Proceedings of Réanimation 2017, the French Intensive Care Society International Congress

© The Author(s) 2017. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. P144 Post traumatic cerebral thrombophlebitis: prospective study about 15 cases Chtara Kamilia, Kais Regaieg, Olfa Turki, Najeh Baccouch, Hedi Chelly, Mabrouk Bahloul, Mounir Bouaziz Réanimation polyvalente, Faculté de médecine de Sfax, Sfax, Tunisia; Réanimation polyvalente, CHU Habib Bourguiba, Sfax, Tunisia Correspondence: Chtara Kamilia ‐ kamilia.chtaraelaoud@gmail.com Annals of Intensive Care 2017, 7(Suppl 1):P144

Introduction The study of the bacterial cartography in thoracic surgery is extremely important for the treatment of post-operative infections due to the severity of the underlying pathology, the fragility of patients after surgery in addition to the choice of the empiric antibiotic therapy.

Materials and methods
We led a prospective study following all the patients who underwent a pulmonary resection surgery for a period of 7 months from January to July 2016, jointly with the microbiology department, CHU Ibn Rochd, Casablanca. The bronchial secretions were collected by a protected distal bronchial sample using a (Combicath) after the intubation.
Results During the period of the study, 92 patients underwent a pulmonary resection, 65% for a neoplastic pathology. The medium age was 43 years ±8 and 58% of our sample were male. 48% of our patients had smoking habits and 16 of them had pulmonary tuberculosis, 12 had repeated respiratory infections. The antibiotics used in pre-operative: 58% of beta-lactams; 22% of fluoroquinolones; 5% of macrolides. Moreover, 60% of our patients were classified ASA1.
Of the 92 obtained samples, 22 were positive (23.9%). The most frequently observed germs were the Acinetobacter baumannii (8.7%), Pseudomonas aeruginosa (6.5%), Klebsiella pneumoniae (4.3%), Staphylococcus aureus (4.3%). The Acinetobacter baumannii was the most resistant germ (60% sensibility to carbapenem). These patients were followed until their D30 after surgery, 12 of them developed a post-operative pneumonitis with 4 cases of multi-resistant Acinetobacter Baumanii, 2 of which deceased. Conclusion Pneumonitis after pulmonary resection are common and severe that's why it is necessary to establish a global prevention strategy mainly based on general patricians and pneumologists' awareness concerning the choice of the prescribed antibiotics, in order to avoid the spread of multi-resistant germs.
Introduction Carbapenemase-producing Enterobacteriaceae (CPEc) are increasingly reported worldwide and constitutes a real challenge antibiotic for clinicians to preserve the bacterial ecology. Its incidence has remarkably increased in our intensive care unit during the last 5 years. The ESBL spread has a major consequence in term of antibiotic choices. Carbapenem antibiotic are regarded as the most effective treatment. However numbers of authors suggest that alternatives antibiotics (i.e. noncarbapenems) could be used in ESBL-PE infections. There are some conflicting data regarding the use of alternatives in case of ESBL-PE infections. Moreover as far as we know, there are no data in ICU.
Objectives the aim of this study was to describe ESBL-PE infections in ICU and therapeutic options chosen in these specific situations. Patients and methods Prospective multicentric observational cohort study conducted in volunteers ICU. All consecutive patients hospitalized in ICU with ESBL-PE infection according to CDC definitions were included. Severity of illness was defines according to bone criteria, SAPS II and SOFA. Demographic datas, empirical and definitive antibiotic therapy (ET and DT), clinical evolution, and outcome were recorded. In vitro antimicrobial susceptibility testing was performed by the disk diffusion method or the Vitek 2 system according to the guidelines of the Antibiogram Committee of the French Microbiologic society.
Introduction Bacterial resistance to antibiotics is a common problem worldwide. In South America, this prevalence is reported to be the highest in the world. However, in French Guyana, there is no data on the epidemiology of colonization and infection caused by extended spectrum B-lactamase producing enterobacteriaceae (ESBL-PE). We conducted this study to investigate the prevalence of colonization with ESBL-PE and subsequent ICU acquired infection in French Guiana. Introduction The implementation of hemofiltration (HF) as a renal replacement therapy in septic shock patients requires the supply of large quantities of replacement solutions. These solutions are either industrially prepared in autoclaved expensive plastic bags (conventional hemofiltration, CHF) or continuously provided in unlimited amounts at the dialysis machine directly from the water treatment plant to form the replacing solutions (on-line hemofiltration, OLHF).The aim of our study was to evaluate the safety and effectiveness of on-line hemofiltration compared to conventional hemofiltration in septic shock patients.

Patients and methods
The investigative protocol was approved by the Institutional Ethics Authorities and all patients or their legally authorized representatives provided written informed consent. It was a prospective, randomized, clinical study, including septic shock patients with acute renal failure. Patients were randomized to receive either on-line hemofiltration (n = 8) or conventional hemofiltration (n = 25) for renal replacement therapy during 4 days. Hemodynamic monitoring was conducted by conventional devises, including: electrocardiogram and a radial arterial catheter for invasive arterial pressure every 6 h during period study. We collected serum samples also every 6 h (urea, potassium and sodium levels, troponin, hemoglobin, platelets, C-reactive protein and lactates).
Results The evolution of heart rate (HR), mean arterial pressure (MAP), biological markers were comparable between the two groups over time except a significant decrease in MAP in the OLHF group compared to CHF group only at H6 (P = 0.008) and H12 (P = 0.015) and a significant decrease in C-reactive protein level in the OLHF group at H48 (P = 0.02). Conclusion On-line hemofiltration seems to be a safe and reliable method of renal replacement therapy in septic shock patients. It may be associated with attenuated pro-inflammatory cytokine profile (C-reactive protein).

Competing interests
None.
Introduction Therapeutic plasma exchange (TPE) is crucial for the management of auto-immune diseases like thrombotic thrombocytopenic purpura or myasthenia gravis. TPE is performed either by centrifugation, with specific machines which are not routinely available in ICUs, or by using specific plasma separation membranes with widely spread in ICUs hemofiltration machines. Regional citrate anticoagulation for TPE is well established with centrifugation but has been seldom described for membrane TPE. We are reporting the experience of our ICU in this field. Patients and methods Retrospective study including all patients who received TPE with citrate regional anticoagulation between 2013 and 2016 in an 18-bed ICU. TPE is performed solely in the ICU in our institution.
Results 26 patients were included. TPE was required for thrombotic microangiopathy (13 patients), vasculitis (6 patients), hyperviscosity syndrome (2 patients), Guillain-Barré syndrome (2 cases) and others (3 patients). Mean SAPS2score was 32 [standard deviation (SD) 16.6]. 281 TPE were performed, with a mean number of 10.5 (SD 11.5; range 2-57) TPE per patients. Coagulation of the circuit of TPE occurred in 10 (38%) patients. Coagulation of the circuit occurred in 10.3% (29/281) of the TPE. Minor adverse events have been reported in two patients: one had a rash during the first TPE (no recurrence during the 56 next TPEs) and the other had paresthesia during the first two TPEs (the calcium infusion was increased and there had been no recurrence during the 25 next TPEs). No serious adverse events related to citrate were observed. Conclusion Regional anticoagulation with citrate allowed us to perform TPE in 26 patients, without significant adverse events. The rate of circuit coagulation was 10.3% per TPE.

Competing interests
None.
Introduction A reduced incidence of membrane thrombosis after injection of Anti-thrombin (AT) has been reported in septic patients with acquired deficit in AT undergoing continuous hemofiltration. As this strategy was routinely performed in our unit until 2012, we investigated its cost-effectiveness.
Patients and methods Data about the use of hemofiltration, the consumption of AT and hemofiltration devices during 2011 (period with routine use of AT) and 2012 (period with use of AT only if a membrane thrombosis occurred) were extracted from the administrative database of the institution. A decisional tree was built to modelize the impact of AT on the consumption of hemofiltration devices and blood products. The decisional tree took into account the probability of membrane thrombosis with and without AT and the probability of transfusion after membrane thrombosis. Costs were obtained from the pharmacy of the institution (AT, hemofiltration devices) and from the literature (blood products). Results During 2011, 77 days of hemofiltration were performed, with the use of 45 doses of AT (23,202€) and 76 hemofiltration devices (11,632€). During 2012, 76 (−1%) days of hemofiltration were performed, with the use of 5 (−89%) doses of AT (2578€) and 85 (+10%) hemofiltration devices (13,443€). The mean cost of 1 day of hemofiltration decreased from 449€ to 211€ with the diminution of the use of AT. According to the decisional tree, AT was almost never cost-effective. The only circumstances associated with a benefit for the use of AT was the association of a probability of thrombosis with AT inferior to 0.1, of a probability of thrombosis without AT equal 1, of a probability of transfusion after thrombosis equal 1 and a cost of transfusion of 424€. In these extremely favorable circumstances, AT could decrease the daily cost of hemofiltration of 2.22-19.30€. Discussion The model has several limits: the losses of utility related to transfusion and to interruption of hemofiltration due to thrombosis were not taken into account; the cost of AT measurement was not estimated; the work load of changing a membrane and of transfusion after membrane thrombosis was not analyzed. Conclusion Our results suggest that anti-thrombin is not costeffective to reduce the costs of hemofiltration related to membrane thrombosis.

Competing interests
None.
Introduction In Intensive Care Unit (ICU), some patients suffering from acute kidney injury need renal replacement therapy (RRT). It requires the circuit anticoagulation, this could be done by a regional citrate method. Today, this is a recommended approach for the everyday care, even if the technique isn't widespread yet [1]. The ionized calcemia dosing through the filter ("post-filter" ionized-calcemia) is used to monitor the technique efficacy, with a target of 0.25-0.35 mmol/L showing a good filter anticoagulation. The objective of our study was the assessment of efficacy and safety of our regional citrate anticoagulation protocol, with a less restrictive post-filter ionized calcemia target (0.3-0.6 mmol/L). The main goal was the analysis of the circuit lifespan, considering a lifespan above 24 h, as well as the search of some clinical and biological factors affecting the technique efficacy. Moreover, we analyzed the side effects incidence of the protocol (hypernatremia, metabolic alcalosis), and their consequences. The study received the scientific ethical agreement of University Hospital of Toulouse, and is registered with number 18-0214. Patients and methods 57 patients, admitted to one of the two University Hospital ICUs of Toulouse, needing a continuous RRT method, without any need for systemic heparin anticoagulation, and without severe hepatocellular failure, were included in the study. 103 filters included over a 1-year period were analyzed. Results Results show a mean filter lifespan of 48 h, with a lifespan above 24 h for 85.4% of all filters. Coagulation was the cessation reason for 29.1% of filters, most of them before 24 h of the filter use. A value of post-filter ionized calcemia at day 1 below 0.54 mmol/L was the main factor influencing a filter lifespan above 24 h. An age older than 51 and a SAPS II severity score below 80 were other factors conditioning a filter lifespan of more than 24 h. Side effects of citrate were rare and didn't have any clinical impact among our patients. Discussion These results suggest that citrate used for anticoagulation in RRT could have an additional anti inflammatory effect through the induced hypocalcemia, as well as an energetic gain which could lead to a renal protection against ischemia-reperfusion mechanism [2]. Moreover, these results call into question the need of post-filter ionized calcemia dosing for the monitoring of citrate anticoagulation efficacy, since the method safety is monitored by the total-to-ionized calcium ratio. Conclusion During continuous RRT in ICU, a regional citrate anticoagulation protocol with a non-restrictive post-filter ionized calcemia target seems to be efficient and could reduce side effects. These results need to be confirmed with a randomised control study.
Introduction Continuous veno-venous haemofiltration (CVVH) is used to treat acute kidney injury in critically ill patients. To optimize its efficiency, CVVH requires effective anticoagulation. Systemic anticoagulation with standard heparin, the most used, can lead to major bleeding complications. Hemofilters that are able to adsorb heparin molecules on their surface such as AN69ST and oXiris membranes represent an alternative. The objective of this study was to compare these two types of filters in terms of duration, efficiency, dysfunctions and cost. Materials and methods From October 2012 to May 2014, we conducted a retrospective, observational, and non-interventional study. All patients admitted in the intensive care unit needing CVVH were included. The primary endpoint was the filter lifespan: AN69ST versus oXiris. The secondary endpoint was the filter efficiency (urea reduction ratio: URR). The main analysis did not consider the anticoagulation type. We conducted a subgroup analysis taking into account the use or not of an anticoagulation. Results 181 sessions in 93 patients were carried out using 386 filters representing 10,706 h of treatment. The mean AN69ST filter lifespan was 27 ± 20 h and 28 ± 22 h for oXiris filters (p > 0.05). There is no significant difference in terms of duration between the two filters. The subgroup analysis taking into consideration the use or not of anticoagulation did not show any difference either. The mean URR was 48 ± 23% in the AN69ST group and 44 ± 25% in the oXiris group (p > 0.05). Concerning the dysfunctions, there were no significant difference between the two filters. One hundred and seventy-six AN69ST filters were used for a total cost of 24,288 euros. Two hundred and ten oXiris filters were used for a total cost of 39,060 euros. Conclusion The AN69ST and oXiris lifespans are not significantly different. They were as efficient in terms of blood epuration and had as many dysfunctions. The use of an oXiris filter rather than an AN69ST to extend the circuit's lifespan in the same clinical conditions is not justified considering the extra cost generated.
Introduction Because oliguria is a poor prognostic sign in patients with acute renal failure (ARF), diuretics are often used to increase urine output in patients with or at risk of ARF. From a pathophysiological point of view there are several reasons to expect that loop diuretics could have a beneficial effect on renal function. However, a review of literature shows that the use of loop diuretics in patients with ARF has been associated with inconclusive results despite the theoretical benefits [1].
To assess the adjunctive effect of diuretics, to alter the progression to kidney injury or failure, in patients at risk for acute renal failure. Patients and methods This is a retrospective chart review of consecutive patients who developed ARF with oliguria in the intensive care unit. Chart abstractors were well trained residents. Two chart reviewers (senior intensivists) studied all the charts. An explicit protocol was used to precise all needed definitions. Uniform handling of data was ensured especially for conflicting, missing or unknown data. Oliguria was defined as urine output lower than 0.5 ml/kg/h for at least 3 h. RIFLE score was assessed before and after urinary output normalisation. Therapeutic intervention to optimize pre-renal perfusion was described. Mean arterial blood pressure (MBP) before and after therapeutic initiation, oliguria duration, delay from oliguria onset to diuretic administration, delay from diuretic administration to urinary output normalisation were measured. Results 23 patients were studied over a 2 years period.  3] h. The delay from diuretic administration to urinary output normalization was 3 [0. 5,27] h. After resumption of diuresis, RIFLE score was assessed as (patients without risk, 74%; R, 17%; I, 8%; F, 1% L, zero; E, zero) (Fig. 5). Increased serum creatinine level, above 1.5 fold normal range, was observed only in 6 (26%) patients. Conclusion Rapid optimization of pre-renal hemodynamic disturbances associated with short delay administration of diuretics could significantly alter the progression to kidney injury or failure in at risk acute renal failure ICU patients.

Introduction
The ventilator associated pneumonia (VAP) is a common and severe complication of assisted ventilation. It's the leading cause of nosocomial infections in intensive care unit and remain responsible for a high morbidity and mortality because of the emergence of multidrug resistant (MDR) bacterial agent such us Acinetobacter baumannii (AB). The aim of this study was to determine the incidence, risk factors and prognosis of AB VAP. Patients and methods Retrospective study extending over a 5 year period (January 2010-January 2016) that included all patients over Patients were divided into two groups: one consisting of patients who developed VAP to AB and the second developed VAP to another bacterial pathogen. Results One hundred and forty patients developed VAP. The incidence rate of AB VAP was 15.3% with a density of incidence of 20.3 per 1000 ventilator days. Age, male gender, the time between hospitalization and mechanical ventilation and the medical pathology were risk factors for developing AB VAP. AB was resistant to ceftazidime in 100%, to imipenem in 65%, tobramycin in 70% and netilmycin in 35.3%, rifampin in 85% with a sensitivity to colistin in 100% of cases. The resistance of this germ to imipenem increased from 35% in 2010 to 88.5% in 2016. The evolution of patients with AB VAP developed frequently septic shock compared to other patients (44 vs 19.3%; p = 0.038). The AB VAP mortality was higher (50 vs 33%; p = 0.03).
Conclusion The increasing incidence of multi-drug resistant AB VAP is responsible for a high morbidity and mortality. So we need to identify risk factors and to strengthen the means of prevention of hand contamination and cross transmission during invasive procedures.
Introduction Central line associated bloodstream infections (CLABSI) are among the serious hospital-acquired infections. The aim of this study is to determine the incidence of CLABSI, the pathogens and the risk factors that play a role in the development of BSI among patients followed in a Tunisian medical intensive care unit. Patients and methods All patients admitted for more than 48 h were included in the study over a 1-year period in an 8-bed medical ICU. The enrollment was based on clinical and laboratory diagnosis of BSI. Blood samples were collected from catheter hub of all patients for culture, followed by identification and antibiotic sensitivity testing of the isolates. was higher compared with the mean rate of CLABSI in ICU reported by the NNIS system surveillance for 2004, which is 3.9/1000 catheter.days [1]. Duration of catheterization, frequent manipulation of catheter, catheter location, catheter type, underlying diseases, suppression of immune system, and types of fluids administered through the catheter are significant risk factors in development of BSIs [2]. In our study both duration of catheterization and number of attempts are independent factors for CLABSI. Conclusion In a monocenter cohort, CLABSI had a moderate density rate but are associated with poor outcome. Identifying the risk factors is necessary to find solutions for this major health problem.
Introduction According to some studies, field-intubated patients have 1.5-3 times greater risk of ventilator associated pneumonia (VAP). Endobronchial intubation (EI) can be unrecognized by the physicians and may result in complications such as atelectasis which in turn could increase the risk of VAP. The aim of our study was to confirm this hypothesis. Patients and methods This monocentric retrospective study included all consecutive patients >18 years who underwent an out-of-hospital tracheal intubation before their admission to the intensive care unit (ICU) between January 2012 and December 2015. Exclusion criteria were suspected aspiration or pneumonia on admission, patients who died within the first 5 days of ICU stay, extubation in less than 48 h and underlying disease making radiological interpretation difficult for VAP diagnosis. VAP were divided into early onset (<7 days) and late onset (≥7 days) events and were independently diagnosed by two experienced intensivists who had no access to the initial chest X-ray performed to check the position of the tracheal tube, based on the Clinical Pulmonary Infection Score. Onset of ventilator associated tracheobronchitis (VAT) was also noted. Inadvertent endobronchial intubation was determined by another independent physician based on the interpretation of admission chest X-ray.
Results 397 patients were intubated out-of-hospital. Of the 284 patients excluded, 104 had an extubation in less than 48 h, 114 were died within the first 5 days, 22 had a suspicion of pneumonia, 28 a suspicion of aspiration and 8 an underlying disease making radiological interpretation difficult. Of the 121 patients included, 28 (23.1%) had an EI upon admission. No significant difference was observed between the EI and non-EI group for gender, age, SAPS2, comorbidities and diagnostic category (cardiorespiratory arrest, trauma, coma and cardiorespiratory failure). Early-onset VAP were diagnosed in 43% in the EI group and in 29% of non-EI patients (p = 0.085). Adding early onset VAT, the respiratory infection rate was 61% in the EI group and 44% in the non-EI group (p = 0.061) (Fig. 6). Late-onset VAP were observed in 8.6% in the non-EI group and 7.1% in the EI group, without difference between groups (p = 0.403). There was no inter-group difference in the duration of ventilation, duration of ICU stay and ICU mortality. Staphyloccocus aureus was the most prevalent pathogen in patients with early-onset VAP (23.1%, only one strain was methicillin-resistant). Conclusion This study found a high rate of inadvertent prehospital endobronchial intubation with a higher incidence of early-onset VAP. These results support the implementation of specific procedures to decrease the incidence of EI.
Introduction Ventilator-associated pneumonia (VAP) is associated with increased hospital stay and high morbidity and mortality in critically ill patients. The classic dichotomy between early and late onset VAP is no longer helpful available. The aims of this study were to determine the incidence of multidrug-resistant pathogens in the first episodes of VAP and to assess potential differences in bacterial profiles of subjects with early-onset versus late-onset VAP.
Patients and methods Retrospective cohort study over a period of 18 months including all patients who had a first episode of VAP confirmed by positive culture. Subjects were distributed into 2 groups according to the number of intubation days: early-onset VAP (<5 days) or late-onset VAP (≥5 days).The primary endpoint was the nature of causative pathogens and their resistance profiles. Results Sixty patients were included, 29 men and 31 women. The average age was 38 ± 16 years. The IGS 2 at admission was 40.5 [32; 44] APACHE 19 [15; 22]. Monomicrobial infections were diagnosed in of 46 patients (77%).Two different bacteria were isolated in 14 cases (13%).
A. baumannii was the most frequently isolated in 53% (n = 32) of patients; followed by P. aeruginosa in 37% (n = 22), Enterobacteriaceae in 28% (n = 17) and S. aureus in 5% (n = 3). The isolated bacteria were multidrug-resistant in most cases (58/60). The VAP group comprised 36 episodes (60%) of early-onset VAP and 24 episodes (40%) of late-onset VAP. A. baumannii was isolated in 47% of early VAP (n = 17) versus 62% of late VAP (n = 15) (p = NS), P. aeruginosa in 36% of early VAP (n = 13) versus 37% of late VAP (n = 9) (p = NS) and Enterobacteriaceae in 30% of early VAP (n = 11) versus 25% of late VAP (n = 6) (p = NS). For the resistance profile of the different pathogens isolated, there was no difference between early and late onset VAP. Conclusion According to new data from the literature, there were no microbiological differences in the prevalence of potential multidrugresistant pathogens or in their resistance profiles associated with early-onset versus late-onset VAP.

Introduction
The bacterial nosocomial infection is a major cause of morbidity and mortality in burned. The bacterial ecology in an ICU has a major impact in terms of morbidity and mortality, particularly in the center of burned or length of stay of patients is increased compared to a general intensive care.

