique increases the success rate and decreases the number of complications by novice and experienced operators. However, several surveys showed that only 50 % of the CVC procedures are performed using the ultrasound-guided technique. The reason why physicians continue to use landmarks was the lack of formation and the absence of ultrasound device. In our area (northwest of France), ultrasounds have been taught to our residents and learning courses were implemented for senior physicians. Therefore, we designed a survey to elicit information on physician’s characteristics, experience in CVC placement, training on ultrasound technique, the use of ultrasound for CVC placement, reasons for nonuse of ultrasound and their opinion on the necessity to continue to teach the landmark technique to residents.

Patients and methods This survey (14 questions) has been electronically addressed by email (using the Survey Monkey online software) to every physician belonging to the group of research on clinical practice in the northwest of France composed of 8 university’s and 20 community’s ICU. The survey was sent to the 289 physicians (166 seniors and 121 residents) working in those ICUs.

Results We received 190 responses (response rate 66 %). Among the respondents, 66 % were less than 40 years old, 34 % were residents, 41 % presented an experience in CVC placement ≤5 years and 53 % declared putting more than 1 CVC per week. The preferred site of insertion was the jugular for 70 %, subclavian for 21 % and femoral for 9 %. Seventy-one percentage of the residents declared having learned both landmark and ultrasound-guided techniques during their residency. Only 18 % reported using always the ultrasound to put CVC. The main reasons why they do not use ultrasound were as follows: (1) they think they do not need it (36 %), (2) the ultrasound was not available (33 %) while 3 % declare the absence of ultrasound device in their setting and (3) 11 % reported the absence of formation on ultrasound technique. The ultrasound technique used was ultrasound-guided in 66 %. Coagulation abnormalities (64 %), obesity (54 %) and anatomical difficulties (52 %) were the main motivations to use ultrasound reported. Fifty-three percentage of the respondents declared having confronted at least one time during the last 12 months to an urgent situation where the ultrasound was not available quickly enough. Finally, 91 % think that the landmark technique should still be taught to the residents.

Conclusion Our survey shows that a large majority of the intensivists declare still performing landmark procedures while almost every physician received a formation on the ultrasound technique. Ultrasound device is present in the large majority of the institution; however, their availability is still scarce. Despite the several guidelines, physicians still do not use ultrasound for every procedure and think that landmark technique should still be taught to residents.

Competing interests None.

P76 Ultrasound-guided venous catheter insertion. What do French intensivists really do?

Correspondence: Hafid Ait-Oufella -

Annals of Intensive Care 2016, 6(Suppl 1):P76

Introduction Accumulating evidence suggests that ultrasound-guided venous catheter insertion decreases insertion failure and immediate mechanical complications rates. However, due to misleading opinion, ignorance or absence of training, this method has not been widely adopted. Moreover, the exact utilization of this method among French intensivists is unknown. The aim of this prospective survey was to assess knowledge and utilization of ultrasound guidance for venous catheter insertion.

Patients and methods A questionnaire (16 questions) assessing knowledge, training, mastery and use of ultrasound guidance for central venous catheter (CVC) insertion, constructed by the educational section of the National Intensivists French College (CNER), was electronically sent to French intensivists between May and June 2015 (CNER and SRLF Web site).

Results Answers to the questionnaire were obtained from 300 intensivists [attending physicians (50 %), residents (16 %), fellows (16 %), professor or assistant professor (16 %)] with initial training in anesthesiology (56 %), pulmonology (13 %), internal medicine (10 %) or cardiology (7 %) and who had for most of them (84 %) inserted more than 100 CVCs. Ultrasonography use was significantly higher for internal jugular vein (75 %) than for femoral vein (45 %) or subclavian vein (29 %). Dynamic ultrasonic guidance isolated use was reported by 38 % of respondents while 45 % used both screening before cannulation and dynamic guidance. Use of short-axis approach, as well as the long-axis one according to the site, was reported by 56 % of respondents. The majority of respondents (85 %) reported having never used a specific guiding device. Whereas 88 % of respondents considered specific training as mandatory or absolutely mandatory, only 44 % of them (essentially the youngest ones) reported having received a specific training (most of the them on an inanimate mankind).

Conclusion Despite insufficient training, most of French intensivists report relevant knowledge on ultrasound guidance for venous catheter insertion.

Competing interests None.

P77 Can we trust new connected devices recording heart rate and oxygen saturation? A concordance assessment in intensive care unit

Correspondence: Yacine Tandjaoui-Lambiotte -

Annals of Intensive Care 2016, 6(Suppl 1):P77

Introduction Connected devices have become increasingly common, including for healthcare purposes, since some items now allow a continuous measure of the heart rate (HR) and percutaneous oxygen saturation (SpO2). These devices are sold for wellness, and to date, they are not approved by health authorities as medical devices. While their use for chronically or acutely ill patients would be of great interest, there are no existing data regarding the reliability of their measurements.

Materials and methods We performed a prospective monocentric study in the intensive care unit (ICU) of an university hospital to assess the concordance of HR and SpO2 measured using the three connected devices available on the market to an ICU monitor as gold standard. All consecutive patients hospitalized in ICU during September 2015 were included except for those under vasopressor or mechanical ventilation.

  • Connected devices:

Withings Pulse Ox

Safe Heart Smartphone iOximeter

iHealth Wireless Pulse Oximeter

  • Monitor:

Philips MP60 or MP70.

Measurements were taken during the first 24 h after ICU admission during nurses’ usual round or when an intercurrent event occurred. A maximum of 12 measures per patient per day were performed, and measures could be registered over several days. At each time point, HR and SpO2 were measured using the three connected devices and the gold standard.