Materials and methods
We conducted an observational study spread over 7 months in ICU for severe burned burnt including any who have spent more than 48 h with nosocomial infection (modified CDC criteria), and in which all biological and bacteriological samples were taken. The different types of infections studied were: skin, urinary, lung and bloodstream infections. They excluded all patients belatedly supported or having stayed in other healthcare facilities. Results One hundred twenty (120) patients showed nosocomial infection during this period. The sex ratio (M/F) was 1.7 and the mean age was 39 ± 23 years. Bacteremia was present in 44.84% of cases, followed by the urinary tract infection that was present in 21.21% of cases, followed by the cutaneous infection in 10.30% of cases, and last pulmonary infection in 9% of cases. Infection was polymicrobial in 14.5% of cases. The main bacteria identified were: Acinetobacter baumanii (43.45%) of which 74% is resistant to imipenem, Enterobacteriaceae (31.5%), Pseudomonas aeruginosa (24%) of which 83.25% is resistant to ceftazidime and 68.2% is resistant to imipenem, Enterococcus (16%) and Staphylococcus Aureus (14.29%).
Conclusion The incidence of nosocomial infection is very high compared to literature. The rate of resistance to common antibiotics is very high. A drastic management of antibiotics in our context, the selection of patients and the frequent use in the operating room for skincare allow a better management of these patients.
Introduction Acinetobacter baumannii (AB) ventilator-associated pneumonia (VAP) is common in critically ill patients. The aims of this study were to describing the epidemiological characteristics of AB-VAP, to identify risk factors for acquisition and factors predictive of a poor outcome. Materials and methods A retrospective-prospective study was conducted at the Medical Intensive Care Unit of the University Hospital Ibn Sina, Rabat-Morocco from January 2013 to December 2015. They were included in the study that all patients developed VAP with identified germ. For identification of risk factors of acquisition of AB VAP, two groups of patients were compared: patients with AB VAP versus patients with VAP caused by other germs. To identify factors associated with mortality, two other groups were compared: Survivors versus died.
Results 122 patients presented VAP among which 60 were caused by Acinetobacter baumannii. Among isolates of AB, 8.3% were drug susceptible, and 16.7% were multidrug-resistant while 75% were extensively drug-resistant. They were Independent risk factors for acquisition of AB VAP in multivariate analysis: the presence of a central venous catheter before the occurrence of VAP, duration of prior hospitalization ≥4 days and ICU duration of stay ≥5 days. The mortality rate of AB VAP was 85%. The independent risk factors for poor outcome in multivariate analysis were: duration of antibiotic treatment >7 days, the reintubation and the presence of a previous hospitalization. Discussion Our data were similar to those of the literature with a high incidence of VAP due to the AB (49%) and a high rate of resistance to this bacterium particularly to carbapenems. However, and compared to the literature, the VAP AB were responsible for a death rate much higher (85%). Conclusion Our data were similar to those of the literature with a high incidence of VAP due to the AB (49%) and a high rate of resistance to this bacterium particularly to carbapenems. However, and compared to the literature, the VAP AB were responsible for a death rate much higher (85%).
Introduction Ventilator-associated pneumonia (VAP) is common in critically-ill patients. In fact, 10-20% of patients requiring invasive mechanical ventilation develop this complication. The onset of VAP has been reported to be associated with increased mortality. However, data related to critically-ill elderly patients are scarce. The aim of this study is to assess the prognostic impact of VAP in critically-ill elderly patients.
Patients and methods Mono-center, retrospective study conducted from 01/012014 to 30/09/2015. All old patients (age ≥65 years) requiring mechanical ventilation were included. Two groups were compared: Patients who developed VAP (VAP (+) group) and those who did not develop VAP (VAP (−) group).
increased health-care costs. Among pathogens responsible of VAP, Acinetobacter baumannii which is characterized by its ability to spread in the hospital environment and to acquire resistance leading sometimes to therapeutic impasses is associated with a particularly high mortality reaching 30-75%.
Objective To describe the epidemiological characteristics of A. baumannii VAP, to determine their prognosis and identify factors associated with mortality. Patients and methods It is a monocentric observational study conducted over a period of 13 years in a Tunisian intensive care unit (ICU) including mechanical ventilated patients for more than 48 h with confirmed A. baumannii VAP.
Results One hundred and twenty-three patients were included in the study. A. baumannii was responsible for 31% of VAP in our ICU. The VAP were late in 59% of cases. More than 90% of isolates pathogens were resistant to ticarcillin, piperacillin, piperacillintazobactam, ceftazidime and ciprofloxacin. Sixty percent of germs were sensitive to imipenem. Resistance to imipenem has increased consistently from 0% at the beginning of the study to 88% in 2015.
All pathogens were susceptible to colistin. A. baumannii VAP was complicated by septic shock in 63% of cases. The median duration of mechanical ventilation and of ICU stay were 17 (IQR: [11][12][13][14][15][16][17][18][19][20][21][22][23][24][25] and 25 days (IQR: 17-41) respectively. The use of parenteral nutrition was the only factor associated with the occurrence of A. baumannii VAP resistant to imipenem (odds ratio 2.27, 95% CI [1.07-4.80], p = 0.033). ICU mortality was 45%. It was higher in patients with A. baumannii VAP resistant to imipenem (55 vs 39%, p > 0.05). In the multivariate analysis, the age, the use of renal replacement therapy and the occurrence of VAP relapse have been identified as factors associated with mortality. Conclusion A. baumannii resistance to imipenem became threatening. The use of parenteral nutrition was the only factor associated with the occurrence of A. baumannii VAP resistant to imipenem. The choice of empiric antimicrobial for VAP caused by this pathogen must take in consideration the epidemiologic data of each country and each ICU.
A. baumannii VAP was associated with high mortality. The age, the use of renal replacement therapy and the occurrence of VAP relapse have been identified as predictive of poor outcome.

Competing interests
None.

P182
Admission in intensive care unit for severe adverse drug event: what finding? Julien Arcizet 1 , Bertrand Leroy 1 , Caroline Abdulmalack 2 , Catherine Renzullo 1 , Maël Hamet 2 , Jean-Marc Doise 2 , Jérôme Coutet 1 Introduction Adverse drug events (ADE) remain a serious public health problem. They represent between 0.16 and 15.7% of hospital admissions and between 0.37 and 27.4% of intensive care unit (ICU) admissions. They are defined as any injury related to a drug, and include both adverse drug reactions, expected or not, but also underuse, overuse and misuse, unintended or undesired, preventable or not. Indeed, mortality from iatrogenic event would rise between 2.0 and 28.1%, whereas these ADE that resulted in ICU hospitalization could be prevented in 17.5-85.7% of cases. These unplanned admissions overload ICU, limit access to health care for other patients and have serious economic consequences for the health system. It is therefore necessary to study these ADE to know their main causes and attempt to find a solution to avoid them.
The main objectives of our study were to clinically and pharmaceutically analyze and stratify the different ADE leading to hospitalization in our ICU.

Patients and methods
This is a monocentric prospective study, between June 2014 to January 2016, in medico-surgery ICU. From all admissions, we had included patients admitted in our hospital for involuntary ADE (plausible, likely and very likely causal). We had collected clinical aspects (Failure mode, IGSII score, mortality in ICU) and pharmaceutical aspect (number of drug, offending drugs) at daily medical staff meeting. Conclusion Hospitalizations in ICU for ADE are still too common despite their preventability for most cases. Many patients with known cognitive disorder manage their treatment themselves and this is probably one of the reasons of iatrogenic events. Anticoagulants and antiplatelet agents, by side effects, misuse, underuse or overuse are very often involved. The onset of kidney failure from dehydration and the continuation of nephrotoxic and antidiabetic treatment also remain one of the most common causes. Consequently, it is necessary to continue and develop primary, secondary and tertiary prevention strategies to prevent their appearance, to limit their consequences and to reduce recidivism.

Results
Introduction Intensive care unit (ICU) is usually identified as a place of acute care, concentrated over a short period. For many reasons, a prolonged stay in the ICU has a pejorative connotation for the intensivist physician. The aim of our study is to describe the epidemiological, clinical, paraclinical profile of patients hospitalized for a long time in ICU (over 15 days) and to identify the main prognostic factors and those that can predict the duration of stay in ICU.

Patients and methods
We conducted a retrospective study, over a period of 5 years and 6 months (January 2010 to June 2015), enrolling patients whose length of stay was greater than or equal to 15  Introduction Despite an improvement in prognosis of patients with hematologic malignancies for the last decade, mortality of such patients admitted to the intensive care unit (ICU) remains high. Yet, it seems that a first ICU stay does not modify prognosis of the malignancy. Until now, there is no data on readmission in the ICU of such patients and its effect on short and long term prognosis impact. Patients and methods This retrospective, single-center study conducted on a 5 years period in the medical ICU from our university hospital included 265 patients with hematological malignancies admitted for a first stay. Objectives were to evaluate the ICU, day 28 and 6 months mortality, to identify prognostic factors associated with mortality within uni-and multivariate analysis, to evaluate readmission rate within the 60 days after discharge, to indentify the admission risk factors associated with ICU readmission and the prognosis factors associated with mortality during the second ICU stay.  multivariate analysis poor performance status,  IGS II, HLH, MV and anti-fungal administration were associated  with increased ICU mortality, infections with Pseudomonas were  associated with higher day 28 mortality. Catheter related infections were associated with better ICU survival and CR was associated with lower day 28 mortality. 38 of 132 (28.9%) candidate patients for ICU readmission after a first stay were readmitted within the 60 days following discharge. Median overall survival was lower in readmitted versus non readmitted patients. 6 months mortality was 73.8% for readmitted versus 13.8% for no readmitted patients (p < 0.0001). The second ICU stay mortality was 60.5% and 6 month mortality was 78.9%. By multivariate analysis, only MV was associated with prognosis. The 6 months mortality rate of patients who survived to the second ICU stay was significantly higher than the patients who survived to the first admission but were not readmitted (46.7 vs 13.8%, p = 0.0007).

Results
Conclusion Main features, short and long term mortality and prognostic factors associated with ICU admission are in lines with previous studies. Early readmission rate was high with a negative impact on survival. Despite admission in the ICU of patients with hematologic malignancies seems not to affect long term prognosis, early readmission seems to have a pejorative impact on the course of the malignancy.
Introduction Lung cancer is among all types of cancer, the most common solid tumour admitted in intensive care [1]. Recent studies showed that the prognosis of patients with lung cancer during intensive care unit (ICU) stay has improved [2]. The aim of our study was to determine the causes of ICU admission of lung cancer patients, their prognosis and to identify factors predicting hospital mortality and survival after hospital discharge. In fact, temporary full-code ICU management in patients with relapsed AML seems to be appropriate. None of the life-sustaining interventions at admission and on day 3 were able to predict survival. An ICU trial of 3 days might not be enough to appraise precisely the outcome. Bone marrow transplant was associated with a high mortality in our study.

Patients and methods
In case of relapsed AML with BMT, ICU management is still challenging.

Introduction
The growing population of chronically critically-ill patients has a poor prognosis despite all the resources mobilised [1]. Our primary objective was to analyse the prognostic value of different definitions used to describe them. Our secondary objective was to look for early clinical and biological factors that could be associated with the in-hospital mortality.

Patients and methods
We conducted an epidemiological prospective study in 3 intensive care units (neurosurgical, cardiosurgical and medical) of a large French teaching hospital (Henri Mondor, Créteil). We included all the patients hospitalized for at least 7 days. We tested 5 definitions: the prolonged mechanical ventilation, the definition taken up by Kahn et al. [2], the prolonged length of stay, the persistent critical illness and the persistent inflammation-immunosuppression and catabolism syndrome. Two biological examinations were performed: upon entering the study and 1 week later. The study endpoint was the in-hospital mortality. Results Thirty patients were included between April and July 2016. Among them, only 40% matched the definition of prolonged mechanical ventilation, which is still the most used in the literature. Further, it was not associated with the mortality, but the prolonged length of stay was, with 59% of these patients, that did not survive to their hospital stay. Other parameters that were significantly different between the patients who died and those who survived were an advanced age, an elevated IGS II score at hospital admission, an elevated SOFA score at study entry, a late healthcare-associated infection and several biological variables: a high C reactive protein, low albumin and prealbumin and a poor percent of monocytes expressing HLA-DR, all measured at day 7.
Conclusion The in-hospital mortality of chronically critically-ill is still high. A prolonged length of stay is the only definition who may be helpful to identify the patients with the poorest outcome. Among the early factors associated with mortality, we found a late healthcareassociated infection and a low percent of monocytes expressing HLA-DR, pointing to the value of studying the immune system of these patients.

Competing interests
Introduction As a result of demographic transition, the proportion of «very elderly» (≥90 years) patients is increasing worldwide and more of these patients are nowadays admitted to intensive care units (ICU).
Among physicians the discussion about appropriateness of these ICU admissions still remains controversial mostly due to questionable outcome, limited resources and costs. The aim of the study was to determine and evaluate the clinical characteristics and outcome in a very old population admitted to a medical ICU in an urban teaching hospital.

Patients and methods
We present here a monocentric, retrospective and observational study. We reviewed the charts of all patients (≥90 years) admitted to a medical ICU between 2000 and 2015 (16 years). We collected epidemiological, clinical and biological parameters and all therapeutic measures during the ICU stay. A longterm survival follow-up was also performed. Two hundred eighty-four patients were included for statistical analysis. Multivariate Cox regression was also performed to identify risk factors for 28-day outcome. Results A total of 284 patients were included, which represented 1.8% of admissions to the ICU during the period of the study. The mean age was 92.6 ± 2.1 years, the sex ratio was 0.41. Most of patients (41%) were admitted from the Emergency Department. 20% of these admitted patients suffered of previous dementia. The mean Charlson comorbidity score was 7.7 ± 1.7 and the mean McCabe score was 1.33 ± 0.5. The admission diagnosis in the ICU was mainly respiratory distress (51%), septic shock (11%), cardiac arrest (10%) and coma (8%). The mean SAPS-II score within 24 h of ICU admission was 55.9 ± 21.7.
Half of these patients required support by mechanical ventilation (mean duration 7.3 days) and vasoactive drugs and 6% of patients received renal replacement. ICU and in-hospital mortality rates were 38 and 44% respectively. Overall survival at 6 months after hospital discharge was 33%. Multivariate regression revealed necessity of catecholamines and mechanical ventilation as independent risk factors and urinary sepsis as protective factor for 28-day outcome.
In fine, for 34% of these patients, a limitation of active treatment was decided (on average after 2 days of stay). For all others there was no justification for limiting care because of a well-established treatment plan (with family, GP, ICU team). Conclusion The proportion of elderly patients remains low, but they are increasingly being treated in intensive care units. Nevertheless, the in-hospital mortality is high compared to the average mortality in our ICU over the same period (20%). The prognosis is often not as poor as initially perceived by physicians. The indication for ICU treatment in our study was mostly justified; in the setting of consistent patient care and good clinical practice. It remains therefore appropriate to discuss every single ICU admission of elderly patients without any restriction related to age. Thus, the ongoing cluster-randomized trial of ICU admissions for the elderly patients (ICE-CUB 2 study) is deeply awaited to confirm or not these results [1].
Introduction Regardless of the route of delivery, the postpartum hemorrhage (PPH) is defined as blood loss ≥500 ml after childbirth, and severe PPH as blood loss ≥1000 ml. PPH is the leading cause of maternal mortality in Africa. The aim of this prospective study was to assess the quality of the initial management of PPH in Algeria in Oran EHU and to determine the factors of care with the severity of this complication.

Patients and methods
We conducted a prospective cohort study between April 2014 and September 2014 at the EHU ORAN. All women who delivered vaginally and showed HPP including the suspected cause was uterine atony were included. The severe PPH was defined as bleeding that required invasive surgical treatment (hysterectomy, arterial ligation), a transfusion, a transfer to an intensive care unit or death of the patient. The quality of care was evaluated using objective criteria defined by a delay of diagnosis and care and mortality. Results Among the 466 women who delivered vaginally during the study period, 23 had a PPH, link with uterine atony alleged at diagnosis, 18 of which presented signs of severity. In 41% of cases, the delay in diagnosis of PPH was less than 30 min; 70% of women received oxytocin within 10 min after diagnosis. The tranexanique acid was used in 1 case. The examination of the cervix, uterine exploration and uterine massage was performed in 67, 99 and 97%, respectively. The failure of first line treatment involved 24% of patients. Among them, the time between the diagnosis of PPH and administration of blood derivatives was greater than 1 h in a third of cases. The administration of oxytocin delay exceeds 10 min multiplied by 2.5 the risk of severe PPH. However we had 2 deaths in our series. Discussion In our study the optimal period of care was not adequate, obtaining blood derivatives in our institution remains among the factors aggravating Among the main risk factors for PPH, uterine atony was the main source of complication. Bleeding postpartum aggravated in our two patients has led to the deaths from late diagnosis and care that was not optimal. These hemorrhages PP is the leading cause of mortality: 21% of obstetric deaths (25% in the confidential survey 1996-1997) [1]. A hysterectomy was indicated after failure to conservative treatment. The death rate is estimated at 8% following a disorder complicated hemostasis of disseminated intravascular coagulation (DIC). In some series, the mortality rate is estimated between 2 and 4% [2]. Conclusion The management of PPH in obstetrics gynecology service The EHU Oran was not optimal. The issue of timing of diagnosis and initial treatment is crucial. Solutions must be sought locally to ensure the administration of essential medicines in time, especially the injection of oxytocin within 10 min after diagnosis.
Introduction Chronic obstructive pulmonary disease (COPD) is a common pathology that would represent the third cause of death worldwide by 2020. Its evolution is interspersed with episodes of acute exacerbations (AECOPD) that may indicate an admission in intensive care unit in the most.
Objective To study the evolution of management modalities of patients admitted in our intensive care unit for AECOPD, to determine their prognosis and to identify factors associated with mortality. Patients and methods It is a retrospective, monocentric study, performed in a Tunisian intensive care unit (ICU) over a period of 10 years. We including all patients admitted in ICU for AECOPD. Parameters collected were demographic features, comorbidities, regular treatment, dyspnea assessed by the MRC scale, initial clinical severity reflected by SAPS II and APACHE II scores, modalities and ICU admission deadlines, initial arterial blood gas analysis, management of patients in the ICU (ventilation modalities, prescription of antibiotics, use of vasoactive drugs) and their outcomes (incidence of nosocomial infections and their sites, length of stay and ICU mortality). Results A total of 512 patients, which represents 17.5% of all hospitalizations, with mean age of 72 years (IQR: 66-77) were admitted for AECOPD during the study period. The mean SAPS II and APACHE II were respectively 32 (IQR: 24-45) and 18 (IQR: 14-24). Of these, 60% were ventilated with NIV whose overall failure rate was 48% with a significant decrease between the beginning and the end of the study (94 vs 31% p = 0.001). Sixty-four percent of patients received antibiotics at admission. The prescription rate of antibiotics has decreased significantly over the years from 82 to 36%. The incidence of nosocomial infections was 18%. It remained steady between 11 and 27%. Their sites were pulmonary in 83% of cases. ICU mortality was 16%. In multivariate analysis, ICU admission deadlines, NIV failure and the use of vasoactive drugs were identified as factors associated with mortality. Conclusion Our study showed the importance of AECOPD in the activity of our ICU. The management of these patients has evolved over the years, which was reflected by the significant decrease in the prescription of antibiotics and the enhancement of NIV success rate. This result could be attributed to the combination of several factors: precocious management of patients, experience of the healthcare team and the use of efficient ventilators. ICU admission deadlines, NIV failure and the use of vasoactive drugs were identified as factors associated with mortality.
Introduction Aim. Investigate the effect of music therapy on the tolerance of non-invasive ventilation (NIV) during its introduction. Currently, 50% of the trauma are intubated. Thirty-three percent of the patient admitted in intensive care suffers from acute respiratory distress syndrome (ARDS). The FMHS chose oxygen concentrator as oxygen source in addition to oxygen pressurized bottles. Their supply can be uncertain in conflict areas. Insufficient data are available concerning the use of oxygen concentrator in intensive care unit. The primary endpoint was to determine over the total duration of oxygen therapy, the number of days on which the use of pressurized oxygen was needed for patients oxygenated by oxygen concentrator. The secondary endpoints were to identify when pressurized oxygen was needed, describe the characteristics of the population with oxygen therapy and estimate the oxygen quantity economised thanks to the use of oxygen concentrator.

Materials and methods
The study took place in the forward surgical unit of Bouffard. It's a French role 3 located in Djibouti Republic in Africa. All patients over 15 admitted in the intensive care and needing oxygen therapy were included. All the patients were oxygenated with an oxygen concentrator. The oxygen concentrators used were Sequaltm Integra 10 OM, that could deliver up to 10 l/min of normobaric oxygen. The ventilator used were Pulmonetictm LTV 1000 and 1200. Results Thirty-six patients were included over the 6 months' study period. Sixty percent of the patients were men with an average age of 38  Two hundred and fifty-one days represents the total number of days of oxygen therapy divided into 142 days of invasive ventilation, 15 days of noninvasive ventilation and 94 days of oxygen mask. The use of pressurized oxygen was necessary 19 times over the 251 days of oxygen therapy which represents 7.5% of the total time. The causes of its use were in ten cases (52.6%) criteria of severe ARDS, in six cases an emergency intubation and in three cases a transfer. One dysfunction of an oxygen concentrator happened during our study. The oxygen concentrator produced 1024 m3 of oxygen over the study period, which represents 104 oxygen pressurized bottles of 50 litres. This enabled an economy of 10,000 euros. Conclusion It is safe to use oxygen concentrator to take care of critically ill patients in limited resources environment. The use of pressurized oxygen is still compulsory in two situations: in case of electricity failure and in case of high FiO 2 (above 60%). Oxygen concentrators are sufficient in 92.5% of the time. They enable to deliver oxygen any time which is essential when supply is uncertain in conflict areas.

Competing interests
None.  Table 4). For the same MV and level of OFR, FDO2 was In our experiment, with an OFR of 5L/min, when IFR = 18 L/min (MV = 6 L/min and Ti/Ttot = 0.33), the FDO2 is equal to 41% (±1%) (see Table 5). To this value of IFR, the FDO2 is in accordance with the formula of ATS, but when IFR increase beyond 18 L/min, the FDO2 decrease and the formula is not in accordance with ATS. This can be explain because during inspiratory phase, air room (Fractional oxygen = 0.21) entry in airway mixes with OFR (FO2 = 1), which modifies the FDO2. In this case, when IFR increase then FDO2 decrease and vice versa. Medical and paramedical staff must be aware that with patients who receive OFR by nasal cannula, any change of OFR and/or inspiratory flow changes the FDO2. In this case, for maintain the same FDO2, it is necessary that modify the value of OFR.

Introduction
The actual FiO 2 delivered under oxygen mask in patients with acute respiratory failure and the factors that may influence the FiO 2 are poorly known. In clinical practice, different methods including formula or conversion tables based on oxygen flow can be used to estimate delivered FiO 2 . We aimed to assess first the factors influencing measured values of FiO 2 , and second the best method to estimate FiO 2 in patients breathing under oxygen mask.

Patients and methods
We included ICU patients admitted for acute hypoxemic respiratory failure from a previous prospective trial [1] in whom FiO 2 was measured under oxygen mask using a portable oxygen analyzer. We collected demographic variables and respiratory parameters that may influence measured FiO 2 . Low FiO 2 was defined according to the median measured FiO 2 . For each patient, measured FiO 2 was compared to "Calc + 3%" formula (FiO 2 = oxygen flow in liters per minute × 0.03 + 0.21) to "Calc + 4%" formula (FiO 2 = oxygen flow in liters per minute × 0.04 + 0.21), and to a conversion table [2]. A ± 10% limit of agreement for each estimation method was arbitrarily considered acceptable. Results Among the 265 patients included, median measured FiO 2 was 65% [60-73]. After adjustment on oxygen flow, the three variables independently associated with low measured FiO 2 using multivariate analysis were patient's height, a low PaCO 2 , and a respiratory rate greater than 30 breaths/min. Using paired analysis, each estimation methods differed significantly from measured FiO 2 (p < 0.0001 for each). Values outside the limits  Introduction Acute hyperglycemia is common in intensive care. It was associated with poor prognosis and increased mortality. The purpose of our study is to investigate the frequency of hyperglycemia in our ICU, to determine the main causes of high blood sugar and to analyze the impact of this hyperglycemia.

Patients and methods
Our study is prospective during 3 months. It was conducted in the intensive care unit of the University Hospital Habib Bourguiba Sfax-Tunisia. Were included in our study all patients admitted to the service during the period of the study. For each patient included were collected from the ICU admission, clinical and biological data. Results During the study period, 194 patients were hospitalized in our ICU and the diagnosis of hyperglycemia (>8 mmol/l) was admitted in 93 patients (48%). The comparison between patients who developed hyperglycemia and those free hyperglycemia group showed that, the patients of the first group were significantly older (p < 0.001). Additionally, hyperglycemic patients had more medical history including history of diabetes (p < 0.001), a higher SAPS II (p < 0.05), a more significant frequency of active infections (p < 0.05). Moreover, the presence of hyperglycemia was associated with shock (p < 0.05) and respiratory distress (p < 0.05). Their evolution was marked by the significantly higher frequency of infectious complications (p < 0.05), thromboembolic complications (p < 0.05) and acute renal failure (p < 0.05). The average duration of mechanical ventilation and the length of stay were also significantly prolonged in hyperglycemia group patients (p < 0.05 for both). Finally, the presence of hyperglycemia was significantly associated with a higher mortality rate. Conclusion We concluded that hyperglycemia is correlated with poor prognosis of morbidity and mortality. But strict glycemic control remain controversial. Thus, further studies on this subject will be recommended to define the exact place of glycemic control in intensive care.

Competing interests
None. The RRT was prophylactic in four cases started when phophatemia was more than 2 mmol/L, and therapeutic for renal failure and established TLS in three cases. The median duration stay in ICU was 5 [4][5][6][7] j. Thirteen patients left the ICU without major metabolic dysfunction. Two patients deceased due to infectious complications. Discussion Monitoring of electrolytes was done on average, three times a day which is hard to do in onco-hematology unit. The early use of rasburicase and the aggressive IV hydration helped to prevent TLS for seven patients. The aggressive IV hydration was made according to echocardiography data and close monitoring of vital signs and urine output which has allowed to avoid volume overload and acute pulmonary edema. The early prophylactic RRT prevented renal failure and metabolic complications. Conclusion Early management of TLS in ICU can prevent TLS and most of its serious complications and should be considered in TLS prophylaxis recommendations.

Competing interests
None.

P202
The Both urinary (expressed as the ratio of NGAL on urinary creatinine) and plasma NGAL were predictive of AKI Stage 1. Predictive value of plasmatic measurements was higher than the urinary one (AUC of 0.627 and 0.758, respectively, p = 0.0273 between AUC), but not higher than either baseline serum creatinine (AUC = 0.737) or 24 h diuresis (AUC = 0.735). Backward multivariate regression showed that plasma NGAL concentration was associated with serum creatinine, CRP and albumin, whereas urinary NGAL was associated with leucocyturia and baseline creatinine. Discussion Previous positive studies with NGAL did not compare the performance of this costly biomarker with simple usual clinical parameters to predict AKI. Moreover, several parameters were associated with NGAL concentrations with a high risk of collinearity (CRP) and/or false positive results (leucocyturia).

Conclusion
Our data do not support any added value of NGAL concentration over baseline serum creatinine or urine output to predict AKI.
Introduction Acute renal failure (ARF) is a common entity in intensive care, concern that the heavy morbidity and mortality it is associated [1]. Early diagnosis of this entity remains difficult, neither diuresis and creatinine are early parameters in the diagnosis of ARF. The kidney is an organ that suffers long to become faulty, the priority is to recognize renal aggression and to achieve a therapeutic allowing reversibility of the infringement. A number of markers have been developed for the diagnosis of the IRA but costs remain high not allowing their routine use. The measurement of resistance index with the renal Doppler could be a solution for the diagnosis of aggression and also of the etiology. The elevation of creatinine was seen later within 48 h after the IR >90. Discussion In our series the resistance index has a value of early diagnosis of renal prognosis aggression in the occurrence and development of renal failure. Renal doppler associated with a strictly applied standardized protocol achieves the two goals of monitoring who aid in the diagnosis and guide treatment. Although the recommendations of experts to this tool provides that it should probably not use the resistance index measured by renal doppler to diagnose or treat an IRA (Grade 2) [2]. Identifying the cause of kidney aggression is a prerequisite before any therapeutic action. Hypovolemia and soda hydro overload are the causes principales. Excess filling hyper intra thoracic pressure and hypoxia are the main causes of kidney congestion. Conclusion Doppler is an early renal medium in the diagnosis of renal aggression. A larger series could assert this observation.

Competing interests
None.  ), had significantly more pre-eclampsia, 4/5 (80%) versus 7/33 (21%) p = 0.0187. PE were started at an average of 1.6 days after foetal extraction, and with an average of 5 sessions. Patients of the PE group had significantly lower nadir of hemoglobin but also lower hemoglobin level at day 2 and day 5. Nadir of platelets count was also lower and level remain lower at days 1, 2, 3 and 5. Acute kidney injury (using Kdigo classification) was more frequent with a higher rate of dialysis in ICU, in the PE group (4/5 (80%) vs 3/33 (9%) p = 0.0022) with a more frequent need for dialysis at the exit of ICU. Proteinuria was significantly higher in the PE group (862.8 mg/mmol vs 308.41 mg/mmol, p = 0.0134). ADAMTS13 dosage was done only in patients with PE. We find a diminution of ADAMTS13 activity (before PE) with an average of 53% [40-70] in this group. There was no death, and adverse effects were not significantly different. Discussion This study shows that PE was used when diagnosis was uncertain in the most severe form of PP-TMA. Low hemoglobin, low platelets, acute kidney injury and high level of proteinuria are the main factors associated with the decision to begin PE. This technique was safe and not associated with major adverse events. Several studies show that there are physiopathological crossovers between diseases associated with PP-TMA, for example low ADAMTS13 activity in HELLP or mutation in alternative complement pathway which induced HELLP. Moreover, studies and case reports show a benefit of PE in HELLP syndrome. Our study did not find significant difference in adverse events (maybe due to a lack of power), but this is another argument to discuss PE in the management of PP-TMA in severe patients. The main limits of our study are that none of the patients who had a plasmatic exchange had a diagnosis of PTT and that diagnosis tests were not performed in all patients with PP-TMA (complements level, ADAMTS13…). Conclusion PP-TMA treated with PE has lower hemoglobin, lower platelets, higher rate of kidney injury and proteinuria than those treated without PE. No difference were found for adverse events.
Begining of PE should be discussed for management of a PP-TMA without amelioration after foetal extraction.

Competing interests
None. Introduction Diffuse alveolar damage (DAD) is the typical histological feature of acute respiratory distress syndrome (ARDS). However, in a previous study including 356 patients with criteria for ARDS, we found that only 45% of them had DAD at autopsy exanimation [1]. It has been shown that patients with ARDS and DAD on open lung biopsy had higher mortality than those without DAD [2]. Thus, we aimed to identify markers associated with DAD in patients with ARDS.

Patients and methods
We included the 356 patients who met criteria for ARDS at time of death in our large database of clinical autopsies [1]. We assessed the proportion of DAD according to the severity of ARDS including the degree of hypoxemia and the 4 ancillary variables from the Berlin definition: use of high levels of positive endexpiratory pressure (PEEP at least 10 cmH 2 O), radiographic severity (3 or 4 quadrants on chest radiograph), altered respiratory system compliance (≤40 mL/cmH 2 O), and large dead space defined as a corrected expired volume per minute (≥10 L/min).
Results DAD was associated with all the severity markers abovementioned using univariate analysis. After multivariable logistic regression, the three markers independently associated with presence of DAD were the gender with an odds ratio ( Conclusion DAD was significantly more frequent in females. In addition to the severity of hypoxemia, diffuse infiltrates involving the 4 quadrants was a significant marker of DAD. Introduction Ventilation Induced Lung Injury (VILI) is responsible for an increased mortality in ARDS [1]. Mechanical ventilation may trigger an inflammatory response, comprising alveolar macrophage activation and recruitment, which may be specifically, repeatedly and spatially assessed by functional imaging techniques such as Positron Emission Tomography combined with Computerized Tomography (PET/CT) [2]. 11C-PK11195 is a PET radiotracer with potential to quantify macrophage inflammation. We aim to assess its performance to detect lung macrophage recruitment in an experimental highvolume VILI model. Materials and methods VILI was performed in 5 anesthetized pigs under neuromuscular blockade by rapidly increasing the tidal volume (Vt) to obtain a transpulmonary pressure (TPP) between 35 and 40 cmH 2 O under zero end-expiratory pressure. PET/CT acquisitions were performed before (T1) and after 4 h of high-volume ventilation (T2), and image-derived measurements were realized on the whole lungs, and regionally on 6 distinct lung regions (divided along the anteroposterior and the cephalocaudal axes). 11C-PK11195lung uptake was estimated using the Standardized Uptake Value (SUV), normalized to the CT-derived tissue fraction in the region of interest (ROI). Mechanical lung aggression was estimated by CT-derived dynamic and static strains, and tidal alveolar hyperinflation (expressed as a fraction of the tidal variation in the ROI volume). After euthanasia, alveolar damage and macrophage recruitment were assessed in the 6 lung regions, using semi-quantitative scores.
Results Between T1and T2, VT and TPP significantly increased from 6.0 ± 0.1 to 49.4 ± 2.9 ml/kg and 8.6 ± 1.5 to 37.8 ± 4.4 cmH 2 O, respectively. SUV on the whole lung significantly increased from 1.8 ± 0.6 to 3.0 ± 0.5 between T1and T2 and dynamic strain from 0.36 ± 0 to 2.1 ± 0.2, whereas static strain did not significantly vary. Tidal alveolar hyperinflation significantly increased from 18 ± 4 to 68 ± 3% on the whole lung between T1 and T2. Regionally, dynamic strain, and tidal alveolar hyperinflation significantly differed between regions, as well as between T1 and T2. Regional SUV differed between T1 and T2 but not between regions. Regional static strain did not differ between regions, nor between T1 and T2. In multivariate analysis, regional SUV was independently and significantly associated with dynamic strain and tidal alveolar hyperinflation. Histologic analysis showed significant regional differences in alveolar damage but not in macrophage recruitment. SUV was positively associated with macrophage recruitment but not with alveolar damage. Discussion In this experimental VILI model, 11C-PK11195 SUV was significantly increased after 4 h of injurious ventilation, and was significantly and positively associated with high-volume CT-derived mechanical parameters, such as dynamic strain and tidal alveolar hyperinflation. The radiotracer's specificity for macrophages is confirmed by the SUV significant association with macrophage recruitment and the lack of association with alveolar inflammatory edema.
Conclusion 11C-PK11195is a macrophage-specific PET radiotracer, with potential to dynamically and specifically assess alveolar macrophage inflammation induced by high-volume ventilation.

Competing interests
Research founded by the French society of intensive care medicine (SRLF) and La Fondation pour la Recherche Médicale (DEA20140630499).

Introduction
The reverse triggering (RT) is the term used to name the contractions reflexes of the muscle diaphragmatic provoked ("triggered") by the periodic insufflations, delivered by the ventilator, at sedated patients under mechanical ventilation [1]. The RT constitutes a new form of patient-ventilator interaction clinically difficult to detect and little known. The RT could have potential implications during the management of acute respiratory distress syndrome (ARDS). At present, the management of severe ARDS consists among others, on the use of an early and systematic perfusion of neuromuscular blockade agents (NMBA) during a 48 h' period, continuation to the ACURASYS essay which showed a reduction of the mortality in the group of the severe ARDS patient receiving NMBA. The reason of the beneficial effect of curare is not perfectly known. It is possible that the phenomenon of RT is a mechanism implied in the deleterious role of the mechanical ventilation during ARDS. The abolition of this phenomenon by NMBA could explain the beneficial effect of NMBA in ARDS [2]. The objective was to look for the phenomenon of RT in two groups of ARDS patients: a group receiving NMBA and a group not receiving NMBA. Patients and methods Physiological observational and comparative study in intensive care units. We record continuous signals of airflow, airway pressure, and esophageal pressure during 24 h of consecutives patients with ARDS criteria and PaO 2 /FiO 2 ratio ≤150 at a positive end-expiratory pressure (PEEP) of 5 cmH 2 O evolving for less than 48 h under mechanical ventilation.

Results
Recording of esophageal pressure of 21 consecutives moderate to severe ARDS patients were blinded analyzed (group NMBA n = 11; group unless NMBA n = 10). Any phenomenon of RT was observed in the group of mild ARDS patients receiving NMBA (Fig. 8a).
We confirmed the existence of RT on 3 patients of 10 in the group of mild ARDS who not receiving NMBA (p = 0.05) (Fig 8b). Discussion One of the main limits was the quality of the collection of the signal of esophageal pressure. The monitoring of esophageal pressure is technically difficult, and can d influence the quality of the signal and the reliability of the results. Conclusion This study confirms the existence of the phenomenon of reverse triggering among deeply sedated patients not receiving NMBA with a 30% incidence. More research is needed to determine if the reverse triggering is a risk factor independent from VILI, associated with the bad prognosis of severe SDRA patients and, if a strategy of early treatment based on NMBA, could improve the prognosis of reached patients.  after ECMO removal had a significant median reduction of 3 days in the BIPAP-APRV group, p = 0.03 (Fig. 9).

Discussion
We reported the feasibility of a protocol based on BIPAP-APRV aiming at resuming SV as soon as possible in ARDS patients under ECMO. The occurrence of spontaneous inspiratory efforts in ARDS patients can major variability of transpulmonary pressure and as result jeopardise VT and driving pressure control. This might be an issue if protective ventilation is not guaranteed anymore. VT with BiPAP-APRV remains within safe range when the ratio Fig. 9 Circles are PAC group, rhombus are APRV group. MV mechanical ventilation, PSV pressure support ventilation. Data are presented as median (IQR), Comparison between the 2 groups at each time Mann-Whithney Test, *p < 0.05 of spontaneous minute ventilation to total minute ventilation is between 30 and 40% [1]. BiPAP-APRV is more efficient than PSV to increase lung aeration in patients with ARDS [1]. Recruitment of dependent region is more likely to achieve if SV is not supported by synchronized positive airway pressure as during BiPAP-APRV [2]. Our strategy targeting a percentage of SV between 30 and 40% with high PEEP could be viewed as a compromise in order to promote SV and protective ventilation at the same time.
Conclusion Protective ventilation combined with SV under ECMO by using a specific protocol based on BiPAP-APRV is feasible and safe. It may facilitate weaning and thus reduce the time under MV after ECMO.
To what extend this beneficial effect is directly due to the presence of SV deserve further investigations.
Introduction Since the first transplant from a patient in a state of brain death conducted in 2010 at the University Teaching Hospital IBN RUSHD of Casablanca, the number of transplants has increased. However, it is still inadequate meet the growing needs of organs. The refusal of families remains the main obstacle to the developpement of organ transplantation in Morocco. The aim of our study is to monitor and analyse the evolution of family refusal to organ donation in a brain dead patient. Patients and methods This is a retrospective and comparative study from August until December 2014.The data were collected from records of brain dead patients candidates for organ donation at the 7 intensive care units on IBN RUSHED Hospital. The coordination registers were also studied. A questionnaire was distributed to families who refused organ donation to investigate the causes of the refusal. Results During this period, 79 patients with brain death have been identified and 78 families had been approached. 61 families (78%) refused organ donation. The main causes of refusal were: fear of body mutilation (35%), lack of will (24%) and religious causes in 28% of cases. The refusal rate for families decreased from 85% in 2010 to 50% in 2014. Only 2 patients experienced cardiac arrest before transplantation. During this period, 22 cornea transplants from braindead patient were conducted with 20 kidney transplants and two liver transplants. Discussion The evolution of the refusal of families saw a decline through awareness and communication campaigns for organ donation.
Conclusion Improvements to our health care system must be proposed including strengthening detection of potential donors and relationships with the donor's family and effective communication policy.

Introduction
In the ICU, three major actors are involved in the caring relationship: patient, relatives and caregivers. Acting as spontaneous testimonials of the lived experience, thank-you letters from relatives may be considered by ICU teams as a source of original information which could help in improving care for critically ill patients and families. This study aimed to investigate the qualitative content of thank-you letters from relatives of patients who stayed in the ICU. Specifically, our research questions were, with regards to the letters' content, (1) how is the caring relationship tackled and characterized by relatives? (2) to what extent does this relationship impact their experience of ICU? Materials and methods The study took place in a 16-beds ICU during a 6-month period. The research team consisted in a care assistant, a nurse (also clinical research associate), a psychologist (not working in the ICU) and an intensivist. The corpus consisted in twenty thankyou letters received in the ICU. We conducted a qualitative study according to the thematic inductive approach. The process of coding was intended to create established meaningful patterns.
Results Two main themes emerged as specific determinants of the caring relationship: (1) the temporality, comprising the time dedicated to the patients and their family, the time spent with the ICU team, the striking time corresponding to significant events for relatives needed to be shared with the staff, the extension of the link with caregivers by evocating a new life after ICU stay, the writing time as a countergift to the caregivers; (2) the caregivers behaviour, including human skills detailed in many core values (kindness, availability, devotion, attention, goodwill, sensitivity) psychological support, emotional sharing, capabilities to give informations. Relatives feel to be "at the center of all attention" in the same way as their loved ones. Through the narration of ICU experience, the caring relationship is characterized as follows: (1) the caregiver becomes a close person with an equal relationship (feelings of friendship, emotional closeness); (2) the ICU team becomes a new family (contrasting with the poor living environment of ICUs); (3) the relative becomes a caregiver (with appropriation of medical terms or speaking of his loved one as a patient); (4) the caregiver is seen as a "super-hero" through an asymmetrical relationship with an overstatement of personal dedication and investment of the staff members (abnegation, vocation, involvement). The caring relationship impacts relatives' experience of intensive care in several ways: (1) relatives are deeply touched by caregivers' human behavior, emotional support being a source of solace and resilience in particular for bereaved families; (2) relatives express the idea that taking care of humans is not a valued and rewarded task and the emerging awareness of hospital realities and difficulties of work in the ICU; (3) the most striking transformational change in relatives is the perception of their own vulnerability and humanity, leading them to exhibit an outward-looking attitude (for example filling out their organ-donation card), and encouraging the ICU caregivers to continue their missions for the others. Conclusion Thank-you letters provide both encouraging and informative messages for ICU teams about relational care for patients and families notably the indivisibility of the families and their critically ill loved ones. The relatives' experience of the ICU appears strongly influenced by the caring relationship in the way they express an authentic revelation of their own humanity and altruistic thoughts. The thematic content of thank-you letters questions determinants and fundamental values at stake in the patient-relatives-caregivers relationship.
Introduction Far from medical paternalism, the doctor-patient relationship has now evolved to respect "the autonomy and patients' rights". Changing behavior has been gradual, while the law offered the patient the freedom to consent to care and then of expressing their wishes regarding the therapeutic intensity they would benefit, in critical situations where consent would not be possible, through advance directives (AD) [1]. Their use is of paramount interest for intensivist in many critical situations. Unfortunately, the use of AD remains marginal because of the unfamiliarity of patients with their use and an appropriation default by clinicians [2]. The aim of our study was to investigate the perspective of the coming family physician generation on advances directives. Patients and methods Population of interest was general practitioner fellow (GPF) from class of 2012 to 2014. We built an online questionnaire survey about knowledge and the place they want to give to AD in their forthcoming daily clinical activity. This questionnaire was sent to GPF emails obtained by universities, unions and via the official mailing lists of different regionals classes provided by the first contacted. Descriptive analysis of quantitative data was expressed as mean and standard deviation, qualitative data in number and percentage. The comparison of continuous variables was performed by the Student t-test and the comparison of categorical variables by a Chi2 test. Analyzes were conducted on BiostaTGV website and Microsoft Excel ® . Results 2310 GPF answered the survey, mainly from Ile de France (n = 218), Toulouse (n = 199) and Lille (n = 169). For GPF the majority of patients do not know the AD (81.6%) and 54% think that those who know do not know how to use it. 88.4% of GPF think writing AD by patients requires better information.
According to them, the information should concern the support offered in the ICU (77.1%), the use of mechanical ventilation (58.7%), dialysis (48.2%) and the evolution of patients after hospitalization in ICU (50.6%). Nevertheless information on the prognosis of chronic diseases or organ failure seems interesting for only 37 and 20.7% of them respectively. 89.8% of GPF wish to propose the drafting of AD to their patients. However, only 60.1% of them are willing to suggest AD to patients with cancer or hematologic malignancies, 56.2% to patients with neurological and/or degenerative disorders, 20.5% to elderly patients. Discussion Despite the low proportion of the population we think these observations to be of interest because we probably selected the GPF the most interested in AD as the participation was not mandatory. Conclusion A large majority of young of future general practitioner is willing to be involved in the implementation of AD with their patients, however the target population remains very limited, considering that half of them do not want to discuss AD with patients suffering from diseases potentially associated with ICU admission or therapeutic intensity discussion.  This study was conducted in adult intensive care units in public or private hospitals in four countries: Canada, France, Italy, Spain. In each country, 40 health care professionals were solicited for an exploratory interview about the sources of stress in the work environment: 10 senior physicians, 10 residents, 10 experienced nurses (with more than 2 years of experience in the service) and 10 inexperienced nurses (with less than 2 years of experience in the service). All the interview transcripts were analysed using an inductive coding approach. Results One hundred and sixty professionals (80 physicians and 80 nurses) were included in the study. Eight themes emerged from the analysis, and they concerned the stress linked to (1) patient (2) care, (3) team, (4) family, (5) institutional context, (6) environment, (7) organizational context, (8) individual dimensions. In each theme, sub-themes have been identified and determine more precisely the difficulties at work. Discussion Our findings emphasize the complexity of work in ICUs and show the specifics factors not taken into account in the generic stress scales such as stress in relation with family relationships, the end of life decisions and inequity of health care.
Conclusion The specific stress scale should allow to better identified stress in ICU and to develop measures of prevention and support and training programs.
Introduction Intensive care units (ICU) is a place where caregivers face many constraints that can affect their physical and mental health due to the use of specific care and strong emotional charge related to patient death and pain of the families. The aim of the present study is to detect anxiety disorders and/or depression among staff working in ICUs.