A concordance analysis was performed to assess the reliability of each connected device compared to the gold standard. Intraclass correlation coefficients and 95 % confidence intervals (CI) were estimated using a single pair of measures per patient. Bland–Altman diagrams for repeated measures were used to illustrate concordances.

Results Twenty-four patients, among which 16 (67 %) men, were included in the analysis. The median age was 60 years [interquartile range (IQR) 54–72], and the main cause admission was a medical issue [23 (96 %)]. Overall, 119 different measures were analyzed with a mean of 5 per patient ranging from 1 to 12 measures per patient. The median SpO2 measured on the monitor was 96 [IQR 94–99]. The corresponding measures from devices 1, 2 and 3 were, respectively, 93 [IQR 90–95], 94 [IQR 91–96] and 98 [IQR 97–99]. The median HR measured on the monitor was 98 [90–109]. The corresponding measures from devices 1, 2 and 3 were, respectively, 95 [IQR 86–105], 95 [91–102], 98 [90–109]. For SpO2, the intraclass correlation coefficient between the monitor and the device 1 was 0.12, between the monitor and device 2 was 0.32 and the monitor and device 3 was 0.17. Concerning HR, the intraclass correlation coefficient between the monitor and device 1 was 0.68, between the monitor and device 2 was 0.67 and the monitor and device 3 was 0.08. Moreover, devices failed to provide any SpO2 measure in 0, 0.8 and 23 % of cases for devices 1, 2 and 3, respectively. Device 3 also failed to measure any HR measure in 12 % of the cases, while devices 1 and 2 always succeeded in measuring HR.

Conclusion We observe a poor correlation between devices and the gold-standard measures, especially for SpO2. These results need to be confirmed in larger studies. For now, these connected devices should not be recommended for SpO2 and HR measurement in critically ill patients admitted in ICU.

Competing interests None.

P78 Hemoglobin monitoring with HemoCue©, spectrophotometry and blood cells analyzer: comparison of agreement, completion time and costs

Correspondence: Benjamin Sztrymf -

Annals of Intensive Care 2016, 6(Suppl 1):P78

Introduction Anemia is frequent in the critical care setting. It is the result of several mechanisms such as hemodilution, hemorrhage or inflammation consequences. Blood spoliation due to repeated blood punctures is also one of the factors leading to anemia. The reference method to determine the hemoglobin concentration is the cyanohemoglobin method performed with an automatic hematology analyzer (HA). The main pitfalls of this method are to require a certain volume of blood, further enhancing the blood spoliation, and to be completed in a laboratory, increasing the lag time to diagnosis. Alternate methods such as HemoCue© (HC) or spectrophotometry with the blood gas analysis (BGA) have been developed to address these issues. Nevertheless, accuracy of these methods remains debatable. Therefore, we conducted a study to compare the aforementioned techniques agreement, completion time and costs in ICU.

Patients and methods This is a single-center prospective study. All patients undergoing both a blood cell count and an arterial blood gas analysis were included. A capillary HemoCue© was performed just after the blood puncture and the capillary hemoglucotest. The completion time was registered, as well as the hour of the blood puncture. We also registered the hours at the blood sample arrival at the laboratory and at the availability of the result on the hospital intranet. The HA was considered as the reference method. Costs were estimated according to the French health system charts. Agreement between the methods has been estimated with Bland and Altman test. Results are given in mean ± SD. Comparisons have been made with Student’s t test, ANOVA or Chi square test as appropriate.

Results Fifty-one samples have been compared in 37 patients (age 70 ± 15 years, SAPS II 48 ± 20). We found a difference when comparing the hemoglobin concentration with HA, BGA and HC (respectively, 9.9 ± 1.8, 10.2 ± 2.1 and 9.5 ± 2.5 g/dl, p = 0.003). The comparison between HA and BGA evidenced a systematic positive deviation (+0.25 g/dl). Limits of agreement (±2 SD) were ±1.94 g/dl. Three measurements were outside these limits. The comparison between HA and HC evidenced a systematic negative deviation (−0.4 g/dl). Limits of agreement were ±2.92 g/dl. Three measurements were outside these limits. A transfusion was indicated once according to recommended transfusion thresholds by HA. This transfusion was also indicated by BGA. HC would have indicated nine transfusions, only one being confirmed by HA.

HC was performed in 1.2 ± 0.5 min (min). The time between the blood sample and the result’s availability was significantly lower for the BGA as compared to HA (34 ± 19.9 vs. 54.5 ± 32.6 min; p = 0.01). It was the result of a faster laboratory technique (23.7 ± 15.5 vs. 31.6 ± 20.2 min; p = 0.03), whereas the transport time of both samples was not different.

The costs of HA, BGA and HC were estimated to be, respectively, 426.87, 1032.75 and 45.9 euros. Nevertheless, when summing these “direct costs” with the price of the transfusions that would have been indicated according to the tests, HC became the most expensive tool (5022.9 euros) as compared to HA (979.87 euros) or BGA (1585.75 euros).

Discussion Among the three methods, HC was the fastest to perform, but seemed to be the less accurate as well. Its precise position in the ICU setting must be discussed, since the reference test appears to be mandatory to indicate a transfusion. Its utility may be more substantial in the pre-hospital setting where no other hemoglobin measurement tool is available. The blood gas analysis, often performed in ICU, offers the advantage to be more accurate than HemoCue© and to provide other useful information.

Conclusion Alternate methods of hemoglobin monitoring allow a faster result, but their accuracy is debatable. Their respective utility might depend on the clinical setting.

Competing interests None.