Materials and methods
On September 2016, a questionnaire was distributed to staff (medical and paramedical) operating in 5 ICUs in the university hospital Fattouma Bourguiba Monastir, Tunisia (1 medical ICU, 1 surgical ICU, 2 cardiologic CCUs and 1 nephrologic intermediate care unit). This questionnaire included demographic data of participants (age, sex, marital status, length of service, psychiatric history, consumption of anxiolytic and/or antidepressant) and the Hospital Anxiety and Depression Scale (HAD: scale composed by 14 items to screen the anxiety (A) and/or depression (D) among hospital staff ).
Results During the study period, 106 participants completed the questionnaire (82%), 58% of them were women, the median age was 32 years ± 8.4. Forty-nine participants were doctors (the majority of them residents: 41/49). 12.3% of participants (all paramedics) worked on night shift, seniority of more than a year in the ICU was found in 59% of participants. 53.8% of staff interviewed were married and only 14.2% of them reported consumption of anxiolytics and/or antidepressants. 65.9 and 63.2% of the participants had respectively symptoms suggesting anxiety and depression. The median HAD score was 19 (IQR = 8); The medical function seems to be significantly associated with the occurrence of symptoms of anxiety and depression compared to paramedics, however the type of ICU (medical/surgical ICUs vs cordiologic/nephrologic ICUs) does not appear to be related to the occurrence of symptoms of anxiety or depression (Table 7). Conclusion Anxiety and depression are common symptoms among caregivers in ICUs. Improved conditions of work in these units should be a target to avoid Burn Out Syndrome.

Competing interests
None.

Table 7 Relationship between the type of ICU, the function and the occurrence of symptoms of anxiety or depression in ICUs
Anxiety, n (%) Depression, n (%) Introduction Carbon monoxide (CO) poisoning is one of the common causes of poisoning specially in the cold season, which leads to a significant morbidity and mortality.

Patients and methods
We retrospectively reviewed the medical data of patients who presented to the toxicology emergency department with CO poisoning during January 2015 to March 2016. We analyzed patients' characteristics, management, and outcomes. Results A total of six hundred and sixty-six patients (523 female and 143 male), aged of 34 ± 14 years, were included; poisoning occurred between December and February in 75% of cases, secondary to an indoor heating system exposure in the majority of cases (89%). The estimated duration of exposure was 2.5 ± 2 h [0.5-13 h], with a mean carboxyhaemoglobin (COHb) level on arrival at 18.6 ± 17%. Neurological changes were the most presenting symptoms including headache (n = 652, 98%), dizziness (n = 272, 40%), seizure (n = 15, 2.3%) and loss of consciousness (n = 81, 12.2%). Digestive disorders involving vomiting and nausea were observed in 32.1% (n = 214). One woman without cardiovascular risk factors developed non STsegment elevation myocardial infarction complicated by lung edema. The majority of patients (n = 601, 90%) received normobaric oxygen during 6 h (n = 535) and 12 h (n = 66). Hyperbaric oxygen therapy was administered at 2.5 ATA during 1 h to 65 patients for neurological changes (n = 24), pregnancy (n = 22) and elevated COHb ≥ 25% (n = 19). Mechanical ventilation was required for 5 patients, and admission into intensive care unit in 39 patients (6%). Death occurred in 5 cases (0.7%).
Conclusion The carbon monoxide poisoning is a common reason for emergency department visits in winter. The physician should be aware of the serious neurological and cardiovascular complications, if symptomatic treatment and oxygen therapy regimens were not respected.

Competing interests
None.

P222
Neuro-respiratory toxicity of baclofen in the rat: study of the concentrations/effects relationships and role of GABAergic Introduction Baclofen, a GABA-B receptor agonist is used as muscle relaxant agent and recently for the treatment of alcohol dependence. The number of poisonings has significantly increased since this new indication. Clinical presentation of poisoning mainly includes sedation, hypotonia, respiratory depression and seizures. To characterize the neurorespiratory toxicity of this molecule at high doses, we aimed at investigating alterations in Sprague-Dawley rat ventilation and brain electrical activity after baclofen administration and studied their reversal by GABA-receptor antagonists. Materials and methods Rat ventilation was investigated using plethysmography and arterial blood gas analysis while brain electrical activity was studied using EEG with one implanted frontal electrode. Three baclofen doses were used including 43.5 mg/kg (30% lethal dose-50%), 72.5 mg/kg (50%) and 116 mg/kg (80%). Baclofen concentrations were obtained using HPLC-MSMS assay. We modeled baclofen pharmacokinetics and analyzed the doses/effects and effects/concentrations relationships. Results Baclofen induced early-onset and prolonged dosedependent sedation (p = 0.0002), hypothermia (p = 0.004), EEG and respiratory depression (0.001) characterized by significant increase in the inspiratory (p = 0.0001) and expiratory times (p = 0.02). Significant increase in PaCO 2 and decrease in arterial pH and PaO 2 were observed at 116 mg/kg (p = 0.001), peaking at 240 min. EEG showed signal slowing, burst-suppression aspects and spikes peaking at 5-6 h post-injection without normalization at the end of the experiment at 24 h. We did reverse baclofen-induced decrease in tidal volume with saclofen (a GABA-B receptor antagonist) and interestingly no alteration of baclofen-induced respiratory depression was observed with flumazenil (a GABA-A receptor antagonist). Pharmacokinetic parameters of baclofen were obtained at the three doses and were dose-dependent. Significant but non-linear relationships were observed between baclofen-induced effects and concentrations.
Conclusion Baclofen causes dose-dependent neurorespiratory toxicity in rats. However, due to increased poisonings, its safety profile at high doses remains to be established in humans.

Competing interests
None. Poisoning was deliberate in 100% of cases. Mean ingested dose was 448.5 ± 349 mg. The majority of patients presented to the emergency room at 4.8 ± 5 h after ingestion. Digestive decontamination was performed in 9.6% (n = 2) of patients. Clinical presentation was dominated by neurological symptoms; including coma (n = 15), hypotonia (n = 5), hyporeflexia (n = 5), agitation (n = 5), seizures (n = 3) and delirium in 1 case. Hemodynamic manifestations included bradycardia in 12 patients, three of them required atropine infusion. One patient presented with hypotension responding to vascular resuscitation. Sixteen cases required mechanical ventilation. Aspiration pneumonia was noted in 6 cases. Mean duration of ventilation was 36.6 h ± 27. Mean hospital length of stay was 62 h ± 25. Complications included ventilation associated pneumonia in one case and moderate rhabdomyolysis in 3 cases. All patients evolved favorably. There is no correlation between coma and assumed ingested dose. Conclusion Baclofen overdose causes mainly neurological effects and except for bradycardia cardiovascular effects were uncommon. Prognosis is good if full supportive care is administered properly.

Competing interests
None. Introduction The lack of an effective treatment for the maintenance of abstinence from alcohol has led physicians to take an interest in baclofen. Beyond efficacy, safety of baclofen, prescribed in high doses, is a concern, especially in case of drug overdose. Indeed, patients with chronic alcohol abuse frequently develop psychiatric disorders, and are at risk of voluntary drug intoxications. Thus, we set up a retrospective study to describe morbidity and mortality associated with baclofen overdose. Conclusion Baclofen, prescribed in high doses, may lead to severe intoxications: self-poisonings frequently require endotracheal intubation and are associated with an increased risk of death. Dialysis decreases baclofen elimination half-time but clinical relevance of this difference could not be determined.

Competing interests
None. Introduction Baclofen, a GABA-B receptor-agonist with muscle relaxant properties established since 1974, has been recently used at elevated doses to treat dependence to ethanol. The number of prescriptions has exponentially increased without an exact evaluation of its toxicity. We aimed to describe acute baclofen poisoning requiring intensive care unit (ICU) admission and study the relationships between the toxic encephalopathy and the plasma baclofen concentration.

Patients and methods
We conducted a single-centre retrospective study including all baclofen-poisoned patients admitted to the ICU in 2013-2016. When requested by the clinical situation, repeated electroencephalograms and measurements of the plasma baclofen concentrations were performed. Toxic EEG encephalopathy on a scale of zero to five was graded according to the international rating system (Markand, 1984). Plasma baclofen concentration was determined using liquid chromatography coupled to mass spectrometry in tandem developed with a Quantum Ultra apparatus (Thermo Fisher Scientific) and electrospray source ionization in positive mode (limit of quantification: 5 ng/mL). Linear regression and Chi-2 or Mann-Whitney tests were used as requested for subgroup comparisons. Baclofen pharmacokinetics and the relationships between the toxic encephalopathy and the plasma baclofen concentration were modeled using WinNonlin software v. 5 ) were closed to the observed values reported at therapeutic doses. The relationship between baclofeninduced encephalopathy as a function of the baclofen concentrations was described using a sigmoidal Emax model. Conclusion Baclofen poisoning may be life-threatening. Toxic encephalopathy is well-described with EEG and its grade correlated to the baclofen concentration. Prescribers should be aware of the dangers of baclofen which benefits to treat dependence to alcohol are still lacking.

Competing interests
None. Results Initial examination suggested that an illness other than bacterial meningitis was the cause of patients' complaints. First hypothesis was meningitis receiving uncomplete dosage regimen of antibiotics. Thereafter owing to apparent loss of consciousness with abnormal eyes movements, non-tonico-clonic seizures were considered meanwhile. The ratio of individuals less 5 y-o to those equal to and greater was 33/67%. The male to female ratio was 46/53%. The mean duration of hospitalisation was 3.4 ± 0.8 days (extremes 1-10 days). Extrapyramidal syndrome predominant on the upper part of the body was noted by paediatrician neurologists who suggested considering a genetic disease. However, signs and symptoms were present in people from different families in different areas at the same time. The definitive diagnosis made on pictures and videos of children and adults and was facio-troncular dystonia resulting from drug-induced adverse effect. Four urine samples were collected in children and sent to a toxicological laboratory in France. All urine samples were positive for haloperidol meanwhile the other causes of facio-troncular dystonia were excluded, including other neuroleptics, metoclopramide, antidepressants, amodiaquine, anti-histaminic drugs, anti-epileptics, and cocaine. From January to August 2015, 1021 hospitalisations were recorded in 925 patients. Looking for the source of haloperidol showed that tablets sold as 'diazepam' and consumed by symptomatic patients contained haloperidol as the sole active pharmaceutical ingredient, suggesting that this large outbreak was due to haloperidol toxicity from falsified diazepam. Initial treatment was diazepam to relieve severe facio-troncular dystonia which was efficient but resulted in long-lasting sedation more especially in children. A dosage regimen using bipéridène administered by intravenous and oral route was refined to prevent adverse effects related to this anticholinergic agent used in children.

P226
The complete reversal of the facio-troncular dystonia was the antidotal evidence supporting the toxicological diagnostic. The mortality rate was less than 5% meanwhile the direct causal relationship with ADR is questionable. An epidemiological study, including toxicological analysis in controls in ongoing. Indeed, facio-troncular dystonia induced by haloperidol does not result from a drug overdose but is an ADR occurring in about 20% of patients treated with haloperidol. WHO is involved in the inquiry related to this counterfeature involving different countries. The cause of the error is presently under investigation. Discussion This outbreak emphasizes the need to consider toxicity resulting from counterfeatured medicines when facing collective atypical signs and symptoms in countries with unrestricted access to medication with limited control of qualities of the medicinal drugs. Conclusion Counterfeatured medicinal drug may result not only in poor efficacy but also in onset of unexpected outbreak of unknown diseases that should suggest a toxic origin.

Introduction
In late 2014-early 2015, Médecins Sans Frontières (MSF) had to face an outbreak of severe facio-troncular dystonic syndrome (FTDS) in North-East Congo. This outbreak resulted from counterfeature of pills sold as diazepam. Toxicological analysis revealed one pill contained about 10 mg of haloperidol. FTDS induced by haloperidol does not result from a drug overdose but is an adverse drug reaction (ADR) occurring in about 20% of patients treated with haloperidol. Nine-hundred and twenty-five individuals were admitted in MSF structures for 1021 FTDS. The ratio of individuals less than 5 y-o and equal to or greater of age was 33/67%, including 35 (3.4%) of children less than 1 y-o. Initial treatment was based on diazepam which relieved FTDS but resulted in long-lasting sedation, preventing given any drug by the oral route. Owing to the definitive diagnosis, a shift to the use of a more specific antidote was chosen. Biperiden was selected as existing in the intravenous and oral form in the Swiss pharmacopea. The study was approved by the Ethical Committee of the Ministery of Health of the Republic Democratic du Congo. Patients and methods As a whole, biperiden was used in 223 cases (84% of the total). Treated children presented with severe dystonia as evidenced by inability to cooperate and to swallow. Verbal informed consent was obtained from relatives. The dosage regimen to treat drug-induced dystonic syndrome in the Swiss pharmacopea is as follows: for parenteral use in children, intravenously or intramuscularly: 0.040 mg/kg or 1.2 mg/m 2 BSA every 30, according to response and tolerance; a maximum of four doses per day should be used. The internal MSF recommendations for biperiden use in children were 0.01-0.05 mg/kg of body weight that might be repeated four times a day. Initially, biperiden administration was administered under medical supervision by the MSF referent at the scene. Results There was no pediatric preparation of biperiden. Accordingly, the adult preparation was used in children. The preparation contained 5 mg of biperiden in one milliter of solvent. The initial planned dose for children of 1 y-o and less and those up to 5 y-o were 1 and 2 mg, respectively. The 5 mg (1 ml) of biperiden was diluted in 4 ml of saline resulting in a final dilution of 1 mg/ml. Six children were treated according this dosage regimen. However, the one 1 mg dose was either of limited efficacy while being associated in others of signs suggestive of ADR, including agitation, heart rate greater than 160 b/ min, the upper limit for children aged of 1 y-o and less. Two children greater than 1 y-o presented severe abnormal behavior resulting in an attempt at escape. Owing to question about safety, the dosage regimen was changed, as follows: 5 mg (1 ml) of biperiden was diluted with 9 ml of saline resulting in a final dilution of 0.5 mg/ml. An initial dose of 0.5 mg was administered intravenously as a bolus dose. The effects were looked for over 15 min. In the absence of improvement in facial dystonia, a second bolus dose of 0.5 mg was administered, a third dose could be considered 15 min later if the FTDS did not resume. The cumulative initial dose should not be greater than 2 mg. In addition to the reversal of facial dystonia, the therapeutic effect of biperiden included the return of swallowing to normal allowing to give further doses of biperiden by the oral route for three days. The first oral dose was administered no less than 12 h after the last initial dose at a dose equal to the efficient initial cumulative dose. The following doses were halved every 12 h. No ADR related to biperiden were reported using this dosage regimen. The mean duration of hospitalisation was 3.4 ± 0.8 days.
Discussion The bioavailability of biperiden by the oral route is equal to 33%. Accordingly, the corresponding intravenous dose should be divided by a factor three. Dosage regimen of anticholinergic drugs in children are poorly documented. The dosage regimen recommended by the pharmacopea resulted in frequent and severe ADR. Titration of biperiden resulted in efficient and safe dosage. Conclusion When biperiden administration is required by intravenous route in children of 5 y-o and less, biperiden should be administered intravenously and titred using bolus dose of 0.5 mg till the therapeutic effect is obtained.
Introduction Severe poisoning by rodenticides is frequent. It represents nearly 30% of patients admitted to the new intensive care unit (ICU) of the region. That is why we decided to perform this study. The aim of this work was to describe the epidemiology, clinical features and management of all patients admitted to our unit for acute poisoning with rodenticides. Patients and methods It was a retrospective study performed in the year 2013 from January to December. The study included all patients admitted in the ICU for rodenticide poisoning.
Results 32 patients were enrolled in the study. Our patients were young with a mean age of 30 ± 2 years. Poisoning was more common in females (n = 24; 75%). The mean delay between rodenticide poisoning and first medical contact was about 2 ± 2 h in the cases where this information. Most of our patients (91%) attended the emergency department of Zaghouan with a non-medical transportation. It was a suicide attempt in most cases (62%) and an accidental poisoning in 32% of patients. The most frequent cause of poisoning in our study was organophosphorus pesticide (n = 24; 75%). The second cause was alpha-chloralose poisoning with seven cases (22%). One patient ingested accidentally an anticoagulant rodenticide. Most of patients had ingested (oral route) the rat poison (n = 23; 78%). Clinical examination found normal vital signs in ten cases (31%). Nine patients (28%) had a shock, eight patients (25%) had an acute metabolic disorder and five patients (16%) had acute respiratory failure or were comatose. All patients enrolled in the study were admitted in the ICU for a period of clinical observation of 24 h. Stomach pumping (gastric lavage) was performed in 30 patients (93%). An antidote which was atropine was needed in twelve patients. Three patients (9%) who ingested alpha-chloralose needed intubation and mechanical ventilation. All patients had a good outcome and were discharged from ICU and from hospital. The mean ICU length of stay was 2 ± 3 days. Conclusion This is the first study of acute poisoning with rodenticides admitted in the new ICU. The results of our study were similar to those published in recent literature. Cases of acute poisoning with rodenticides reported in this work were not severe.

Competing interests
None. Introduction The systemic arterial load imposed to the left ventricle (LV) is a major determinant of normal/abnormal cardiovascular function. The LV mean ejection pressure (LVMEP) is the best estimate of load faced by the LV throughout ejection. The contribution of the steady and pulsatile blood pressure (BP) component of arterial load to LVMEP is debated. We studied the hemodynamic correlates of LVMEP using carotid tonometry. Intensive care unit patients equipped with an indwelling catheter were studied, thus allowing precise calibration of the tonometer. Patients and methods Carotid tonometry (Complior Analyse ® ALAM Medical, France) was prospectively performed on 28 hemodynamically stable, spontaneously breathing patients (12F, mean age ± SD = 64 ± 18 years). Carotid waveforms were calibrated from diastolic BP and time-averaged mean BP invasively obtained at the radial (n = 18) and femoral (n = 10) artery. All patients were free of aortic stenosis. LVMEP was the area under the systolic part of the carotid pressure waveform divided by ejection time.

P229
Results LVMEP (111 ± 17 mmHg) was strongly related to central systolic BP (126 ± 21 mmHg; r 2 = 0.97) and was also related to mean BP (r 2 = 0.82), peripheral systolic BP (r 2 = 0.83), peripheral (r 2 = 0.35) and central (r 2 = 0.50) pulse pressure (each P < 0.05). The LVEMP was not related to age, heart rate and stroke volume. Systolic pulse wave amplification ratio from carotid to periphery was 1.07 ± 0.08. Conclusion LVMEP was most strongly related to central systolic BP, which combines the influences of the steady and pulsatile components of central arterial load (r 2 = 0.97). LVMEP was less strongly related to peripheral systolic BP, which may be less informative given variable systolic pulse wave amplification across patients.
Introduction Myocardial dysfunction is one of the main predictors of poor outcome in septic patients, with mortality rates next to 70%. Many pathological findings were found in the sepsis induced cardiomyopathy including myocardial ischemia, alterations in microcirculation and proinflammatory cytokines. The aim of this study was to assess the prognostic value of a recently developed highly sensitive cardiac troponin I (hsTnI) assay in patients with septic shock.

Patients and methods
We performed a prospective observational study in septic shock ICU patients within 72 h of admission. Exclusion criteria were age >18 years; pregnancy; post-cardiac arrest and braindead. HsTnI was measured soon after admission and 12, 24, 48 and 72 h after. Patients were subjected to transthoracic echocardiography (TTE) at study inclusion and regular biochemical and hemodynamic assessments were performed. Pearson's Chi square and Fisher's Exact tests were used. P < 0.05 was considered significant. Conclusion Circulating hs-cTnI is present in patients with septic shock. A rise of HsTnI may be an indicator of poor outcome. Also, right heart functional abnormalities exist in patients with septic shock.

Competing interests
None.

P231
Evolution of the right distribution width as a pronostic marker during the differents state of shock Introduction Right distribution Width (RDW) has been recently proposed as a pronostic factor in different pathologic situations and especially to the septic patients who stay in ICU. Some works substantiate the relationship between an alteration of the red blood cell rheology during the septic shock and a severe state of the disease. No one has studied the RDW between the differents shocks yet.
We are going to determinate the relationship between RDW and APACHE II score, mortality rate in the Intensive Care Unit (ICU), at the hospital, at the day 30 and 90.

Materials and methods
We investigated those parameters near 228 patients who were admitted at the ICU and needed norepinephrine between the first of March and the 31st of December. They were stratified in différent groups: septic shock n = 101, cardiogenic shock n = 100, hemorragic shock n = 20 and obstructive shock n = 7.
Results We did not observe any correlation between the RDW and the ICU mortality, hospital mortality and at the day 30 and 90. Only a poor significant correlation has been found between the cardiogenic shock and the mortality rate: at the hospital (p = 0.01), at day 30 (p = 0.01) and at the day 90 (p = 0.05) but not in the ICU (p = 0.89). The Receiver Operating Characteristics (ROC) curves do not show significant differences between RDW, APACHE II score and ICU mortality rate or intra hospital. The sample of the hemorrhagic shock and obstructive shock was not usable for this calculation. Compared to other studies which were focused on the septic shock where the mortality was approximately 20%, we determinated a mortality rate near 50%. Conclusion The delta of the RDW D3/D1 did not present any correlation with the mortality rate. In our study, the RDW in the different kind of shocks do not look like to be a good predictive marker of the mortality, except for the patients included in the cardiogenic shock where a poor significant correlation could be highlighted. Conclusion Cardiogenic shock was the most frequent complication of AMI who led to ICU admission, whereas mechanical complications are rare at the era of early coronary reperfusion strategies. In addition to severity score, serum creatinine and cardiogenic shock appeared as independent factors of hospital death.

Competing interests
None. Introduction Pulmonary embolism (PE) in high-risk is a partial or total obliteration of the pulmonary arterial network by a fibrin-clot cruoric more than 50%, the management requires a rapid reduction of pulmonary arterial resistance and right ventricular post load through rapid revascularization by thrombolysis. Our aim is to determine the value of thrombolysis in pulmonary embolism and describe the clinical, paraclinical and outcome pulmonary embolism at high risk. Patients and methods This is a descriptive study of 20 cases of pulmonary embolism at high risk admitted to the cardiology department to CHU Oran between 2008 and 2014. Signs of gravity of (PE) comprising: syncope, circulatory collapse, cardiogenic shock or acute pulmonary sonographic sign of heart. It was confirmed in chest CT. All patients received thrombolysis using the protocol accelerated by two types of molecules: streptokinase or actilyse.

Results
The sex ratio was 0.11; mean age 44 years, ranging from 20 to 80 years; Risk factors were dominated by contraception was 35% and the postoperative 30% the clinical picture was dominated by cardiogenic shock in 75% of cases. 20% cardiovascular collapse and syncope in 5%; Doppler echo all patients had signs of dysfunction of the right ventricle represented by the dilatation of the right cavities and pulmonary hypertension. The CTA found a (PE) bilateral in 70% right in 20%. Thrombolysis using actilyse in 12 patients and streptokinase in 8 cases.
The outcome was favorable in 16 patients; with two cases that are complicated by chronic pulmonary heart and the death of 2 patients with cancer. Discussion The female predominance is explained by the increase of risk factors hormonal contraception, whose first generation combination hormonal. Our patient had a high probability with clinical signs of severity based on the score WELLS [1]. This diagnosis was confirmed by chest CT; which shows the vascular bed obstruction degree with a very good sensitivity and specificity. The suspect patients with severe PE and that presented signs of acute pulmonary heart ultrasound have effectively (PE). The indication of thrombolysis was chosen on hemodynamic criteria; success is found in 80% of patients with improved hemodynamics dice the early hours. This success is explained by the role of thrombolytic in lysis clot to obtain pulmonary arterial revascularization; and reduce pulmonary arterial resistance and the right ventricular afterload which accelerates the healing of right heart failure and improvement of pulmonary capillary volume. The 02 cases who developed a chronic pulmonary heart; it was done immediately a right ventricular dysfunction with pulmonary arterial outset of very high pressures suggestive that the embolism occurred on an already pathological right heart. No cases of massive bleeding were noted in our series. Conclusion Severe pulmonary embolism is burdened with high mortality; diagnosis is based on the stratification of risk score, was facilitated by the non-invasive strategies that articlent around the Doppler echocardiography and CT angiography; thrombolysis can reduce the high mortality related to severe pulmonary embolism.
Introduction Hypertension is a frequent motif for admission to emergencies. The diabetic is increasingly exposed to this risk [1]. The objective of this study is to evaluate the proportion of diabetic patients presenting to the emergency department with high blood pressure (BP) and to identify their epidemiological and clinical characteristics. Introduction Sepsis associated liver dysfunction (SLD) is usually attributed to systemic and/or microcirculatory disturbance. Hypoxic hepatitis, also known as shock liver or ischemic hepatitis, is a life threatening event associated with high morbidity and mortality. Doppler ultrasonography is a non invasive method to measure Doppler hepatic hemodynamic parameters. The primary objective of this study was to assess the accuracy of the hepatic hemodynamic parameters (portal venous blood flow PVBF and resistance index of the hepatic artery HARI) in predicting SLD in septic shock patients. The secondary aims were to identify factors associated with SLD, investigate the effects of volume expansion (VE) on systemic and intrahepatic hemodynamics and to assess the intra-and interoperator reproducibility. We also analyzed 28-day mortality.

Patients and methods
In a prospective design, we included 30 consecutive patients with septic shock (24 males; median age: 36.5 years) admitted to the ICU with septic shock in Charles Nicolle Hospital of Tunis from February to July 2015. All patients were resuscitated following the Surviving Sepsis Campaign guidelines. We measured systemic hemodynamic variables (Mean arterial pressure (MAP), and cardiac index (CI)) and performed hepatic Doppler before and after volume expansion. We measured PVBF and computed the HARI. We recorded the liver function tests (ALT, AST and Bilirubin) for 48 h. SLD was defined as an increase in serum Bilirubin ≥ 20 µmol/l (Hepatic SOFA ≥ 1). Accuracy of the hepatic hemodynamic parameters to predict SLD was measured by the area under the ROC curve. P < 0.05 was taken to indicate statistical significance.

Results
The median SOFA score at T0 was 8 points and the median IGS2 score was 38 points. The sources of infection were as follows: the lungs (n = 19), the abdomen (n = 4) and the urinary tract (n = 3). The incidence of SLD in our cohort was 33.3% (n = 10). There was no significant difference between "SLD group" and "No-SLD group" in all hepatic hemodynamic parameters especially the PVBF and the HARI. Lactate levels were significantly higher in patients with SLD (median 3.55 vs. 0.85 mmol/l). Similarly, the platelet count was significantly lower in the "SLD group" [mean (± SD) 149.2 ± 103.9 (109/l) vs. 242.8 ± 104.1 (109/l); p = 0.039]. There was no difference in duration of mechanical ventilation, ICU length of stay and 28-day mortality between the 2 groups. The PVBF was significantly lower in patients who died before D28 (median: 558 vs. 826 l/min in the survivors; p = 0.014). Volume expansion caused a significant increase in CI, mean hepatic artery velocity and the PVBF. The intra-and interoperator reproducibility was good to excellent for the systolic and mean velocities of the hepatic artery, portal vein diameter and the PVBF. Conclusion Our results don't support the hypothesis that the hepatic sonography is predictive of SLD in septic shock. Our pilot study showed higher lactate levels and hematologic SOFA in SLD group. The PVBF was significantly lower in patients who died before D28. More experience will be necessary to define the ultimate role of Doppler ultrasonography in the evaluation of hepatic perfusion in patients with septic shock.
Introduction Early surgery is the current trend for management of patients with valvular disease. That said many of them, particularly from developing countries, are still operated at a very advanced stage of disease. Despite improvements in myocardial protection and surgical techniques, postoperative care after multiple valve surgery (MVS) for advanced rheumatic heart disease (RHD) remains to be a clinical challenge. We conducted a study to determine postoperative complications and morbidity-mortality risk factors in this subgroup of patients. Results Sixty-two patients were included: 25 with out-of-hospital refractory cardiac arrest and 37 with in-hospital refractory arrest. The initial rhythms was shockable rhythm in 25 (40%) cases. At ECLS initiation, the mean no flow was 0.8 ± 3.9 min and mean low flow (time between the time of refractory cardiac arrest and time at which an ECLS flow was provided) was 71 ± 37 min. The mean ECLS flow rate was 4.17 ± 1.31 L/min. Initial blood test results were: arterial pH = 7.04 ± 0.20 and plasma lactate = 12.8 ± 5.9 mmol/L. Eleven (18%) patients survived (4/30 (13%) acute coronary syndrome, 4/8 (50%) severe poisoning due to drug intoxication, 1/4 (25%) dilated cardiomyopathy, and 2/20 (10%) others). Survival was lower for patients with out-of-hospital refractory cardiac arrest, 2 of 25 (8%), than for patients with in-hospital refractory cardiac arrest, 9 of 37 (24%), respectively, p = 0.17. As expected, out-of-hospital refractory cardiac arrest was associated with a more prolonged low flow (89 ± 40 min vs 60 ± 30 min, p < 0.01) and a more profound acidosis (pH 6.9 ± 0.2 vs 7.1 ± 0.2, p = 0.01 and arterial lactate 14.7 ± 6.1 vs 11 ± 5, p = 0.05).

Patients and methods
In univariate analysis, survival was lower for patient with refractory cardiac arrest unrelated to drug intoxication, 13 vs 50%, respectively, p = 0.03. In addition, mortality was associated with arterial pH (7.01 ± 0.2 vs 7.14 ± 0.15, p = 0.025) and low flow (75 ± 40 vs 55 ± 12 min, p = 0.005 Conclusion In a highly selected group of critically ill patients with refractory cardiac arrest, the potential beneficial effect of ECLS could be due only to its clinical impact on reversible causes of circulatory failure (i.e. severe drug intoxication in our cohort). Further studies are needed to clarify whether the use of ECLS could be considered as a disproportionate tool, specifically in patients with out-of-hospital refractory cardiac arrest due to acute coronary syndrome or associated with prolonged low flow or a profound acidosis.

Competing interests
None.

S2
Post-cardiac arrest shock treated with veno-arterial extracorporeal membrane oxygenation: an observational study and propensity-score analysis Wulfran Bougouin 1 , Nadia Aissaoui 2 , Alain Combes 3     Average time between introduction and removed of the ECD was 50 h (42-56). Among the 16 esogastroduodenoscopy performed, 9 (56%) were strictly normal. Endoscopy showed minor gastric injuries in 6 patients (38%). Within these patients, 3 (19%) also presented minor esophageal injuries. Esogastric injuries characteristics were mostly similar to usual orogastric probe injuries. One patient (6%) experienced a serious ulcerous esophagitis mimicking a peptic esophagitis, not firmly related to the ECD. No patients necessitated hemostatic local procedure and no significant gastrointestinal bleeding was observed. Eight patients (47%) were alive at D90, including 5 patients (29%) with a Cerebral Performance Category score of 1. This compares favorably to outcomes from previous studies. Conclusion ECD seems an interesting and safe semi-invasive method of cooling in OHCA patients treated with 33 °C-TTM. Although it seems slower than more invasive devices to reach 33 °C, ECD was able to strictly maintained the TT within the maintenance phase of TTM. Further studies will be necessary to define the exact place of this new device within the cooling strategy in patients necessitating a precise TTM-strategy.

Competing interests
None. Fig. 10 See text for description Introduction Since post-cardiac arrest care might influence the outcome, characteristics of receiving hospitals should be integrated in survival evaluation of patients transported in hospital. We aimed at assessing the influence of care level center on survival at discharge in a regional registry of out-of-hospital cardiac arrest (OHCA).

Materials and methods
We prospectively collected Utstein and in-hospital data for all non-traumatic OHCA patients, in whom a successful Return of Spontaneous Circulation (ROSC) had been obtained, from a large metropolitan area (Great Paris). Receiving hospitals were categorized in 3 groups (A, B, C) depending on their respective characteristics (annual volumes, 24/7 catheterization availability and temperature management use). We compared patients' characteristics in the 3 groups and performed a multivariable logistic regression using discharge survival at endpoint. Results During the study period (May2011-Dec2013), 1476 patients were admitted in 48 hospitals (917 in group A, 428 in group B and 91 in group C). Overall survival rate at discharge was 433/1436 (30%). Patients' baseline characteristics significantly differed, as hospitals from group A treated younger patients and more frequent shockable rhythms (p < 0.001). Unadjusted survival rate differed significantly among the 3 groups of hospitals (respectively 34, 25 and 15.4% for A, B, C, p < 0.01). However in multivariable analysis, the category of hospital was no longer associated with survival. Conclusion In this population-based study, characteristics of receiving hospitals are not associated with survival rate at discharge. This could result from the strategy used for triage, which aims in matching patients' characteristics and resources.
Introduction Acute kidney injury (AKI) commonly occurs after cardiac arrest and is associated with an increased mortality and a delayed awaking. Early recognition of AKI remains challenging, given that serum creatinine increases belatedly after aggression. Introduction Out-of-hospital cardiac arrests (OHCA) are an absolute urgency and have a very poor prognosis. Pediatric guidelines differ from adult guidelines for cardiac arrest management. Since 2005, adult guidelines apply from the onset of puberty. The main objective was to describe the epidemiological characteristics and outcome of OHCA victims while taking puberty into account. The secondary objective was to determine the prognostic factors for survival at D30. Materials and methods All patients less than 65 years of age, victims of OHCA between July 1, 2011 and September 1, 2015 care by a Mobile Emergency and Resuscitation Service (SMUR) participating in French National Cardiac Arrest registry (RéAC) were included. Patients were split into 3 groups: Prepubescent patients (named "Children": Girls 0-9 years, Boys 0-11 years), Pubescent patients (named "Adolescents": Girls from 10 to 17 years and Boys from 12 to 17 years) and "Adults" (men and women 18-64 years). The "Adolescents" group was consecutively compared to the "Children" group and to the "Adults" group.
Results 644 children, 256 adolescents and 16,566 adults under the age of 65 have been included. OHCA in Adolescents occurred more often on public roads (30%) or in public places (15%) and were more often traumatic (46%) than those in children and adults. Respiratory causes were more frequent in children (27%) than in adolescents (18%) and adults patients (18%). The proportion of shockable rhythm increased with age (3, 7 and 10% for children, adolescents and adults respectively). Survival at D30 was greater in adolescents (12%) than in children (7%) and adults (8%) (p = 0.01 and p = 0.02 respectively). In the 3 studied groups, initial shockable rhythm was a survival factor at D30 (respectively OR 18 [11.39-17.48] for children, adolescents and adults). Other risk factors are described in Table 11. Conclusion Adolescents had better survival at D30 than the 2 others groups. Adolescents and adults had shockable rhythm more often than children. Moreover, respiratory failure was less frequent in adolescent and adults patients compared to children. Puberty seems to be a good limit to differentiate pediatric patients with OHCA.

Competing interests
None. Introduction Non-invasive ventilation (NIV) is an effective alternative to endotracheal mechanical ventilation (MV) in the management of acute respiratory failure (ARF) patients. Nevertheless, it can be still difficult to assess its real feasibility, application and outcome in daily clinical practice. Therefore, we report our clinical experience with routine use of NIV since the last national recommendations (2006). Our aims were to evaluate the clinical efficacy and outcome of NIV, and to identify predictive factors for NIV failure based on a daily use. Patients and methods We conducted an observational retrospective single-center cohort study by reviewing all medical records from January 2006 to December 2013 in our 22-bed medical intensive care unit (ICU). Eligible patients were those having received NIV during their ICU stay. Two groups were defined according to the indication of NIV: NIV for hypoxemic or hypercapnic ARF (ARF-NIV), and NIV used in the post-extubation period for weaning, prevention or treatment of post-extubation ARF (post-extubation NIV).The main evaluation criteria were the incidence of NIV use, success/failure rate of NIV and risk factors for NIV failure in each group. NIV failure was defined as the need for stopping NIV whatever the reason (intubation, intolerance, death) within 3 days after its initiation.  (2; 9), and was longer in the post-extubation NIV group (11 days (6; 19)) than in the ARF-NIV (5 days (3; 9) for hypoxemic ARF, 4 (2; 6) for hypercapnic) (p < 0.001). The overall ICU mortality was 11.0% (15.8% in hypoxemic group, 8.1% in hypercapnic group, and 6.9% in post-extubation NIV group) (p = 0.0005). In multivariate analysis, the main risk factors for ARF-NIV failure were: SAPS II on admission (p < 0.0001), absence of cardiologic history (p = 0.0274) and the cause of ARF (p = 0.0016) with a higher failure rate for pulmonary infections than acute cardiogenic pulmonary edema (OR 2.94, p = 0.0004). For post-extubation NIV, the only independent risk factor for failure was normocapnia before NIV initiation (p = 0.0011). Conclusion Our large longitudinal study demonstrates the feasibility and efficacy of NIV applied in daily clinical practice. Provided it is performed in a suitable environment by an experienced team, NIV should be considered as a first-line ventilatory treatment in various etiologies of ARF and a very useful ventilatory support in the postextubation period. Nevertheless, risk factors for NIV failure should be known by ICU clinicians, hypoxemic ARF remaining the more difficult indication to manage with NIV. 1 Réanimation médicale, Hôpital Saint-Louis, Paris, France; 2 Service de biostatistique et information médicale, Hôpital Saint-Louis, Paris, France; 3 Réanimation, Institut Paoli-Calmettes, Marseille, France; 4 Réanimation  Introduction Acute respiratory failure (ARF) is the leading cause for ICU admission in immunocompromised patients. In these patients, oxygenation strategy is of major interest to avoid the need for mechanical ventilation (MV), which is associated with high mortality rates. In that setting, use of non-invasive ventilation (NIV) and oxygen therapy with High Flow Nasal cannula (HFNC) could be interesting alone or in association, but data about initial ventilation strategy in immunocompromised patients are controversial.

Patients and methods
To assess how initial oxygenation strategy actually influences the risk of MV on the coming day within the three first days of ICU stay. The study end-point was the need for MV on the coming day. We restricted analyses to these first three ICU days given, based on our own experience, most of MV was expected to occur by then.
We performed a post hoc analysis combining three prospective studies of critically ill immunocompromised patients (two randomized control trials, the IVNICTUS and the MINIMAX studies and one prospective cohort, the TRIAL-OH study). We only considered patients with ARF and a delay between ICU admission and study inclusion less than 48 h. We excluded patients who required invasive MV within the first day, those with an ICU stay less than 1 day and those with acute pulmonary edema diagnosis at ICU admission.
In order to estimate and compare the causal effect of daily respiratory management strategy on the probability of intubation in the coming day, we computed inverse probability of treatment weights (IPTW) using propensity-score, defined as the probability of actual treatment selection conditionally on observed covariates. To handle confounding in such dynamic regimens, we considered marginal structural models (MSM), which have been proposed to estimate the causal effect of a time-dependent exposure when time-dependent covariates that can be affected by the previous treatment are present. Two treatment exposure models were considered: NIV versus oxygen therapy regardless the device (model 1) and HFNC alone, NIV alone versus NIV + HFNC versus standard oxygen therapy alone (model 2).
Results 847 patients were included in the study. In model 1, there was no difference between NIV and oxygen groups on MV whatever the landmark time. In model 2, while the unweighted OR for intubation at day 1 was significantly higher in the NIV group (OR 2.05, 95%CI 1.29-3.29) and HFNC group (OR 2.85, 95%CI 1.37-5.67) than those in the standard oxygen alone group, these differences disappeared in the weighted samples. Using MSM, no effect of the oxygenation strategy on MV was found, regardless of the oxygenation devices but the landmark time was associated with a reduced occurrence of MV.
Conclusion We found no evidence of any significant difference from several oxygenation strategies on mechanical ventilation probability during the first 3 days of ICU in a large cohort of immunocompromised patients with ARF.

Competing interests
None. Introduction The role of noninvasive ventilation (NIV) is debated in the management of patients with acute hypoxemic respiratory failure. A recent study showed that patients treated with high-flow nasal cannulae oxygen therapy (HFNC) had lower intubation and mortality rates than those treated by the association of HFNC with NIV (1). High tidal volumes (VT) delivewred with NIV may be associated with an increased risk of intubation (2). We aimed to identify risk factors associated to intubation, in hypoxemic patients with acute respiratory failure and especially the role of VT under NIV. Patients and methods This is an ancillary study from a multicenter, randomized, controlled trial including patients with acute hypoxemic respiratory failure (FLORALI-study). We focused on only patients with moderate or severe hypoxemia (PaO 2 :FiO 2 ratio ≤200 mmHg) and we excluded those with mild hypoxemia. The criteria for intubation were predetermined including worsened or persisted respiratory failure, impairment of neurologic status and hemodynamic instability. Results After adjustment on the oxygenation strategy, the two factors independently associated with intubation were the presence of bilateral pulmonary infiltrates at admission (OR 2. Simulation conditions enables to reproduce its occurence, using different types of tools, from physiological parameters to heart rate variability and psychocognitive tests. Future research is required to evaluate the impact of these parameters on teaching.

Competing interests
None. with stratification by centre and operator experience. An only inclusion criterion was: "Patients must be admitted to an ICU and require mechanical ventilation through an endotracheal tube". Patients were excluded if: contraindication to orotracheal intubation (e.g., unstable spinal lesion); insufficient time to include and randomize the patient (e.g., because of cardiac arrest); age <18 years; pregnant or breastfeeding woman; correctional facility inmate; patient under guardianship; patient without health insurance; refusal of the patient or next of kin to participate in the study; previous enrolment in a clinical randomized trial with intubation as the primary end point (including previous inclusion in the present trial). Post-hoc analysis was performed to assess occurrence of SpO 2 <90% during intubation procedure between 4 groups of preoxygenation: BVM (at a minimum flow of 15 L/min, NIV (100% FiO 2 ), HFNC (at a minimum flow of 60 L/min, with 100% FiO 2 ), and NRM (at a minimum flow of 15 L/min). Between-groups difference in desaturation occurrence was adjusted for baseline covariates significantly associated with the group membership (P < 0.2). Multivariate analysis of the occurrence of a desaturation (<90%) was performed using logistic regression. Bag-valve mask was considered reference.

S12 Preoxygenation for intubation in the ICU: an ancillary study of the MACMAN Trial
Results Baseline characteristics were showed in Table 12. Groups were similar at baseline except for PaO 2 /FiO 2 ratio. In univariate analysis, age (P = 0.08), SAPS2 (P = 0.02), PaO 2 /FiO 2 ratio (P = 0.03),SpO2 (P = 0.001) and method of preoxygenation (P = 0.08) were associated with occurrence of desaturation below 90%. In multivariate analysis, SpO2 at randomization and method of preoxygenation were only predictors of desaturation below 90%. BVM and HRM were associated with similar risk of desaturation occurrence whereas NIV (OR 0.

Fig. 11 See text for description
Introduction Intubation procedure is a challenging issue in intensive care unit (ICU) [1]. Cardiac arrest related to intubation in critically ill adult patients has been poorly studied. The studies were not powered to conclude on this rare outcome [2]. The main objective of our study was to establish the incidence of cardiac arrest and to assess the risk factors of cardiac arrest in a large prospective database of intubation procedures performed in ICU.