P79 Activated partial thromboplastin time and anti-Xa measurements in heparin monitoring among critical care patients

Correspondence: Frédéric Jacobs -

Annals of Intensive Care 2016, 6(Suppl 1):P79

Introduction Monitoring the anticoagulant effect of unfractionated heparin (UFH) is mandatory. This monitoring can be done by the mean of the activated partial thromboplastin time (aPTT) or by anti-Xa level measurements. Compared with anti-Xa levels testing, aPTT is more frequently impacted by preanalytic variables and biologic factors (increased levels of acute-phase reactants, consumption coagulopathy) often encountered among critically ill patients. We studied the agreement of both test results in unselected critically ill patients.

Patients and methods aPTT and anti-Xa levels were simultaneously monitored in patients treated by continuous intravenous infusion of UFH. Blood samples were drawn into sodium citrate tubes (Greiner Bio-One SAS, France). aPTT was measured with TriniCLOT Automated aPTT reagent (Tcoag, Ireland) and anti-Xa levels with Biophen Heparin (LRT) (HYPHEN Biomed, France). An aPTT of 2–3 times the control and anti-Xa levels between 0.3 and 0.7 IU/ml were defined as therapeutic.

Results Forty-four patients (mean age 71.13 ± 15.7 years; mean SAPSII 39.1 ± 14.4) were included. Reasons for admission were medical in 31 and surgical in 13. The indications for UFH therapy were atrial fibrillation (26), venous thromboembolism/pulmonary embolism (13), thrombophilia (2), acute coronary syndrome (1), and arterial thrombosis (2). Paired measurements of aPTT and anti-Xa were taken on 353 samples. Linear regression analysis was used to evaluate the relationship between aPTT and anti-Xa.

The correlation between aPTT and anti-Xa levels was low (r = 0.495).

The concordance of tests results is shown in Table 17.

Table 17 Cross-tabulation of clinically relevant aPTT and anti-Xa levels

Concordant aPTT and anti-Xa values were observed in 213 (60.3 %) data pairs. aPTT was discordantly high in 96 (29.9 %) data pairs and discordantly low in 44 (12.5 %) ones.

Discussion In an unselected population of critically ill patients, the concordance of aPTT and anti-Xa levels was low. Considering anti-Xa as gold standard, monitoring anticoagulation treatment by aPTT leads to a high risk of misdosing. aPTT is frequently impacted by biologic factors. Although less commonly, anti-Xa levels can also be influenced by biologic cofounders.

Poor correlation between aPTT and anti-Xa could result from alterations in FII and FVIII activity.

Conclusion Use of aPTT and anti-Xa levels to guide heparin therapy may lead to different estimates of UFH concentration in the same patient.

Both aPTT and anti-Xa have limitations when used for UFH monitoring and may not accurately assess anticoagulant status. Further investigation (using thromboelastometry or thrombin generation assays) could be useful to determine the optimal anticoagulation testing protocol in critically ill patients.

Competing interests None.

P80 Agreement between the Point-Of-Care Siemens RAPIDPoint® 500 Blood gas system and central laboratory measurement of hemoglobin, hematocrit, glucose and electrolytes in ICU patients

Correspondence: Jérôme Allardet-Servent -

Annals of Intensive Care 2016, 6(Suppl 1):P80

Introduction Several blood analyses are performed in critically ill patients. Rapid detection of abnormal values, by point-of-care testing, allows clinicians to promptly deliver the appropriate therapy. We sought to investigate the agreement between the last generation of blood gas analyzer and central laboratory measurement of electrolytes (sodium, potassium, chloride, bicarbonate), hemoglobin, hematocrit and glucose.

Patients and methods Three hundred and fourteen paired samples were collected prospectively in 51 patients. All samples were drawn from an arterial line in the morning at the 06:00 am. Biological tubes were collected through a Vacutainer® and analyzed in the central laboratory (Lab) using Beckman Coulter® automates (DXH and AU 5800). BD Preset® heparinized syringe was filled with 1.6 ml of arterial blood and analyzed immediately in the ICU using the Point-of-Care Siemens RAPIDPoint® 500 Blood gas system (POC). Measurements obtained by the two methods were compared using the Bland–Altman method (bias and limit of agreements), Deming regression analysis and Pearson correlation. We use the proficiency testing criteria to define acceptable analytical performance (CLIA 1992): ±4 mmol/l for sodium, ±0.5 mmol/l for potassium, ±5 % for chloride, ±0.6 g/l for glucose, ±6 % for hematocrit and ±7 % for hemoglobin.

Results The main results are presented in the Table 18.

Table 18 See text for description

Conclusion The Point-of-Care Siemens RAPIDPoint® 500 Blood gas system satisfied CLIA criteria of acceptable analytical performance for all tested parameters except for hemoglobin (difference = 7.6 %). We identified a systematic overestimation of hemoglobin (mean 0.8 g/dl) with the RAPIDPoint® 500 Blood gas system.

Competing interests None.


1. Ehrmeyer SS, Laessig RH, Leinweber JE, Oryall JJ: 1990 Medicare/CLIA final rules for proficiency testing: minimum intralaboratory performance characteristics (CV and bias) needed to pass. Clin Chem 1990, 36:1736–1740.

2. CLSI. Measurement Procedure Comparison and Bias Estimation Using Patient Samples; Approved Guideline-Third Edition. CLSI document EP09-A3. Wayne, PA: Clinical and Laboratory Standards Institute; 2013.

P81 Clinical features and prognosis of pesticide poisoning

Correspondence: Neila Maaroufi -

Annals of Intensive Care 2016, 6(Suppl 1):P81

Introduction Deliberate self-poisoning with pesticide continues to be a major public health concern in many developing countries. This study aimed to evaluate the data on cases of acute pesticide poisoning and to compare different variables between survivors and non-survivors.

Patients and methods In this cross-sectional study, medical records of all pesticide-intoxicated patients were reviewed from 2009 to 2013, retrospectively. Demographics, clinical features and laboratory findings were evaluated. The variables compared between survivors and non-survivors were the amount of pesticide ingested, occurrence of vomiting after ingestion, time and place of hospital admission, length of hospital stay, leukocytosis, serum creatinine level and the outcomes.

Results A total of 154 patients were evaluated. The mean ± standard deviation of patients’ age was 32 ± 18 years. The length of hospital stay was 4.76 ± 4.8 days. Most poisonings occurred in spring and summer. The in-hospital fatality rate was 36 %. Statistically significant associations were found between the outcome of patients and the amount ingested (P = 0.002), vomiting (P = 0.003), early need to intensive cares (P = 0.005), leukocytosis (P = 0.001) and serum creatinine levels (P = 0.001).

Discussion and conclusion Prompt vomiting, early need to intensive cares, leukocytosis and multi-organ failures are major determinants for fatal outcome of pesticide poisoning. It may be useful to educate health professionals and the general population about the serious consequences of exposure to pesticide.

Competing interests None.

P82 Variability of neurobehavioral toxicity of naphyrone, a new synthetic cathinone, according to its acute or binge administration

Correspondence: Bruno Megarbane -

Annals of Intensive Care 2016, 6(Suppl 1):P82

Introduction Naphyrone is a new psychoactive substance and a cathinone derivate. Naphyrone was reported to be used by approximately 2 % of recreational drug users and is frequently present in bath salts. Scarce data are available regarding its toxicity, pharmacokinetics and effects on brain monoamines. Our objectives were to study naphyrone-related effects on behavior and brain monomaine content according to two administration modalities (acute and binge) mimicking its use in humans.

Materials and methods We performed an experimental study on the effects of naphyrone-induced effects after acute and repeated administration (binge) on the locomotor activity, anxiety (openfield), resignation (forced swimming), memory (Y-maze), hedonic status (sucrose consumption) of Swiss mice. We also investigated the effects of naphyrone-induced on monoamines in the prefrontal cortex. Plasma naphyrone concentrations were measured using high-performance liquid chromatography (HPLC) coupled to mass spectrometry (MS) and brain monoamine concentrations using HPLC coupled to fluorometry. For each animal and each time, we calculated the difference between the parameter value at that time and baseline and the area under the curve of its time course. Comparisons were made using two-way ANOVA followed by posttests using Bonferroni correction. Pharmacokinetics (PK) was modeled and parameters calculated using WinNonlin® software.

Results Naphyrone induced dose-dependent stimulation of locomotor activity that appeared more marked and prolonged than MDPV- and cocaine-related effects (used as positive controls), additionally increasing after its binge administration (p < 0.001). Significant increase in the distance walked at the periphery of the openfield was observed up to 24 h postinjection (p < 0.001), corresponding to the behavior compensation of enhanced frighten. During binge administration (3 times per day, 3 successive days), increased locomotor effects at day 3 (after the 9th injection) in comparison with day 1 (after the 1st injection) supported naphyrone-induced hyper-sensitization process. Similarly, significant effects were observed on mice depression (p < 0.0001) and memory (p < 0.005) but not on hedonic status. Significant dose-dependent increase in aggressively social relationships was also reported among naphyrone-treated mice (p < 0.001). The neurochemical study revealed significant increase in dopamine and norepinephrine concentrations in the prefrontal cortex, without significant modifications in serotonin concentrations. This monoamine profile was similar after repeated naphyrone administration. Naphyrone PK profile was described after acute and binge administration and effects correlated with plasma concentrations.

Conclusion Neurobehavioral disorders induced by acute and repeated naphyrone administration mainly consist in the stimulation of locomotor activity. The increase in the cortical dopamine concentrations may suggest an addictive potential that should be further investigated.

Competing interests None.

P83 Acute acenocoumarin poisoning: an epidemiological and clinical study

Correspondence: A. M’rad -

Annals of Intensive Care 2016, 6(Suppl 1):P83

Introduction Acenocoumarin is the most commonly prescribed oral anticoagulant in our country. Because of the paucity of data on acute acenocoumarin poisoning, we undertook this study, which aimed to determine the epidemiologic, clinical, therapeutic characteristics and outcome of this intoxication.

Patients and methods It was a retrospective study performed over a 9-year period (from June 2006 to June 2015). Demographic characteristics, acenocoumarin long-term treatment history, clinical signs, treatment and outcomes of this intoxication have been analyzed.

Results Fifty-eight patients aged 28 ± 11 years were eligible; their sex ratio was of 1. All cases were self-inflicted. Among them 11 males were detained. Acenocoumarin long-term treatment was noted in 40 % (n = 23) patients, for thromboembolic disease (n = 17) or valvular replacement (n = 6). Coingestion of other drugs was reported in 43 % of cases. The mean ingested dose was 56 ± 48 mg (extremes 8 and 240 mg). The average time of consultation was 9 ± 11 h. In the majority of cases, patients were asymptomatic. Bleeding was reported in 4 detained patients but not observed by physician. Gastric lavage was performed in 17 % of cases within 1 h after ingestion. Vitamin K was administrated in 52 % of cases. No patient required PPSB perfusion or transfusion. The outcome was favorable in all cases. No deaths were reported. Length of stay was 30 ± 24 h.

Discussion Acenocoumarin overdose was well described in the literature, and the treatment is well codified.

However, few studies were interested in acute poisonings and their management. Our study suggested that acute acenocoumarin poisoning is often not serious if patients were early managed. The administration of vitamin K could be necessary and safety, especially for patients without long-term anticoagulant treatment.