Patients and methods
Five prospective studies were included, with similar data collected before, during and after intubation procedures using the same methodology. The primary outcome was the incidence of cardiac arrest related to intubation. The secondary outcomes were the death (cardiac arrest without return of spontaneous circulation (ROSC)), the cardiac arrests with ROSC, the complications related to intubation, the length of ICU stay and the 28-day mortality. The factors associated with cardiac arrest related to intubation procedures were assessed by univariate and multivariate analysis based on patient, provider and practice characteristics. Results Among the 1847 intubation procedures included, 49 cardiac arrests (2.7%) occurred, including 35 with ROSC (71.4%) and 14 without ROSC (28.6%). Main patient, provider, procedure characteristics and outcomes according to cardiac arrest related to intubation are presented in Table 13. In multivariate analysis, the independent predictors of cardiac arrest related to intubation were low systolic blood pressure prior to intubation, hypoxemia prior to intubation, no preoxygenation, overweight or obesity and age >75 years. Mortality rate at day 28 was significantly lower in patients intubated without cardiac arrest (30.1%, 541 of 1798) than with cardiac arrests overall (73.5%, 36 patients of 49, p < 0.001) and cardiac arrest with ROSC (63%, 22 patients of 35, p < 0.001).
Conclusion Cardiac arrest related to intubation in adult ICU is not a rare event occurring in 2.7% of cases with high immediate mortality of 28.6% and at day 28 of 73.5%. We identified five independent risk factors to cardiac arrest which 3 of them could be modifiable. Optimal preparation to intubation procedure could help to prevent those cardiac arrests.
Introduction Naasotracheal intubation (NTI) has been progressively given up in favour of the orotracheal intubation (OTI) in intensive care unit (ICU). This could be explained by more frequent infectious (sinusitis and ventilator associated pneumonia) and non-infectious (epistaxis, turbinates bones injury) complications the former being thought to be more frequent with NTI. However, whereas infectious sinusitis is a risk factor for VAP, no study has yet demonstrated that OTI decreases the infectious sinusitis rate compared with NTI. Furthermore, nasal route could improve patient comfort and decrease auto-extubation. Finally NTI can be performed without laryngoscopy with less risk of lips and dental injury. In this prospective study, we aimed to compare the complication of NTI and OTI and to assess the comfort of the patient.

Patients and methods
We performed a prospective observational study in a 12-bed medical ICU including patients requiring endotracheal intubation. The intubation route was let at the discretion of the physician in care of the patient, however OTI was compulsory in case of cardiac arrest, severe hypoxemia (P/F < 150 when available) and clotting perturbation. For each patient, age, sex, SAPSII, mechanical ventilation duration. intubation route were recorded as well as complications during the placement of endotracheal tube. Infectious and non infectious complications during invasive ventilation period were also recorded. In patients who were successfully extubated, pain, burning feeling, dryness and the wish of tube removal were assessed using visual analogic scales (VAS  Conclusion Despite its small size, and the absence of randomization, the present study suggests that nasotracheal intubation improves the comfort and the tolerance of tracheal intubation and is not associated to higher rates of VAP.

Competing interests
None.

S15
Effect of mode of hydrocortisone administration in patients with septic shock: a prospective randomized trial Oussama Jaoued 1 , Rim Gharbi 1 , Najla  The baseline characteristics of patients were similar between the two groups. Sepsis was secondary to community-acquired infection in 54% of cases. There was no difference in mortality between groups (74% in continuous groups and 59% in discontinuous group). SOFA score was significantly higher at days 1, 2 and 5 in discontinuous group. Length of stay, duration of mechanical ventilation, number of day without vasopressors, and the occurrence of adverse events were similar in the two groups.
Conclusion The mode of hydrocortisone administration in patients with septic shock has no influence on morbidity or mortality. The occurrence of adverse events was similar.
Introduction Widespread activation of coagulation with platelet consumption is a pathophysiological feature of severe sepsis and septic shock. Thrombocytopenia, either defined by platelet count below 150 G/L or by a significant relative 20-40-percent decrease in platelet count is a potent poor prognostic factor in sepsis. Besides their role in hemostasis, platelets also carry various immune and inflammatory functions that are likely to impact on host defense against infections. We aimed to assess whether changes in the platelet count induced by sepsis is associated with the development of subsequent nosocomial infections. Patients and methods Patients were obtained from two prospective studies about immuno monitoring of dendritic cells and innate-like lymphocytes in critically ill septic patients (1,2). Adult patients with severe sepsis and septic shock were included. Exclusion criteria were any immunosuppressive condition (hematological malignancy, HIV infection at any stage, bone marrow or solid organ transplantation, daily corticosteroid therapy >0.5 mg/kg prednisone-equivalent, chemotherapy or any other immunosuppressive treatments), pregnancy, do-not-resuscitate orders on admission. In addition patients who died or who received platelet transfusion during the first week after ICU admission were also excluded. Platelet counts were collected on the day of sepsis diagnosis (D1) and then on D3, D5 and D7. The relative variation in platelet count at day n compared to day 1 was calculated as follows: (count at day n − count at day 1) × 100/ (count at day 1 between between D1 and D3, between D1 and D5 and between D1 and D7 were also similar between patients with and without ICUacquired infections (Fig. 12). Discussion In this preliminary study from selected cohorts of nonimmunocompromised patients, sepsis resulted in mild alterations in platelet counts, making it unlikely to become associated with the development of nosocomial infections. It would be relevant to address this question in larger cohorts of non-selected patients, as well as the impact of platelet transfusions in this setting.
Conclusion Changes in platelet counts were not associated with an increased susceptibility towards ICU-acquired infections in non-immunocompromised patients with severe sepsis and septic shock.
Introduction Sepsis is the leading cause of mortality in the intensive care unit (ICU) patients despite the progress regarding their care. The immunodeficiency due to sepsis with the consequent profound lymphocyte alterations is now well proven. The objective of this work was to determine the prognostic impact of lymphocytopenia in septic patients in ICU.

Patients and methods
Retrospective study including all patients hospitalized for sepsis or septic shock between 01/01/2013 and 31/05/2015. The sepsis and septic shock definitions were adjusted with the third international consensus definitions for sepsis and septic shock. Were excluded from the study patients of onco-hematology. Lymphocytopenia was defined as an absolute lymphocyte count less than level of 1500/mm 3 during the first 24 h of hospitalization.
The prognostic factors analyzed for the lymphopenic and non lymphopenic patients were in hospital mortality, the occurrence of nosocomial infections and hospital length of stay.
Results Among the 52 patients, aged 58 ± 17 years, 23 patients were with septic shock and 29 patients with sepsis. IGSII score and SOFA score were respectively 42 ± 19 and 5 ± 3. Four patients were immunocompromised due to HIV infection in one case and an immunosuppressive therapy in 3 cases. Lymphocytopenia was observed in 41 patients (78%). Twenty-eight patients (53%) died within an average of 11 ± 13 days. It was noted the occurrence of 10 nosocomial infections. The median length of stay was 7 days with extremes of one and 74 days. The lymphopenic patients were comparable to non lymphopenic patients in terms of medical history and severity scores. Mortality was comparable between the 2 groups with a rate of 51% (n = 21) in lymphopenic patients and 64% (n = 7) in non-lymphopenic patients (p = 0.51). The earliness of death was correlated with the duration of lymphopenia (R 2 = 0.32, p = 0.007). The occurrence of nosocomial infections was not different between the two groups: 17% (n = 7) for lymphopenic and 27% (n = 3) for non lymphopenic patients. The hospital length of stay was not different between the two groups but was correlated with the duration of lymphocytopenia (R 2 = 0.29, p = 0.000).
Conclusion Lymphocytopenia is frequently found in sepsis. Lymphocytopenia was not associated with excess of mortality nor with the subsequent occurrence of infectious complications during the ICU stay. His persistence was associated with an earlier death and a more prolonged hospitalization.

Competing interests
None. Introduction Relative adrenal insufficiency (RAI) is common in ICU patients, particularly during septic shock (1). It has been shown that the RAI also occurs during cardiogenic shock (2). Septic cardiomyopathy occurs in a significant proportion of septic shock patients. The aim of this study was to evaluate the role of RAI on septic cardiomyopathy. Patients and methods Prospective observational study conducted in the intensive care in one university hospital in France. Patients meeting the criteria for septic shock without prior corticosteroid therapy and without chronic heart disease were included. Total blood cortisol levels were assessed immediately before (T0) a short corticotropin stimulation test (0.25 mg iv of tetracosactrin) and 30 and 60 min afterward. Δmax was defined as the difference between the maximal value after the test and T0. RAI was defined as an inappropriate adrenal response with Δmax < 9 µg/dl and septic cardiomyopathy as the appearance of cardiac systolic dysfunction (left ventricle ejection fraction <45%) within the first 3 days of septic shock. We performed a multivariable analysis using backward stepwise logistic regression to identify independent predictors of septic cardiomyopathy. Discussion Although the definition of RAI is not consensual, a threshold of Δmax at 9 µg/dl has been widely used in septic shock, with or without the use of T0 (1). The usefulness of substitutive doses of steroids in septic shock is controversial, but many authors assume this treatment has a potential in reversing overt vasoplegia. Our data suggest an implication of RAI in septic cardiomyopathy. Conclusion We found RAI to be an independent predictor of septic cardiomyopathy. These findings may suggest a new role for substitutive doses of steroids in the hemodynamic management of septic shock.

Results
Introduction Regional perfusion parameters, like lactate, pyruvate and glycerol, may predict outcome in septic shock patients. Continuous venovenous haemofiltration (CVVH) has been considered beneficial in septic shock patients. The aim of our study was to investigate whether CVVH, in comparison to intermittent haemodialysis (IHD), is able to improve regional perfusion in septic shock patients studied by muscle microdialysis. Patients and methods It was a prospective, randomized, clinical study, including septic shock patients with acute renal failure, aged over 16 years. Patients were randomized to receive either CVVH (n = 10) or IHD (n = 10) for renal replacement therapy. Intermittent haemodialysis was carried out during the first 4 h of the 24 h study period. Systemic haemodynamics and interstitial tissue concentrations of lactate, pyruvate, glucose and glycerol were obtained at baseline, 6, 12, 18 and 24 h after initiation of renal replacement by using muscle microdialysis.
Results Regarding systemic haemodynamics parameters, CVVH caused a decrease in heart rate in contrast to IHD after 6 h (−4 ± 2 vs +6 ± 3/mn). There were no changes in vasopressor support throughout the 24-h study period and so systolic blood pressure remained stable in both groups. During the 24 h of all renal replacement therapies there was no significant change in muscle pyruvate and glucose levels.
During CVVH muscle lactate decreased constantly, as did muscle glycerol levels. This decrease reaches a significant levels at H12 for muscle lactate and at H24 for muscle glycerol (Fig. 13). Conclusion Our results suggest that among septic shok patients, CVVH may improves regional perfusion in comparison with IHD.

Competing interests
None. Introduction Acquired hypernatremia (H-Na) is an independent risk of death among ICU patients (1). In the RCT "Hyper2S" study, we compared normal to 3% hypertonic saline during the first 72 h in patients with septic shock with normal serum Na concentration (SNa) at baseline. The study was prematurely stopped for potential harmful effect associated with more frequent H-Na. We assessed the role of H-Na on mortality.

Fig
Patients and methods Data are a post hoc analysis of the "Hyper2S" study database including 434 patients. SNa was measured at H0, every 12 h for 3 days and then daily until D7. Study fluids were stopped if SNa >155 or >12 mmol/l increase over 24 h. Mild, moderate, and severe H-Na were defined as SNa >145 mmol/L, >150 mmol/L and >155 mmol/L, respectively. SNa profiles were compared between D28 survivors and non-survivors. Acute kidney injury (AKI) was defined by doubling serum creatinine and/or need for dialysis. Results 431 patients with available data were analysed. 234 (54%) developed H-Na (mild: 28%, moderate: 16%, severe: 10%). No matter the absence or presence and its severity, H-Na did not affect mortality (43, 34, 41, and 35%, respectively without, with mild, moderate, and severe H-Na, p = 0.41). SNa profiles were similar between survivors and non-survivors (Table 14). A sensitivity analysis performed among survivors at D3 did not change the results. Compared to patients without H-Na, AKI occurred in 42% of patients with H-N vs 40% (p = 0.58), atelectasis in 8 versus 10% (p = 0.58) and ICU acquired weakness in 7 versus 10% (p = 0.23). Conclusion Hypernatremia occurrence is not associated with an increased risk of morbidity and mortality during hypertonic fluid resuscitation in septic shock.

Competing interests
None. Introduction Guidelines about the moderate hypokalemia treatment (between 2.5 mmol/l and 2.9 mmol/l) are based on experts estimations, and non-specific ones for patients in the intensive care units (ICU). The aim of this study was to evaluate the correction of the hypokalemia in an ICU and the compliance of recommendations. Materials and methods An observational epidemiological, retrospective and monocentric trial has been realized during a period of 13 months (from January 2015 to February 2016). The study population included hospitalized patients in the ICU who have shown a first moderate hypokalemia episode, all cause considered. Patients who have presented an acute renal failure with a KDIGO (Kidney Disease: Improving Global Outcomes) score of three the day of their inclusion were excluded. The main primary study endpoint was percent correction of the serum potassium after 24 h. The secondary study endpoints were the incidence rate of moderate hypokalemia and the efficacy about the hypokalemia correction in accordance with the achieved treatment consistent or not with recommendations. Results 140 patients had at least one episode of hypokalemia. The incidence rate of the hypokalemia first episode was 1.8%. The study population included 106 patients. IGS2 score was 47.1 (± 16). 48 patients required mechanical ventilation at the inclusion. The serum potassium was greater than or equal to 3.5 mmol/l after 24 h about 34 patients (32.1%) (corrected group). At 24 h one patient had a serum potassium higher than 5 mmol/l. The average total potassium was respectively 72.  infusion of potassium and 7 (6.6%) patients have been a management compatible with the most common recommendations (input potassium chloride of 100 mmol, use of the enteral administration and lack of continuous intravenous infusion). The percent correction of the hypokalemia after 24 h was respectively of 2/7 (28.6%) in the group in which recommendations had been respected and of 32/99 (32.3%) in the other one (p = 1.00). Discussion In our knowledge there are no previous studies that have specifically focused on the correction of the moderate hypokalemia in critically ill patients. In our study the incidence rate of the moderate hypokalemia was lower than data from the literature because we have only considered the first episode of the hypokalemia [1]. Among patients without contraindication to the enteral administration, this one was used in less than half of the cases. 62% of these patients received potassium with a continuous intravenous infusion and only 7 patients received medical care conform to the guidelines. The medium potassium quantity provided was very lower to the guidelines. Only 32% of the patients have been corrected after 24 h without any difference in the medium potassium quantity which has been provided in relation to the uncorrected group. Conclusion Only 32.1% of moderate hypokalemia in ICU are corrected after 24 h. The intravenous way is considerably used (in 72% of cases) with a poor return. A wide-ranging study is necessary to determine the best correction modes.

Competing interests
None.  Results 1028 patients were included. Mean ± SD age was 65 ± 16 years, 64% were male, mean ± SD SAPS II was 47 ± 20. ICU length of stay was 7 ± 11 days and ICU mortality rate was 23%.
During the first 7 days in the ICU, 60% of patients received at least one nephrotoxic drug. 31% of patients received one, 18% received two, 9% received three and 3% received more than three nephrotoxic medications. Diuretics, antibiotics and iodinated contrast media were the nephrotoxic drugs most frequently administered to, respectively, 35, 22 and 15% of patients. AKI (KDIGO stage 1 or higher) occurred in 58% of patients during the first 7 days in ICU. The proportion of patients with AKI increased with the number of nephrotoxic drugs received: 189/410 (46%) of the patients not exposed to nephrotoxic drugs developed AKI whereas, respectively, 189/315 (60%), 136/184 (74%), 62/91 (68%) and 24/28 (85%) of the patients receiving one, two, three, and more than three nephrotoxic drugs developed AKI.
The univariate association between the number of nephrotoxic medication and AKI persisted in the multivariate analysis adjusted on baseline SAPS II score (p < 0.001). Conclusion The significant proportion of patients exposed to nephrotoxic drugs and the observed association with AKI warrants further investigation. Statistical adjustments for multiple potential confounders is needed in order to assess a potential causal relationship which would lay foundations for interventional studies.

Competing interests
None.  (1) the minimal kidney aggression by current monomeric nonionic low-osmolar contrast media, late serum creatinine increase being explained by the occurrence of later (between the 6th and the 72nd hour) kidney injury due to critical illness or its therapy or (2) insufficient sensitivity of early (6 h) measurements of this biomarker to detect contrast-associated AKI.

S24 Relationship between ionized calcium and its determinants in septic shock
Competing interests partial financial support, no implication in data analysis and interpretation.
Introduction Diabetic ketoacidosis, generally resulting from an absolute deficiency of insulin, is a frequent cause of hospitalization in intensive care unit. Recommendations for diagnosis of diabetic ketoacidosis, care and site of admission have been published by the English society of diabetology. ICU admission are recommended if one of the following criteria is present: GCS < 12, systolic arterial pressure (SAP) <90 mmHg, SpO 2 < 92%, ketosis >6 mmol/L, HCO 3 < 5 mmol/L, pH < 7.1, potassium level < 3.5 mmol/L or anion gap >16 mmol/L. However, it is suspected that adhesion to recommendations remains low.
In this study, we aimed at describing patients admitted for diabetic ketoacidosis in ICU. We looked at adhesion to published recommendations regarding admission and care. We also described metabolic complications and looked for an association between complications and dose of initial insulin therapy. Complications Hypoglycemia (<3.9 mmol/L) was observed in 48% of patients within the first 24 h in which 19% were <2.9 mmol/L. This was 47 and 15% of patients between 24 and 48 h of ICU stay. Hypokalemia below 3.5 mmol/L happened in 44% of patients within the first 24 h and in 52% between 24 and 48 h. Neither hypoglycemia nor hypokalemia were correlated with initial insulin bolus or initial dosage of continuous intravenous insulin. Hypophosphatemia <0.30 mmol/L was observed in 15% of patients. Discussion In this study, admission to ICU was consistent with British recommendations since most patients presented at least one clinical or biological criterion indicating ICU admission. Arterial blood gas were sampled in the large majority of patients despite consistent data showing that venous blood gas might be sufficient in non-hypoxemic patients. Also, initial insulin bolus and sodium bicarbonate perfusion were performed in a significant subset of patients despite absence of convincing data or recommendations supporting their use. Finally, significant hypokalemia and hypoglycemia were frequent in these patients. These complications are in theory favored by insulin therapy but we did not observe a correlation between administration of an insulin bolus or the dose of continuous intravenous insulin perfusion.

Patients and methods
Conclusion In this retrospective multicentre study, patients admitted in ICU for diabetic ketoacidosis were correctly oriented regarding the British recommendations. Metabolic complications (hypoglycemia and hypokalemia) were frequent but not correlated with initial dose of insulin.

Introduction
The appropriate rate for hypernatremia (H-Na) correction is unknown. Under-correction could be associated with worse outcome. Experts recommend a rapid correction of acute (<2 days) and sever (> 150 mmol/l) H-Na with a rate of −2 mmol/l/h until Na < 146 mmol/l (1). Correction should be, therefore, obtained within 24 h. In patients with septic shock resuscitated with iso-or hypertonic saline and who acquired acute severe H-Na, we assessed if the correction rate was associated with mortality. Patients and methods Data are a post hoc analysis of the RCT "Hyper2S" database comparing normal to 3% saline for 72 h in septic shock. Serum Na (SNa) was measured at H0, every 12 h for 3 days and  ). H-Na correction rate was more rapid in non-survivors, p = 0.05 (Table 15). Over-correction occurred similarly in survivors (15%) and non-survivors (10%). The time to reach SNa normalization was shorter in nonsurvivors (p = 0.05). After adjustment for SAPSII and MacCabe scores, more rapid correction rate remained significantly associated with mortality: OR 0.16; 95% CI (0.03-0.86), p = 0.048. Conclusion In the context of acute severe H-Na induced by fluid resuscitation, a rapid correction rate might be associated with even aggravated rather than improved mortality.
Introduction Systemic Capillary Leak Syndrome (SLCS) is a rare disease characterized by recurrent life-threatening attacks of capillary hyper permeability in the presence of a monoclonal gammopathy (MG). During acute episodes, the leak of fluid and proteins from the intravascular compartment to the interstitium results in clinical signs of both acute hypovolemia and interstitial edema. Biological profile is pathognomonic with marked hemoconcentration and paradoxal hypoproteinemia. Hypovolemic shock is the classical feature of severe SCLS attacks. However, beside this typical hemodynamic profile, several case report described myocardial dysfunction during SCLS attacks. The objectives of this study were to assess frequency, characteristics and outcome of myocardial involvement during severe SCLS attacks. (64%) mechanical ventilation, 6 (43%) renal replacement therapy, 5 (36%) veno-arterial extracorporeal membrane oxygenation, 1 (9%) intra-aortic balloon pump and 1 (9%) an IMPELLA. Compartment syndrome occurred in 5 (45%) patients and 4 (36%) died in ICU. We then compared the 11 patients with myocardial involvement to the 28 without Clinical and biological manifestations were similar in between groups. However, chest pain (54 vs 11%, p = 0.008), dyspnea (82 vs 39%, p = 0.03) and respiratory failure (54 vs 18%, p = 0.044) were more frequent in patients with myocardial involvement than in others. There was no difference between groups regarding treatment received in ICU, complication and outcome except for the use of VA-ECMO (36.4 vs 0%, p = 0.004). Conclusion Myocardial involvement seems frequent in patients with severe SCLS attack, occurring in 22% of the cases. Such patients exhibited classical features of SCLS attacks. Myocardial involvement was responsible for altered LVEF or transient ventricular hypertrophy. Myocardial dysfunction could be severe, even requiring mechanical circulatory support. SCLS attacks should be known as a cause of severe reversible myocardial dysfunction and hypertrophy.

Competing interests
None. Introduction In refractory cardiorespiratory emergencies, ECMO appears a good alternative to conventional treatment. Its extracorporeal circuit justifies curative anticoagulation explaining haemorrhagic and thrombotic complications. Activated Clotting Time (ACT) is empirically and commonly used to assess anticoagulation but with large inter and intraindividual variabilities. In practice, antiXa activity dosage is available to approach anticoagulant effect of heparin and is less expensive, but data during ECMO are missing. We sought to demonstrate the lack of correlation between antiXa and ACT in patients under ECMO support.