Conclusion Acenocoumarin poisoning is increasingly frequent in our country; its prognoses are generally favorable. Hospitalization must be indicated in symptomatic patients who presented active bleeding or those with long-term anticoagulant treatment. The others can be managed ambulatory.

Competing interests None.


1. Granditsch G, Pilgerstorfer HW. Acenocoumarin (Sintrom) poisoning in a child Wien. KlinWochenschr. 1971;83(4):62–3.

2. Isbister GK, Whyte IM. Management of anticoagulant poisoning. Vet Hum Toxicol. 2001;43(2):117.

P84 Death by deliberate self-poisoning in Sfax, Tunisia: Victims characteristics and toxic substance profile

Correspondence: Maatoug Samir -

Annals of Intensive Care 2016, 6(Suppl 1):P84

Introduction Deliberate self-poisoning is a serious problem in Tunisia. However, our understanding of the problem of suicidal intoxication remains limited. This study aims to reveal the frequency of suicide by self-poisoning in the region of Sfax, to describe the characteristics of victims and to identify the difficulties often encountered in determining the toxic substance.

Patients and Methods This study is a retrospective study conducted during a period of 08 years (2007-2014), about all cases of self-poisoning collected in the Department of Forensic Medicine of Habib Bourguiba University Hospital of Sfax, Tunisia.

Results During the study period, 299 cases of suicide are compiled. In 42 cases (14 %), suicide is committed by self-poisoning, representing a little more than 5 cases per year. The sex ratio is 0,7. The mean age of victims is 32.9 ± 7,6 years with extremes ranging from 18 to 63 years. In 25 cases (60 %) the victim is single, in 15 cases (35 %) she is married and she is divorced in only 2 cases (5 %). She is unemployed in 18 cases (43 %) and she is a daily-worker in 16 cases (38 %). In half of cases the victim doesn’t have any known psychiatric illness history. The antecedent of suicide attempts is found in 8 cases (19 %).In only 14 cases (33 %) death occurred at home. Most of victims were hospitalized in intensive care unit (28 cases, 67 %). The average time interval between hospitalization and death is 32.45 hours with extremes ranging from 8 to 96 hours.

The forensic autopsy is often performed more than one day after the intoxication date. The post-mortem toxicological samples systematically made during the autopsy were positive in only 5 cases (12 %). Toxic samples are often not made during the hospitalization. The nature of the toxic substance is determined based only on history and clinical information. In our study the nature of the toxic substance is a pesticide in 16 cases (38 %) and a medication in 14 cases (34 %). It is still undetermined in 6 cases (14 %).

Conclusion Toxicological samples performed after a prolonged interval of time from the intoxication moment are often negative. Thus it is fundamental to realize them systematically at the time of admission into intensive care unit in case of intoxication suspicion. The determination of toxics allows not only to orientate the therapeutic management but also to strengthen preventive strategies against suicide.

Competing interests None.

P85 Adverse drug reactions (ADRs) collected by medical-staffed ambulances: pilot study

Introduction In Paris area, emergency medical assistance is run by the “SAMU de Paris.” Medical-staffed ambulances go directly to take care of any patient with an acute medical problem after an evaluation of the clinical situation made by phone.

Aim: Collection, analysis, and evaluation of adverse drug events (ADRs), whatever it is, acute intoxication, organ failure, allergy, and whatever its clinical presentation.

Methods: Every month, forms from all medical interventions performed daily by the SAMU de Paris are analyzed retrospectively by a physician trained in pharmacovigilance. The following information is available: patient’s main characteristics, description of the medical event, main diagnosis, and outcome. If an ADR is present or even suspected, a copy of the form is made. The pharmacovigilance centers contact medical staff that took care of the patient to retrieve more information about the final diagnosis and the drug involvement. If a drug adverse effect is retained, the case is anonymously registered in the national database, and a specific letter is sent to describe and sum up the case.

Results From January 1, 2015, to July 31, 2015, 274 cases of possible ADRs have been collected.

  • 203 cases of ADRs (74 % of the total) have been immediately diagnosed and registered.

  • The drug causality is pending for 29 cases. For the last 37 cases, a drug implication has been definitely ruled out, with no drug intake, another causative agent, illicit substances and food.

  • For all ADRs retained (n = 203), a fatal outcome occurred in 14 cases (7 % of ADRs). The main organ failure was central nervous system (n = 107, 53 %), cardiovascular (n = 55, 27 %), digestive (n = 21, 10 %), and lung (n = 8).

  • The type of ADRs was voluntary overdose (n = 60, 30 %), bleeding (n = 48, 24 %), hypoglycemia (n = 40, 20 %), allergy (n = 15, 7 %), and medication error (n = 11, 6 %).

  • The following drugs were involved for all ADRs retained (n = 203): psychotropic drugs (n = 73, 36 %); drugs acting on hemostasis (n = 48, 24 %); antidiabetics (n = 41, 6 %); pain killers (n = 12, 6 %); antiinfectious (n = 11, 6 %), and various (15): NSAID (8)—anesthetic agents (3)—corticosteroid (3)—contrast media (2)—potassium (1), vaccine (1), GcSF (1), colchicine (1).

Discussion and conclusion To our knowledge, this is the first time that serious adverse drug reactions are collected directly from the first step of emergency care, before hospitalization. Excepted intentional drug overdose for suicide attempts, the main ADRs observed such as bleeding with anticoagulants or hypoglycemia with insulin, are expected and evitable. Further analysis on cardiovascular effects is pending. This study increases awareness of physicians working in SAMU. The usefulness of toxicological analysis must be discussed.

P86 Prognostic value of plasma concentration of acebutolol in acute poisoning

Correspondence: A. M’rad -

Annals of Intensive Care 2016, 6(Suppl 1):P86

Introduction Beta-blockers could be responsible of severe poisoning leading to death. The knowledge of prognostic factors is essential and could guide the management of these patients.

The purpose of our study is to determine whether the plasma concentrations of acebutolol can predict clinical symptoms and prognosis.

Patients and methods It was a retrospective study conducted over 2 years in a 16-bed toxicological unit including all admitted patients for acebutolol poisoning, presenting with shock or cardiac arrest. A systematic serum measure of acebutolol was done at least once hospital stay.

Results Eleven patients were eligible. The mean age was 23.0 ± 6.6 years with a female predominance (9/11).

The average of the presumed ingested dose was 8050 ± 4812 mg.

On admission, cardiovascular manifestations were predominant; the main presenting symptoms were bradycardia (n = 5), shock (n = 10) and cardiac arrest with favorable evolution in one case.

Electrocardiogram showed a heart block in 7 cases: first degree (n = 4) and third degree (n = 3), wide QRS complex with sodium channel-blocking effect in 7 cases (mean QRS duration was 220 ± 77 ms) and a prolongation of the QT interval in 8 patients (the mean QT interval was 521 ± 79 ms). The mean plasma concentration of acebutolol at admission was 19.3 ± 16.6 mg/l. Six patients had required mechanical ventilation due to hemodynamic instability. Inotrope drugs were required in eight patients; semi-molar sodium bicarbonate was infused in seven patients with an average volume of 500 ± 144 ml.

Three patients died. Two early deaths were related to intoxication, and the later was related to nosocomial complication.

The analysis of prognostic factors showed the absence of correlation between acebutolol plasma concentration either on admission or any time and clinical or electrocardiographic signs in serious poisoning.

Conclusion The severity of acebutolol poisoning must be assessed by clinical and electrocardiographic manifestations. The plasma concentration of acebutolol should not be considered as prognostic factor.

Competing interests None.

P87 Baseline pharmacologic treatment for patients admitted in ICU for AECOPD: trends in use and conformity to the GOLD guidelines

Correspondence: Islem Ouanes -

Annals of Intensive Care 2016, 6(Suppl 1):P87

Introduction The quality of baseline treatment of COPD patients is important in the prevention of acute exacerbation. The aim of our study is to describe trends in baseline pharmacologic treatment (BPT) use guidelines in a cohort of patients admitted to a Tunisian ICU for AECOPD and to assess its conformity with GOLD, which is intended to classify patient’s COPD severity and to guide patient’s management.

Patients and methods In a cohort study of consecutive patients admitted between 2000 and 2012 for AECOPD in our ICU, we analysed BPT and its conformity with GOLD guidelines issued in 2001 and 2007.

Results During the study period, 440 patients admitted for 552 episodes of severe AECOPD requiring ventilatory support (NIV in 67.8 %). Table 19 shows patients’ characteristics and baseline medications used in this cohort:

Table 19 See text for description

Aminophylline use decreased significantly from around 50 % (in 2000) to 21.4 % (in 2012), short-acting ß2 mimetic use remained unchanged (70.1 % in 2000 and 64.3 % in 2012), whereas inhaled anticholinergics and corticosteroids increased significantly during the same period (respectively, from 0 to 18 % and from 13 to 60 %). In patients in whom FEV1 was measured and GOLD classification was possible, BPT was compliant with 2001 and 2007 recommendations in 58.9 and 15.6 %, respectively.

Conclusion In this cohort of severe COPD admitted to ICU for acute exacerbation, compliance with GOLD 2001 and 2007 varied substantially and could be explained by changes regarding long-acting ß2 mimetic use.

Competing interests None.

P88 Blood eosinophils levels and acute exacerbations of COPD

Correspondence: Islem Ouanes -

Annals of Intensive Care 2016, 6(Suppl 1):P88.

Introduction Increased blood eosinophil count (greater than 2 %) has been described by several studies as reflecting eosinophilic airway inflammation exacerbation and a useful marker to guide systemic corticosteroids administration in acute exacerbations of COPD (AECOPD) [1, 2]. The aim of this study is to describe the rate and the patterns of hyper-eosinophilia in a population of patients admitted to an intensive care unit (ICU) for severe AECOPD requiring ventilatory support.

Patients and methods This is a retrospective cohort study, including patients consecutively admitted to our ICU between January and August 2015 for severe AECOPD requiring ventilatory support. COPD patients with an obvious cause of decompensation (pulmonary embolism, acute pulmonary edema, pneumothorax, pneumonia) were excluded. Continuous variables are presented as medians and interquartile ranges (IQRs) and compared by the Mann–Whitney test, whereas dichotomous variables are presented in percentage and compared by the Chi square test.

Results During the study period, 42 patients were admitted for AECOPD [median age was 66.5 years (IQR 60–74) with 83.5 % male]. Hyper-eosinophilia (>2 %) was present in 52.4 % of patients with a median value of 3.25 % (2.2–4.0). Table 20 compares the characteristics of patients with and without hyper-eosinophilia.

Table 20 See text for description

Conclusion Hyper-eosinophilia (>2 %) appears to be common in patients admitted for AECOPD (52.4 % of patients in our study) and seems associated with inflammation markers and a trend toward longer ventilation and ICU length of stay durations. A policy of corticosteroids prescription guided by blood eosinophilia count seems reasonable in our population.

Competing interests None.


1. Saha S, Brightling CE. Eosinophilic airway inflammation in COPD. Int J Chron Obstruct Pulmon Dis. 2006;1(1):39–47.

2. Bafadhel M, McKenna S, Terry S et al. Blood eosinophils to direct corticosteroid treatment of exacerbations of chronic obstructive pulmonary disease: a randomized placebo-controlled trial. Am J Respir Crit Care Med. 2012 Jul 1;186(1):48–55.