Patients and methods
We prospectively include patients supported by ECMO in CHU Toulouse, France, between 01/2014 and 04/2015 for circulatory/respiratory support. Anticoagulation was achieved by unfractionated heparin: initial bolus then continuous intravenous infusion (800-1200 IU/h), for antiXa target of 0.2-0.4. Concomitant dosing of antiXa (laboratory) and ACT (Hemocron ® ) was conducted two times a day on the same sample throughout the ECMO period. Relationship between ACT and antiXa was analyzed by Spearman correlation (Rho). After transformation into categorical variables (obtained target = 1; outside the target = 0), analyzes were completed by a concordance study (Kappa). As recognized on literature ACT's targets were between 180 and 220.
Results 65 patients were included: 46 men (72%), median age 55 yo (53-57). Indications were veno-arterial (n = 42) and veno-venous ECMO (n = 23). ECMO median duration was 5 days (hours to 30 days). Spearman correlation test found low and inconsistent correlation between antiXa and ACT (Rho Spearman < 0.4). This correlation lack present from the day one, worsens over time. Analyzed Kappa showed no discrepancy between the areas "targets" of ACT and AntiXa confirming the results (Table 16). Conclusion Use of ACT for ECMO anticoagulation monitoring doesn't seem appropriate and high price probably justifies preferential use of antiXa in clinical practice. Analyzes of relationships between antiXa and bleeding/thrombotic events are needed to confirm the antiXa place and its target in these indications.
Introduction Postcardiotomy cardiogenic shock (CS) has an incidence of 2% to 6% after routine adult cardiac surgery. In 0.5-1.5% of cases, an venoarterial extracorporeal life support (VA-ECLS) is requested. The 6-month survival rate is 17.6% (1). Survivors may suffer of physical and psychological impairments as well as an alteration of quality of life. This study was designed to assess the outcomes, long-term health- Since ICU discharge, 65% of patients reported physical sequelae., ECLS-related limb pain occurs in 38% of patients while paresthesia occurs in 23% and chronic-tiredness in 69%. Mean Karnofsky score was 68% (Table 17). Conclusion After VA-ECLS for postcardiotomy cardiogenic shock longterm physical and psychological sequelae are frequent in survivor   Discussion Interest for fluid management is growing in critical patients. Nevertheless, no study has yet investigated its impact in selected patients with cardiogenic shock treated with VA ECMO. Our study suggested a possible association between fluid overload and mortality but lack the power to confirm these results with multivariate analysis. Conclusion Fluid management is a key therapy during VA ECMO but fluid overload could be associated with worsen outcomes. Further studies with larger population are warranted before considering fluid restriction trials.
Introduction Extracorporeal life support (ECLS) has taken an important place in the treatment of cardiogenic shock (CS) or refractory cardiac arrest (CA). However, ECLS deplore a high mortality rate in the first days raising important ethic and economic consequences. In this context, continuation of support should be reassessed precociously. The aim of this study was the research of prognostic factors of 30-days mortality, 24 h after ECLS implantation for CS or CA. Materials and methods All patients undergoing ECLS in our tertiary center during a 2-year period were prospectively included. The ECLS were managed with a multidisciplinary protocol based on consensus. Clinico-biological data were collected just before and 24 h after ECLS implantation. These data were compared between survivors and deceased at 1 month. , CPC score was respectively 1 for 6 patients, 2 for 2, 4 for 1. At 12 months, CPC score changed only for the 2 patients with a CPC score at 2 (one died after another suicide attempt, one changed his CPC score to 3). In the group without CA (n = 88), 79 had normal neurological status at 6 months and 78 at 12 months (one patient died because of a cancer). Among these patients, 96% returned at home and 77% returned to work. 16 (18%) patients re-attempted suicide in the year. The major risk factor of mortality is the presence of a cardiac arrest on hanging site. All the other factors found to be related to mortality are well known risk factors in cardiac arrest of other origin. In univariate analysis, risk factors of neurological sequelae at 6 months were a cardiac arrest on hanging site (p = 0.045) an elevated diastolic blood pressure (87 vs 70 mmHg; p = 0.04), a lower initial Glasgow score (4 vs 5; p = 0.04), and an elevated blood glucose (1.39 vs 1.13 g/L p < 0.001) at admission in ICU. Discussion Our cohort of self-hanging patients can be divided in two parts: a) patients with CA in the pre-hospital period with a high mortality and a good neurological recovery in 2/3 surviving patient, but with a small group; b) patients without CA with a very low mortality and a very good neurological recovery. These results seem to be better than in the most important cohort [1] published until now in self-hanging patients without CA and not treated by HBOT (mortality at 9.5% and 3.5% of poor neurological recovery). Conclusion Patients surviving a self-attempted hanging who have not presented CA and treated by HBOT have mainly a good neurological outcome. Randomized control study should be undertaken to confirm HBOT effectiveness in that indication.
Introduction Venoarterial extracorporeal membrane oxygenation (VA-ECMO) is increasingly used to treat refractory cardiogenic shock or cardiac arrest. Acute brain injury (i.e. ischemic stroke, haemorrhage and/or failure to awaken because of diffuse brain injury) may occur in up to 15% of patients on VA-ECMO and is associated with increased mortality and poor functional outcome in survivors. However, early indicators of neurological outcome are lacking in this population. We aimed to assess the prognostic value of early electroencephalography (EEG) alterations during VA-ECMO.

Patients and methods
We conducted a prospective single-center study in the medical ICU of a university hospital on consecutive patients cannulated to VA-ECMO. A standardized clinical neurological evaluation including the RASS score, the GCS score, the Full Outline of UnResponsiveness (FOUR) score and brainstem reflexes was coupled to an intermittent EEG. EEG was recorded as soon as possible within the first 72 h after VA-ECMO cannulation. EEG characteristics were analyzed by a neurophysiologist who was blinded to the patient's condition. A severely altered EEG pattern was defined as a predominant delta frequency, discontinuous, unreactive and/or an isoelectric background. The primary endpoint was poor neurological outcome, defined as the composite of death or acute brain injury on neuroimaging within 28 days. Data are presented as median (interquartile range) or number (percentage). False-positive rates (FPRs, corresponding to 1-specificity) of poor neurological outcome were calculated for each significant predictor, using an exact binomial 95% confidence interval (CI). Results Sixty-nine (age 58 (50-67) years) patients with a SOFA score of 14 (13-17) were included. Main indications for ECMO were: post cardiac surgery (n = 25, 36%), terminal dilated cardiomyopathy (n = 12, 17%), and acute myocardial infarction (n = 11, 16%). Cardiac arrest before ECMO cannulation was noted in 20 (29%) patients. EEG was recorded 1 (1-2) days after VA-ECMO cannulation and 62 (90%) patients were sedated at time of EEG. At day 28, 46 (67%) had a poor outcome (n = 37 deaths and n = 9 patients alive with acute brain injury). In univariate analysis, a lower RASS score (p = 0.003), a lower FOUR score (p = 0.001), a lower score on the motor component of the Glasgow coma scale (p = 0.001), and a lack of cough reflex (p = 0.033) at the time of EEG were significantly associated with a poor outcome. A severely impaired EEG pattern or presence of a discontinuous background activity were also associated with a poor outcome (p = 0.015 and p = 0.002, respectively). Indicators of poor neurologic outcome are presented in the Table 19. Among all parameters, a discontinuous background activity was the only variable that constantly predicted poor outcome (false-positive poor outcome prediction rate of 0%, 95% CI 0-15%). Conclusion Early intermittent EEG has a strong prognostic value for sedated patients on VA-ECMO. Presence of a discontinuous EEG background activity seems to be more accurate than clinical alterations to predict a bad neurologic outcome at 28 days.

Competing interests
None.   Table 20). It was not found a significant association of CTP to mortality (46% in the case group and 30% in control group, p = 0.08). Other factors that increased mortality were coma, seizures, shock, oedema, cellularity in CSF >1200 units/mm 3 . Otherwise, the ventilation length was prolonged with CTP group (8.7 vs 3.4 days, p = 0.034) and neurological sequels namely the epilepsy was more frequent with the group CTP: (23 vs 5%, p = 0.028).

S36 Transjugular intra hepatic porto systemic shunt (TIPS) placement induces modifications of cerebral multimodal MRI in cCirrhotic patients, even in the absence of development of hepatic encephalopathy
Conclusion The occurrence of CTP on bacterial meningitis was significantly associated with CT scan lesions which seems to be an association be in both directions. Also, the positive culture predisposed more to the CTP. Mortality was higher with the presence of CTP but without real significance. The CTP was a factor that extends the ventilation time and exposed to the post infectious epilepsy.
Introduction Acute bacterial meningitis requires rapid triage and therapeutic decision-making. The aim of this study was to assess the overall ability of a point-of-care glucometer to determine bacterial infection in cerebrospinal fluid (CSF).

Materials and methods
We performed a prospective, observational study. We included patients for whom an analysis of CSF was indicated by the physician in charge with blood sampling performed for glucose concentration measurement within 1 h. We simultaneously measured the glucose concentrations in CSF and blood using a central laboratory and point-of-care glucometer. The diagnosis of bacterial meningitis was determined by two physicians after reviewing the complete medical chart. We compared CSF and blood glucose concentrations and CSF/blood glucose ratios obtained at the bed-side with a glucometer versus those obtained by the central laboratory. We determined the performance characteristics of the CSF/blood glucose ratio provided by a glucometer to detect bacterial infection in the CSF immediately after CSF sampling. Conclusion We demonstrated that the CSF/blood glucose ratio measured by a glucometer can serve as a clinical decision support tool for the early detection of CSF with a high probability of bacterial infection. This costless point-of-care method has the potential to expedite medical decision-making for the triage of adult patients with suspected meningitis in the emergency department immediately after lumbar puncture.

Competing interests
None.  Introduction Cardiac arrest remains a frequent cause of admission in intensive care unit. A majority of patients will die during their hospital stay mainly from consequences of hypoxic-ischemic brain injury after a decision of withdrawal of life sustaining therapy support by a prediction of poor outcome. The reliability of prognostication is crucial, but is still a difficult and uncertain exercise. EEG is the most widely used prognostic tool to support a clinical examination and is accessible in most hospitals. It is recommended for both prognostication and ruling out subclinical seizures. There is no high-level evidence for predicting poor prognosis using EEG because of the wide variety of classification systems used and the interrater variability. Our objective is to assess the prognostic value of simple EEG features based on the recent American Clinical Neurophysiology Society (ACNS) standardized classification and to study the interrater variability.

Patients and methods
We conducted a retrospective monocentric observational study in a 12 bed medical intensive care unit of the university Hospital la Timone, Marseille, France. All patients aged of more than 18 year-old admitted for a resuscitated cardiac arrest between November 2012 and July 2014 who underwent therapeutic hypothermia and a full multimodal prognostic evaluation including a EEG were included in the study. Outcome was classified according to the Cerebral Performance Category Score measured at day 28. Unfavorable outcome was defined as death (CPC 5), persistent vegetative state (CPC 4), or severe neurological disability (CPC 3). Favorable outcome was defined as moderate neurological disability (CPC 2), or no disability (CPC 1). EEG was performed in all patients still comatose after rewarming between 48 and 72 h after admission and after discontinuation of sedation. EEG interpretation was made by 2 independent senior neurophysiologists, blind to the outcome. EEG features are based on the latest ACNS classification. For each EEG feature, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) for predicting an unfavorable outcome were calculated. Results During the study period, 122 cardiac arrest were admitted of which 48 patients went through a full neurologic evaluation and were finally included in the study. According to neurological outcome, 19% had a favorable evolution, and 81% had an unfavorable outcome. The presence of burst suppression, and epileptiform activity was constantly associated with an unfavorable prognostic with a 100% specificity and 0% false positive. A non-reactive EEG is strongly associated with an unfavorable evolution with a 89% specificity and 3% false positive. Other features including periodic or rhythmic patterns and low voltage were inconstantly associated with unfavorable outcome. Kappa score for all EEG feature was slight or fair and always under 0.4. Discussion This study allowed us to identify a homogenous cohort of comatose patient after cardiac arrest who underwent therapeutic hypothermia. We identified simple EEG features based on the new classification of the ACNS constantly associated with unfavorable outcome. These features must be known by intensivists to better integrate EEG in the multimodal evaluation of neurological prognostic. There is important interrater variability that must lead to caution and to always use multimodal approach to prognostic an unfavorable outcome. Conclusion Bedside EEG is an excellent tool for predicting outcome of post-anoxic coma through simple EEG features. Burst suppression, epileptiform activity and non-reactive EEG are strongly associated to neurological outcome after cardiac arrest. However, the interrater variability emphasize the need of being well trained for the standardized methods of evaluating EEG parameters.
Introduction Emergent reintubation is a well-known risk of laryngotracheal trauma and of ventilatory acquired pneumonia. To precisely define its risk before extubation for each patient is a part of quality of care in intensive care units. None of these 43 consecutive children representative of PICU activity has been reintubated. The coming prospective muticentric study which aims to validate ALT in childhood must precisely define this criteria of evaluation.
Conclusion The different methods of ALT are feasible in real clinical conditions in PICU. Because of the increasing use of cuffed ETTs in a wide variation of patients with different body weight, the best ALT to use at the bedside must be definitively validated in this population.
Introduction Prolonged mechanical ventilation (PMV) and chronic mechanical ventilation (CMV) in neonates is associated with a high morbidity and mortality. The objective of the study is to identify, among the patients with PMV, those that evolved to CMV, as well as the adverse respiratory, neurological and feeding sequelae.

Patients and methods
We conducted a retrospective study of the last 10 years at the CHU Sainte-Justine (Montreal, Canada). Chart review included patients with PMV (≥21 days) using the paediatric definition adapted from the 2005 NAMDRC consensus conference (1). Demographic and clinical data, including follow-up at 6 and 18 months corrected age, was collected for each included patient. The evolution of PMV neonates with CMV (≥125 days) and without (21-125 days) was compared.

Results
We identified 174 neonates that met criteria for PMV. Patients born between 2004 and 2011 (n = 110, 63% of the cohort) were analyzed. Around half of the patients (5-10 patients a year) are transferred from the neonatal unit to the paediatric intensive care unit. In our center, they represent around 1% of total admissions, but their length of stay is among the longest. Among these 110 newborns, 79% were preterm (n = 87) with 45% (n = 50) born before 29 weeks gestation. Of all patients with a malformation (52%, n = 57), 21 had a thoracoabdominal anomaly and 24 had congenital heart disease. Thirty-six patients had CMV with mean ventilation time of 239 days (range 125-753 days). Survival at 18 months corrected age was 66% (49/74) in the PMV group and 72% (26/36) in the CMV group. At 18 months corrected age, 30% of patients were dependent on artificial enteral feeding (nasogastric tube or gastrostomy), with 18% in the PMV group and 54% in the CMV group. Nine percent of patients had oxygen supplementation (2 patients in the PMV group and 4 in the CMV group), and 7% were mechanically ventilated. Ten percent of patients had a tracheostomy (5 patients in the PMV group and 2 in the CMV group). Discussion Neonates with CMV have more sequelae. Their rapid identification (at 21 days of ventilation) is essential to implement multidisciplinary development care in order to minimize neurodevelopment impairment. Conclusion Most newborns in our PMV cohort have a congenital malformation. Survival at 18 months corrected age appears equivalent in both PMV and CMV group. Artificial enteral feeding is more frequent in the CMV group and most patients have no respiratory support at 18 months corrected age.

Competing interests
None.

Introduction
The value of pressures and volumes in assessing the fluid responsiveness depend on the systolic cardiac function in adult (1). We have studied the relative value of static filling volume and pressure to predict the fluid responsiveness, according to systolic cardiac function in children during acute circulatory failure. Patients and methods Patients under 8 years old with an acute circulatory failure of two intensive care units during a 1 year period of inclusion were analyzed. An exhaustive cardiac echography was performed initially (indexed End-diastolic volume (EDVi) and E/e' from transmitral and tissue Doppler were recorded), and the stroke volume index (SVI) was measured before and after a fluid challenge (a 10 ml/ kg of crystalloid over 10 min Results Twenty-five children with acute circulatory failure were included. Fluid responsiveness occurred in 6 of the 13 fluid loading events with low LVEF, and in 6 of the 12 fluid loading events with normal LVEF. Pressure approach: For low and normal LVEF, the AUC-ROC for fluid responsiveness was respectively 0.83 (CI 0.56-1)/0.73 (0.39-1) for a E/e' .The best thresholds of E/e' in low LVEF was 7.9 with a sensitivity of 87 (CI 62-100) % and a specificity of 67 (CI 33-100) %. For low and normal LVEF AUC ROC was respectively 0.6 (CI 0.21-0.99)/0.52 (CI 0.14-0.9) for the PVC. Volume approach: For low and normal LVEF, the AUC-ROC for fluid responsiveness was respectively 0.69 (CI 37-1) and 0.74 (0.4-1). The best thresholds in normal LVEF was an EDVi below 29 ml/m2 wit a Specificity of 83 (CI 50-100) and a sensitivity of 71 (CI 42-100) %. Discussion Our study shows a variation of the diagnostic value of E/e' and EDVi according to the left ventricular systolic function. Therefore, the systolic function should be taken into account to analysed the E/e' and EDVi value. Few preload dependency markers are validated in Children and none for children in spontaneous ventilation (2). Our study suffers from a lack of power that calls into question the validity of our results. Another limitation is that both approaches with volume and pressure are not very discriminant as it is known for static value in adults. Our study illustrates that, on a pressure-volume curve, when the cardiac inotropism is reduced, the filling of the Left ventricle is moved to the up and right of the curvilinear diastolic function curve. Therefore, pressure variations are larger than volume variations. These values should be monitored on a larger scale to define their exact diagnostic value. Conclusion Static PVC value is a low preload-dependency surrogate. When LVEF is low a pressure evaluation based approach seems more accurate. When LVEF is normal a volume evaluation based approach seems informative as predicted by the slope of the end diastolic pressure volume curve. Those both static approaches remain of poor diagnosis accuracy.
Introduction Acute viral bronchiolitis is a primary cause of respiratory distress in paediatric intensive care unit (ICU). Prone position (PP) is commonly used in neonates to improve respiratory mechanics and has been found beneficial to adult patients with acute respiratory distress syndrome. We aimed to evaluate the effect of PP on work of breathing as compared to supine position (SP) in children with severe bronchiolitis requiring non-invasive ventilation.

Patients and methods
The protocol was approved by our IRB (2015-A01200-49). Fourteen infants (9 boys) with median age 33 days [firstthird quartiles 25-58] with severe bronchiolitis requiring CPAP were included after written informed consent. Children were investigated in PP and SP each applied for 1 h in a random order with a washout period of 10 min between them. Level of CPAP was set at 7 cmH 2 O in both conditions. Oesophageal pressure probe was inserted orally (CTO-2 pressure transducer, Gaeltec, Scotland) to measure oesophageal pressure. Flow and airway pressure (Pmo in Fig. 15) were simultanuously recorded using a Neurovent data acquisition system (Neurovent Inc, Toronto, Canada). One hundred breaths were analyzed in each condition, in which work of breathing was estimated from oesophageal pressure-time product (PTPes) and oesophageal swings (Fig. 15). Data were expressed as median (first-third quartiles) and compared by using the Wilcoxon two-sample paired sign test. A p-value below 0.05 was considered significant. . The eDTB contains data from ventilated patients (invasively and non-invasively) and details concerning ionotropic and sedative treatment during PICU courses. Discussion As far as we know, this eDTB is currently the only one as exhaustive available in PICU worldwide. After almost 3 years of multidisciplinary collaboration, we are able to collect many useful physiological, therapeutic and medical data in an ongoing eDTB. Although many concerns remain concerning data validation, organisation and exploitation, this eDTB already contribute to the development of clinical decision support systems and virtual patient validation and we create international collaborations to further develop these tools. Three research protocols using the database are ongoing including: validation of a neuromonitoring clinical decision support system, validation of a cardio-respiratory simulator, developement and validation of the automatic diagnosis of pediatric acute respiratory distress syndrome and development of SpO2 forecast using artificial neuronal network. Conclusion Thanks to informatics and electronic devices improvement, data gathering in intensive care units has empowered. We hope that our work in PICU will encourage other teams on the way of data gathering, in order to build an international PICU eDTB in a close future.

Competing interests
None. Introduction Severe trauma is rare in the pediatric setting (15% of all trauma in France). However, its morbidity and mortality remain high, in relation to brain injury. Pediatric traumatic brain injury (TBI) prehospital care is challenging for non-pediatric retrieval teams. Though, we disseminated pediatric TBI pre-hospital care regional guidelines and thereafter intended to assess severe pediatric trauma pre-hospital care and secondary cerebral insults control.

Materials and methods
We conducted a retrospective study in a single pediatric trauma center. Children admitted in emergency room with severe trauma and moderate to severe TBI (Glasgow coma scale ≤12) from June 2014 to March 2016 were included. Pre-hospital and hospital data regarding primary care, equipment, medications and secondary cerebral insults control (i.e. blood pressure, oxygenation, CO 2 level, temperature, glycemia) were collected from medical files. Two pediatric transport team experts assessed the quality of pre-hospital care, based on two major endpoints. Results Twenty-nine files were analyzed. Median ISS was 34 . All the children had been referred directly from the trauma scene to the pediatric trauma center. They were all intubated in the prehospital setting, 4 (13.7%) presented with SpO2 < 90% before or at emergency room admission, and 17 (58.6%) presented with a pCO2 >45 mmHg at admission. At least one peripheral catheter was inserted in all the children. Mean total fluid bolus was 26.9 mL/kg (± 14). Nor-epinephrine was administered in 16 (55%) children. Mean blood pressure was below age threshold in 18 (62%) children during transport or at admission. An intracranial hypertension treatment (apart from sedation) was delivered in 17 (55%) children before admission. Body temperature was monitored in 10 patients and 17 were hypothermic at emergency room admission. Experts concluded on sub-optimal care in 17 children: major endpoint was "respiratory care", "hemodynamic care" and "neurologic care" in 13, 12 and 9 patients respectively. Discussion On this small series, we showed pre-hospital sub-optimal care regarding secondary cerebral insults control, especially regarding CO 2 level, blood pressure and body temperature. Our results will help to design new care improvement strategies (e.g. sedation, fluid bolus and ventilation optimization, early use of vasoactive drugs, systematic body temperature monitoring…). Conclusion Data on pre-hospital secondary cerebral insults care are rare in the pediatric setting. Based on our results, we aim to improve quality of care of children presenting with traumatic brain injury, and to reduce its morbidity and mortality.
Introduction Unsuccessful extubation from mechanical ventilation increases mortality and morbidity. To reduce the extubation failures in our Intensive Care Unit we used a mechanical ventilator weaning protocol, based on published data. During the first part of the study, risk factors and incidence of extubation failure were first described.
Afterwards in the second part, our mechanical ventilator weaning protocol was tested to determined its efficiency regarding the extubation failure. Patients and methods A monocentric and observational study, was first conducted. We included 245 children aged from birth to 18 old, during a period of 15 months and collected for each patient their medical history, intubation and extubation parameters, and existing events of extubation failure or extubation complication. The second part of the study was prospective, we include 70 patients extubated by applying our mechanical ventilator weaning protocol. Results Average duration of mechanical ventilation was 51.6 h in the first part of the study. Using a univariate analysis, duration of mechanical ventilation was a risk factor of extubation failure with an average duration of 131. Discussion Our study confirms published data about extubation failure risk factor like duration of intubation, chronic respiratory affection, history of previous intubation, and the administration of benzodiazepine. It is the first pediatric study that shows a reduction of extubation failure by using a specific mechanical ventilator weaning protocol. The mean bias of our its retrospective and prospective character. Conclusion Our study shows the interest of a mechanical ventilator weaning protocol to reduce the incidence of extubation failure. We currently continue the apply our protocol to include more patients in order to confirm our results.