P89 POFE: postoperative outcomes of chronic obstructive pulmonary disease (COPD) frequent exacerbators after pulmonary resection—a pilot study

Correspondence: Suela Demiri -

Annals of Intensive Care 2016, 6(Suppl 1):P89

Introduction After pulmonary resection, a postoperative pulmonary complication is present in 12–40 % of cases. Risk factors such as COPD are well identified. The phenotype of frequent exacerbator (FE) has been recently described. No existing study has evaluated the rate of patients called ‘FE’ among COPD patients requiring pulmonary resection and the relations between exacerbations history and the incidence of acute respiratory postoperative complications (ARPC).

The main objective was to determine the frequency of ARPC (atelectasis, acute respiratory failure, pneumonia) following lung resection in COPD patients. The secondary objectives aimed to determine the frequency of extra-pulmonary postoperative complications and the prevalence of the ‘FE’ phenotype in this population, as well as its relation with the risk of ARPC.

Patients and methods It was a prospective, observational, single-center study of patients with COPD hospitalized for elective thoracic surgery in the center of Hôpital Cochin, Paris, France. The inclusion criteria were as follows: male or female adults, permanent airflow obstruction (VEMS/CV < 70 %), scheduled oncologic pulmonary resection. Collected data were as follows: COPD symptoms and severity scores, preoperative severity scores, comorbidities, per operative data, postoperative complications and 30-day mortality.

Results Since July 2014 to September 2015, 98 COPD patients were included for analysis. Twenty-one were considered as FE (21.6 %). COPD classification was as follows: I—5 (%), II—11 (%), III—2 (%). There was no difference in demographic or COPD severity score (CAT score, IPAQ, MRC) between FE and the rest of the population. There was no difference in the 30-day mortality.

Results are in Table 21.

Table 21 See text for description

Conclusion FE history does not seem to affect the ARPC rate following oncologic pulmonary resection in COPD patients.

Competing interests None.


1. McGarvey L, Lee AJ, Roberts J, Gruffydd-Jones K, McKnight E, Haughney J Characterisation of the frequent exacerbator phenotype in COPD patients in a large UK primary care population. Respir Med. 2015 Feb;109(2):228–37.

2. Nojiri T, Inoue M, Takeuchi Y, Maeda H, Shintani Y, Sawabata N, Hamasaki T, Okumura M. Impact of cardiopulmonary complications of lung cancer surgery on long-term outcomes. Surg Today. 2015 Jun;45(6):740–5.

P90 Optimal delivery of salbutamol with a pressurized metered-dose inhaler within a high-flow nasal therapy circuit

Correspondence: François Réminiac -

Annals of Intensive Care 2016, 6(Suppl 1):P90

Introduction Nasal high-flow therapy (HFT) is increasingly used in intensive care for patients suffering from respiratory failure who also often require inhaled bronchodilators. Pressurized metered-dose inhalers (pMDIs) combined with spacers are convenient and cost-effective to generate aerosols for both ambulatory and mechanically ventilated patients. Therefore, administration of salbutamol within a spacer, placed in the HFT circuit, using a pMDI, could be relevant. The aim of this bench study was to determine optimal settings in this regard.

Materials and methods We assessed the mass of salbutamol delivered downstream of a tridimensional anatomical model reproducing aerosol deposition and leakage at the nose and rhinopharynx level after ten pMDI actuations (delivery of 1000 μg = nominal dose) of a canister connected with a spacer chamber (CombiHaler®) placed in an HFT circuit and synchronized with inspiration. The aerosol was collected on a filter and the mass of salbutamol measured by spectrophotometry. The spacer was placed either before the humidification chamber or immediately before the nasal cannula, and we assessed the influence of different HFT flow rates (30, 45 and 60 L/min) and breathing patterns (“respiratory failure” with mean patient inspiratory flow of 45 L/min vs. “quiet” breathing with a mean patient inspiratory flow of 15 L/min) on the delivered mass of salbutamol (respirable mass). Six experiments were performed for each condition tested.

Results Connection of the spacer before the humidification chamber was always poorly efficient, whatever the HFT flow rate used or respiratory pattern simulated, with masses collected ranging from 1.5 to 2.5 % of the nominal dose. When the spacer was placed immediately upstream from the cannula, the masses collected were 11.8, 10.7 and 6.2 % of the nominal dose for HFT flow rates of 30, 45 and 60 L/min, respectively, with the quiet breathing setting. During simulation of respiratory distress, the amount of salbutamol collected on the filter was higher at the HFT flow rate of 30 L/min (16.8 vs. 11.8 %) while it was lower at 45 and 60 L/min, compared to the quiet breathing pattern (respectively 4.4 vs. 10.7 % and 3.3 vs. 6.2 %): Fig. 30.

Fig. 30
figure 30

See text for description

Conclusion Placing the spacer immediately upstream from the cannula was the most efficient configuration for pMDI salbutamol administration in an HFT circuit, actuation being synchronize with inspiration. The masses observed are likely to produce a therapeutic effect in the clinical setting, even in case of respiratory failure and high patients inspiratory flows. Indeed, an inspiratory flow matching the HFT may minimize aerosol leakage during inspiration. These findings may lay the foundations for the clinical evaluation of salbutamol delivery with a pMDI during HFT.

Competing interests None.

P91 Differences in perceptions between the various healthcare professionals working at the bedside of the risks related to the rehabilitation procedures

Correspondence: Laurent Poiroux -

Annals of Intensive Care 2016, 6(Suppl 1):P91

Introduction In intubated patients, early mobilization reduces the incidence of ICU-acquired weakness but also mechanical ventilation duration and length of stay both in the ICU and in the hospital. Even if a relative consensus about the definition of early mobilization has emerged (i.e., rehabilitation interventions that are initiated immediately after stabilization of physiologic derangements), there is no definite consensus about the safety criteria required to initiate the rehabilitation process in acutely ill patients. This is probably at least partly due to very different perceptions, particularly between the various healthcare professionals working at the bedside of the risks related to the rehabilitation procedures.

Materials and methods The aim of the study was to compare what the different members of the ICU multidisciplinary team consider as contraindications to perform rehabilitation in mechanically ventilated critically ill patients. Practically, as preliminary part of a multicenter interventional study on early mobilization, a questionnaire was sent to the referent nurse, physiotherapist and physician of each participating center. Each healthcare professional was asked to define whether passive or active mobilization, inside or outside the bed, could safely be performed in different clinical situations and with various ICU equipment. Each participant was asked to answer by “No” if the mobilization was contraindicated in the described situation and by “Yes” if the mobilization was possible.

Results Fifty-one questionnaires were analyzed (17 from nurses, 17 from physiotherapists and 17 from physicians). The percentages of “No” answers meaning that mobilization is contraindicated are reported in Table 22.

Table 22 See text for description

Conclusion In general, passive mobilization is considered as safe and possible by all the healthcare professionals in the majority of the clinical situations described. Oppositely, when active mobilization outside the bed is considered, major differences in perception of mobilization associated safety emerged within the various healthcare professionals. In particular, we noticed that the physiotherapists were less afraid by poor respiratory conditions than nurses and doctors. On the other hand, the presence of an arterial femoral line or of a dialysis catheter in femoral position was mainly considered as a contraindication of mobilization outside the bed by the nurses.

Competing interests None.


1. Pohlman MC, Schweickert WD, Pohlman AS, et al. Feasibility of physical and occupational therapy beginning from initiation of mechanical ventilation. Crit Care Med 2010;38(11):2089–2094.

2. Bourdin G, Barbier J, Burle JF, et al. The feasibility of early physical activity in intensive care unit patients: a prospective observational one-center study. Respir Care 2010;55(4):400–407.

P92 Early rehabilitation in critically ill patients: a prospective multicenter assessment of real-life practice

Correspondence: Lorraine Ducroux -

Annals of Intensive Care 2016, 6(Suppl 1):P92

Introduction Early mobilization of critically ill patients is a major issue when it comes to ICU stay shortening for multiple organ failure patients. It is well accepted that “bed-in-chair” sitting, first chair sitting exercise, passive verticalization and walking are crucial steps to better rehabilitation in the ICU and after patients have left the ICU. We aimed at assessing the real-life conditions performance of early rehabilitation practices in ICUs and to study why there are some limitations to a large use of such training for patients.

Materials and methods A prospective telephone inquiry was performed from February to March 2015 in 35 ICU (25 from university hospitals), representing 1046 beds; it recorded data on ICUs and on the real-life practice of rehabilitation in the setting of 5 clinical cases (1 coma patient undergoing mechanical ventilation weaning yet he was on low tapering NE infusion, 1 ARDS, 1 COPD under noninvasive ventilation, 1 acute/chronic liver failure patient infused tapering doses of norepinephrine, 1 postoperative patient). For each case, staff was questioned on the practice of transferring from flatbed to chair; bed in armchair positioning; transferring to edge of bed; standing position; and marching. In addition, we collected the reasons for not practicing early mobilization.

Results One hundred percentage of responses were made by physiotherapists, 64.5 % of which had less than 5-year working experience. The median number of patients taken in care by one physiotherapist was 14 [range (3.5–52)], whereas simultaneously each nurse had 2.5 patients in charge, and each nurse auxiliary 4.2. Shift hours were on 12 h a-day in 58.8 % of the ICUs (Table 23).

Table 23 See text for description

The main reasons for not proposing early rehabilitation are low consciousness level, NE infusion, curare infusion and the simultaneous presence of many surgical drains in the same patient.

Conclusion Despite low sedation, many critically ill patients are readily installed in a bed in armchair position because this position does not include danger and allows easy visual control. Postoperative patients with complicated surgical devices are not often proposed for early mobilization from bed: nurses and physiotherapists bring up the problem of understaffing. Patients with low doses (or tapering low doses) of NE are infrequently proposed for early mobilization; although there are no reliable data concerning the dose–effect relationship with negative side effects in such a population, a specific study is deserved to understand whether such patients would benefit from such rehabilitation in terms of shorter ICU stay and outcome.

Competing interests None.

P93 Extracorporeal membrane oxygenation for pheochromocytoma-induced cardiogenic shock: report of nine cases

Correspondence: Guillaume Hékimian -

Annals of Intensive Care 2016, 6(Suppl 1):P93

Introduction Pheochromocytoma is a rare catecholamine-producing tumor that might occasionally provoke a stress-induced takotsubo-like cardiomyopathy and severe cardiogenic shock. Since the recourse to veno-arterial extracorporeal membrane oxygenation (VA-ECMO) has rarely been reported in this setting, we reviewed the presentation and outcomes of patients admitted to our ICU for pheochromocytoma-induced refractory cardiogenic shock that was rescued by VA-ECMO.

Patients and methods All cases of pheochromocytoma-induced refractory cardiogenic shock managed with VA-ECMO in our ICU between January 2007 and March 2015 were retrospectively reviewed.

Results Nine patients were included, 7 women and two men, aged 31–51 years (median: 43, IQR [36–49]). None of them were previously known to suffer pheochromocytoma. The medical history of these patients was unremarkable except for 4 of them: one had isolated hypertension, one had chronic headache, one had a type 1 multiple endocrine neoplasia syndrome and another reported clinical symptoms suggestive of the disease in the last 12