Introduction
Stroke of the child is formidable though it is ten times rarer than in adults, but this scarcity can have adverse consequences on the speed and quality of the management and the consequences on later psychomotor development. Our goal is to describe the clinical and therapeutic aspects of these pediatric stroke while bringing our experience. Patients and methods Retrospective study of cases of children hospitalized in general intensive care unit to the pediatric hospital Canastel Oran for stroke during the period from January 2014 to January 2016. The clinical, etiological, para clinical, and scalable were studied and transcribed on a standard electronic form.All patients had a brain CT. Magnetic resonance imaging(MRI) was possible in 02 patients for lack of availability of the technical facilities during the study. Results Ten cases were selected. The mean age was 64 months (1 month to 15 years), 66% are male, 2 patients had a history of CHD like tetralogy of Fallot and complicated bronchiolitis myocarditis, one patient had a history of petechial purpura, 1 other was a factor 7 deficiency, headache history was noted in 02 patients, and 04 patients with no particular antecedent was found. All patients arrived comatose 11/15 score on the scale of glasgow, isochores reactive pupils with a motor deficit of hémicorps, 03 patients have degraded their neurological score with onset of clinical signs of hypertension intra cranial namely anisocoria and hypertension requiring osmotherapy, sedation and mechanical ventilation with an average duration of 2-7 day. O1child arrived brain dead, 03 patients had generalized tonic-clonic seizures which yielded after taking a benzodiazepine (Diazepam) and phenobarbital (like gardenal). Cerebral CT was performed in all cases and could we revealed the nature of the stroke hemorrhagic in 07 cases and ischemic stroke in 03 cases. Two patients have benefited from an MRI that found a thrombosis of the artery internal carotid right Sylvian. Besides symptomatic treatment, treatment was initiated based on the type of stroke, 03 patients received low molecular weight heparin (LMWH) at 0.1 ml/kg in addition to symptomatic treatment, 02 patients received vitamin K. Four patients died in an array of autonomic disorders and evolved favorably and six patients were transferred to a pediatric unit. The average length of stay in ICU was 5.5 days (2-10 days). Discussion The mortality rate is important since no specialized center for children, and difficulty especially in the diagnostic imaging field while Suspected stroke should be confirmed by imaging and the diagnostic delay. Which is due to a poor assessment of the initial situation in half of the cases by the parents, the other half by the Swiss magazine consulté.une doctor showed that in a study in 42% of children with stroke, this diagnosis was not primarily discussed and that in 11% of cases the cause of the stroke was poorly evaluated [1]. Heart disease certainly represent the second most important risk factor. A collaboration of a team must be multidisciplinary, death has affected mostly older children whose age is between 11 and 15 years, who have a hemorrhagic stroke against by infants who have an ischemic stroke have evolved and oriented they exceed the acute phase to pediatric services for further investigation and monitoring. Conclusion The child may also be having a stroke, which usually reaches the elderly. This justifies a good knowledge of this disease, and multiply the initial management efforts to reduce mortality and improve prognosis.

S50 Bacterial nosocomial infections incidence in medicosurgical pediatric intensive care unit
Anwar Armel 1 , Benqqa Anas 1 , Samira Kalouch 2 , Khalid Yaqini 2 , Aziz Chlilek 2 Introduction Nosocomial infections are a main problem for public health for their cost as well as for the morbidity and mortality they generate. They are particularly common in intensive care units due to patient's lower defenses and of invasive procedures proliferation. Work's purpose: • Determine the epidemiology of bacterial noso-Comiales infections (IBN) in the medico-surgical pediatric intensive care department of Children's University Hospital of Casablanca. • To identify factors associated with these infections.

Patients and methods
We led a retrospective study of hospitalized patients, spending more than 48 h in medical-surgical pediatric intensive care department, at the University Hospital Ibn Rochd of Casablanca, over a period of 12 months from 1 January 2015 to 31 December 2015. Results During the studied period, 420 patients were admitted at intensive care with a stay of more than 48 h. Thirty episodes of INB were recorded. The incidence rate was 7.1% and the incidence density was 20.6% per 1000 hospitalization's days. The admission average age was 4.6 ± 22-month starting from 1 month to 12 years with a male predominance (60%). Most of admissions (80%) was related to medical background, 50.3% received from other hospital department. Furthermore, 40% of the patients received prior antibiotics, usually prescribed before ICU admission. Invasive procedures (intubation, central catheterization) were used in 93.3% of patients, VVP only in 6.66%, tracheotomy in 33.3 and 6.66% had received surgery. Gram-negative bacilli (BGN) were isolated for a lot of patients, dominated by Acinetobacter baumannii. These bacteria were isolated throughout the study year. Risk factors Analysis underlined that the presence of invasive procedures enhances IN risk, that is central venous catheter and the need for mechanical ventilation. Conclusion Nosocomial bacterial infections are dominated by pneumonia and central catheter infections, and are mainly due to BGN. The factors associated with these infections were identified.

Introduction
The Guillain-Barré syndrome (GBS) is the most common cause of acute flaccid paralysis in children since the acute anterior poliomyelitis eradication. Few studies have been held on the topic and knowledge of GBS in children, although it is recognized that the etiologic mechanisms, and clinicobiological background, are the same as in adults, prognosis remains different. Our work's aim is to study this disease's mortality factors of children hospitalized in pediatric intensive care. Patients and methods It is a retrospective, descriptive, mono centric study to review 35 patients with GBS between January 2009 and December 2015 and hospitalized at pediatric intensive care department of AbderrahimHarouchi hospital of Casablanca. The used software is SPSS 16.0 to compare the bivariate variables, we used the khi2 test, and to compare quantitative variables, the ANOVA to 1 factor test was used. The level of significance was fixed at 5% with 95% confidence interval.

Results
The disease was predominant in male with a sex ratio of 1.7 men/women. After a prodromal event, usually infectious (95.7%) and a free interval of 15 days on average to start motor disorders. These are of two types: either a hypo or areflectic flaccid paralysis of the lower limbs (25.7%) of ascending evolution in 91.4% of the cases. Either flaccid tetraplegia or hypo areflectic, (74.3%). Ventilation was required in 65.7% of the cases, and specific treatments based on immunoglobulins were administered in 88.6% of the cases. Death's rate is still high (22.9%) and mainly due to hospitalization complications. In our study respiratory disease was noted in 65.7% of the cases, also other signs of serious illness such as swallowing disorders (71.4%) and autonomic disorders (14.3%) also noted what led to management in intensive care for all our patients. These patients study allowed to identify some mortality prognosis factors of the disease in intensive care units (such as male gender, Ig administration duration, the occurrence of autonomic disorders like blood pressure instability), the most discriminating remains the occurrence of nosocomial infections. Conclusion It must be underlined, that in view of our strict inclusion criteria, focusing only on patients admitted at intensive care and of the relatively small sample size (35 cases), our results must be qualified and must be enhanced by additional and more varied studies to better understand this disease in children.
Introduction Early surgical treatment is recommended for refractory intracranial hypertension (HTIC) in children to improve vital and functional prognoses, whether traumatic or vascular cause. The main objective of this study was to compare the mortality and morbidity of children with severe intracranial hypertension after severe head trauma (TC) or due to vascular cause after decompressive craniectomy (DC) or medical therapy alone. The secondary objective was to identify the initial severity factors associated with higher mortality. Patients and methods A retrospective study was performed with data collected from patients aged under 18 years-old admitted to our pediatric intensive care unit for severe intracranial hypertension of traumatic or vascular cause, between January 2000 and January 2016. They were divided into 2 groups: patients who received medical therapy alone and those treated with decompressive craniectomy after optimal medical management. Results A total of 83 children were included. Among them, 17 were treated with DC (6 HTIC of vascular cause and 11 HTIC of traumatic cause), and 66 were supported by medical means only (6 HTIC of vascular cause and 60 HTIC of traumatic cause). In the population "traumatic intracranial hypertension", we note that children in the "DC" subgroup are more often in mydriasis upon arrival (p = 0.017) than in the subgroup treated medically. In this same population, children in the "DC" subgroup received higher doses of MIDA-ZOLAM (p = 0.015), of MANNITOL (p = 0.0041) and hypertonic saline (p = 0.0006) than in the other subgroup. In the population "vascular intracranial hypertension" the two subgroups were comparable. In the case of traumatic intracranial hypertension, mortality rate in the "DC" subgroup was 54.5% against 13.34% for children treated medically (p = 0.0015); "DC" children had more metabolic complications such as hypernatremia than "not DC" children, p = 0.043. Mortality rate in the «vascular intracranial hypertension» group was 0% for children treated with decompressive craniectomy, and 83.3% for children treated medically alone (p = 0.015). Patients treated surgically in the «vascular intracranial hypertension» group had longer overall stays (p = 0.009) and longer ICU stays (p = 0.016). POPC Score (Pediatric Overall Performance Category) upon discharge for children with intracranial hypertension of traumatic cause treated with decompressive craniectomy was 4.91 ± 1.37 against 3.75 ± 1.31 among children treated medically, p = 0.08. In "DC" children with intracranial hypertension of vascular cause, POPC upon hospital discharge was 3.66 ± 0.51 against 5.66 ± 0.81 among non-operated children, p = 0.007. The schooling rate was higher among children treated medically for intracranial hypertension of traumatic cause, p = 0.012. The severity factors related with higher mortality identified in the population "traumatic intracranial hypertension" were mydriasis upon admission, a PIM 2 score higher and a lower temperature (<35.5°); the latter being the only factor identified for HTIC of vascular cause. In the case of traumatic intracranial hypertension, ICP monitoring in survivors was 40.35% against 28.6% in children died, with no significant difference. In the population "vascular intracranial hypertension", all the patients who died had not been monitoring PIC. Discussion The severity factors related with higher mortality identified in the population "traumatic intracranial hypertension" were mydriasis upon admission, a PIM 2 score higher and a lower temperature (<35.5°); the latter being the only factor identified for HTIC of vascular cause. Other studies have related other severity factors as initial Glasgow scale, tardive decompressive craniectomy. Conclusion Decompressive craniectomy doesn't seem to improve the mortality rate or the outcome in patients with hypertension of traumatic cause in our study but the DC traumatic subgroup was more serious than the subgroup treated medically. In children with refractory intracranial hypertension of vascular cause DC significantly improves survival and outcome. Further studies are needed to clarify the role of decompressive craniectomy and its timing in the therapeutic management of refractory intracranial hypertension.
Introduction Shortage of heart grafts is a major problem, leading to a significant mortality rate in the national waiting list, essentially for young children with low weight. The potential paediatric brain-dead donors often have myocardial dysfunction (MD), which seems to be reversible. The aim of this study is to assess prevalence, causes and consequences of MD when the potential paediatric donors are taken over, up to multi-organ retrieval, and the evolution after cardiac transplantation. Materials and methods This observational, monocentric, retrospective study included all brain-dead children aged 0-18 years old, who had their myocardial function assessed through a cardiac ultrasound performed by a cardiologist and identified from 2004 to 2016. All adult patients and those who didn't undergo a cardiac ultrasound were excluded. MD was defined as an LVEF ≤50% with or without abnormal segmented cinetic parameters.
The main evaluation criteria was the prevalence of MD in potential identified donors. The secondary evaluation criteria were the causes and consequences of MD on heart retrieval and the origin of this MD. Results Out of 40 included patients, 11 had MD. Prevalence of MD was of 30%. There was no significant difference between 2 groups regarding aetiology of brain death nor administration of catecholamines. Having a cardiopulmonary arrest during intensive care unit stay was associated with a significant risk of presenting a MD (p = 0.034). Having a MD had no consequences on organ retrieval in general (p = 0.25), but was significantly associated with a decrease in heart retrieval opportunities (p = 0.036). The cause of heart grafts refusal was a poor ventricular function in 50% of cases (3 cases out of 6). The cause for non-retrieval was parental refusal in one-third of cases. Evolution of the cardiac grafts was favorable in 16 cases on 18, one transplanted patient died (from a non-cardiac cause) and 1 patient was lost to follow up. Conclusion MD in paediatric brain-dead patients has direct consequences on heart retrieval and transplantation, and otherwise, organ shortage is a major ongoing problem. A better transplant management regarding hemodynamics (with the use of a protocol) could increase the number of heart transplants, especially in small children, and reduce mortality rate in national waiting list.

Introduction
The prone positioning (PP) is a strategy widely used in the treatment of severe forms of acute respiratory distress syndrome (ARDS) in adults. Its early use significantly reduces mortality (1). However, the studies do not strongly demonstrate its prognostic impact in pediatric ARDS. The aim of this study was to describe the prone positioning practices in the French-speaking pediatric intensive care units (PICU). Patients and methods This survey was conducted by email questionnaire to pediatric intensivists belonging to the French Society of Intensive Care Medicine and the French-speaking Group of Pediatric Intensive Care and Emergency Medicine. It was conducted From February to May 2016. The survey was addressed to doctors, nurses, physiotherapists practicing in PICU. It included 29 questions about indications, contraindications, techniques and medical devices used, and complications. Results One hundred and three persons answered (69 doctors and 33 nurses) which work in 28 french hospitals and 1 canadian hospital. Sixty-eight percent of interviewed persons have more than 5 years experience and 57% of them treat each year more than 10 children ARDS. Only 10% of the PICU have a PP medical protocol. Fifty percent of interviewed persons frequently use PP for the medical care of ARDS and 30% systematically use it. Thirty-six percent begin PP at the early phase of ARDS during conventional ventilation, while 42% before the introduction of unconventional ventilatory strategies (OHF); only 12% use it after the respiratory failure unless unconventional ventilatory strategies. Seventy-three percent report that PP is used with prolonged periods (>12 h/day), 22% with short periods (<12 h/day) and 14% with very long periods (>20 h/day). Regarding the weaning criteria, most of interviewed persons seem to use multiple and combinated criteria: 51% use hypoxemia severity parameters (PaO 2 /FiO 2 , PaO 2 , SaO 2 ), 39% use the oxygen level (FiO 2 ) and 38% use the mechanical ventilation parameters (PEEP, P max, P plate). Finally, despite a low level of scientific evidence in children, 87% of the persons gave a strong recommendation for PP as standard care in severe pediatric ARDS. See Fig. 16.

Discussion
The survey confirmed the widely use of PP in pediatric ARDS. However, no specific protocol is avalaible in most of the PICU. The timing of the PP beginning can be different according to children, early and prior to use of the conventional ventilation strategy in most cases. The duration of PP seems more consensual. Most of the centers use extended periods longer than 12 h/day. These results are close to Guérin et al. advocating a duration >16 h/day. Finally, the weaning is a great issue and depends on multiple criteria. In Guerin et al. (2) PP was interrupted if one of the following criteria were present: PaO 2 / FiO 2 ≥ 150 mmHg, with PEEP of ≤10 cm of water and a FiO 2 of ≤0.6; decreased PaO 2 /FiO 2 than 20%, compared to compared to the supine position, or the occurrence of complications. No study has validated PP weaning criteria during pediatric ARDS. Conclusion The prone positioning is a strategy commonly used in pediatric intensive care units for the severe pediatric ARDS. The criterias of implementation and timing are variable, as well as the weaning criterias. More pediatric multicenter randomized studies will be necessary to confirm the benefits of PP in pediatric ARDS and to define clear weaning criteria.
Introduction Allogeneic Hematopoietic Stem Cell Transplantation (HSCT) recipients have profound defects in every immunity compartments that can lead to severe opportunistic infections (OI). 20% of HSCT patients require admission to the ICU because of diverse infectious or non-infectious complications with dismal outcomes. OI specific course in this population has not been described previously and the management of these infections may be a concern. The aim of this study was to investigate risk factors, management and outcomes of IO in HSCT recipients admitted to the ICU. Patients and methods This was a retrospective (2007-2016) single center study of patients admitted to ICU after an allogeneic HSCT. Patients provided written informed consent according to Helsinki declaration. Data regarding the transplant, infections and life sustaining therapy use were analyzed. OI were considered if present at the time or during ICU admission. Results Hundred and ninety-four patients (Pt) were included. Median age was 43 [28; 58] years, 63.2% were males. Reason for transplantation was acute leukemia in 84 (42%) Pt and the hematological condition was still in complete remission at ICU admission in 83% of patients. 78 (41%) and 53 (28%) had received a myeloablative conditioning regimen and anti-thymoglobulin serum respectively. 40% had acute graft versus host disease over grade 2 at ICU admission. OI was documented in 38 patients (20%). An invasive fungal infection (IFI) was found in 19 Pt owing to 3 mucormucosis, 1 trichosporon septicemia and 15 invasive aspergillosis (5 possible, 9 probable and 1 proven according to EORTC criteria). Serum Galactomannane antigen was positive in 10 (67%). Median time from transplantation and ICU admission to IFI diagnosis was respectively 86 [18; 529] and −3 [−8; 0] days. Lung was involved in 90% and patients with aspergillosis were admitted to the ICU for acute respiratory failure in 71% (vs. 33% for others p = 0.2). They did not required invasive ventilation more frequently (36 vs. 31% p = 0.8). 32 and 10% required vasopressors and renal replacement therapy with no difference as compared to others. Median ICU length was 3 [2; 6] days. Demographic, stem cell source, and donor type were not associated with IFI occurrence in this population. However 15/19 had received a total body irradiation (78 vs. 47% p = 0.01). IFI occurrence was not associated with ICU or Day 90 mortality (33 vs. 26% p = 0.7 and 53 vs. 41% p = 0.3 respectively). A viral infection was found in 17 Pt owing to 11 CMV, 1 adenovirus, 1 HSV and 1 VRS infections. Analyses were focused on CMV reactivation. Median time from transplantation and ICU admission to CMV reactivation was respectively 111 [50; 151] and −23 [−38; −10] days. Reactivation was mainly positive blood PCR but 1 pt had CMV colitis. A preemptive treatment was started on the same day in median and lasts 34 [21; 71] days. Patients with CMV reactivation had more frequently multiple organ failure (50 vs. 22% p = 0.06) and higher ICU admission SOFA score (7 [5; 12] vs. 6 [3][4][5][6][7][8] p = 0.05). They trend to have higher admission creatinine serum level (132 [72; 157] vs. 81 [60; 120] umol/L, p = 0.2) and more frequently required emergency renal replacement therapy (36 vs. 7% p = 0.01) mechanical ventilation (64 vs. 30% p = 0.04) and vasopressors (73 vs. 37% p = 0.02). Median ICU length was 4 [2; 11] days and comparable to others. Demographic, stem cell source, conditioning regimen and donor type were not associated with CMV occurrence. CMV reactivation was not significantly associated with ICU or Day 90 mortality (36 vs. 26% p = 0.5 and 60 vs. 41% p = 0.3 respectively). Conclusion OI was found in 20% of allogeneic HSCT recipients admitted to the ICU. IFI were mainly responsible for respiratory distress and CMV associated to multiple organ failure. Non-invasive diagnostic tests were positives in a majority of these patients. In this cohort, IO treatment was started quickly after the diagnostic and we did not find an association with mortality. Intensivists should always consider OI in their diagnostic panel in this specific population.
Introduction Over the last two decades, targeted therapies in patients with solid tumors have both increased their length of survival and significantly altered their immune functions. However, data on opportunistic infections in this setting remain scarce. In this systematic review, we sought to identify published cases of opportunistic infections in patients with solid tumors, with a special interest on clinical findings, trends over time and outcomes. Materials and methods We performed a search of Medical Subject Headings (MeSH) on PubMed using the words pneumonia pneumocystis (PCP), invasive aspergillosis (IA), histoplasma, mucor, geotrichum, cryptococcus, coccidioidomycosis combined with the MeSH term neoplasms (breast, lung, ovarian, urologic gastrointestinal, digestive system, abdominal, brain, carcinoid tumor, sarcoma, testicular, seminoma). We identify published cases of opportunistic infections in non HIV patients with solid tumors between 01/01/1966 and 05/01/2016 included. Results Regarding Pneumocystis jirovecii pneumonia, 94 cases could be identified. There were 32 men and 62 women, aged of 57.5 (19-80) years. Underlying tumors were chiefly brain neoplasms (n = 31, 33%), lung neoplasms (n = 24, 26%) and breast neoplasms (n = 16, 17%). At the time of Pneumocystis pneumonia onset, 52 patients (55%) had a history of chemotherapy, 58 (62%) had received long term or high dose steroids, and 21 (22%) had an history of biotherapy targeting the malignancy. Of note, 18 patients (19%) had received only chemotherapy, 17 (18%) had received steroids alone, 4 (4%) everolimus therapy alone and 3 (3%) received none of these treatments. Regarding invasive aspergillosis 64 cases could be identified. Mean age was 66.7 (37-82) and 46 (72%) were men. Solid tumors associated with invasive aspergillosis were primarily lung neoplasms (n = 31, 48%) and brain neoplasms (n = 24, 38%). At aspergillosis onset, 38 (59%) patients had a history of chemotherapy, 29 (45%) were receiving long term or high dose steroids and 3 (5%) had received targeted therapy. Fourteen (22%) patients had received only chemotherapy, 2 (3%) only steroids, and 1 (1.5%) had received targeted therapy alone. For both infection, there was a trend for a higher number of reported cases throughout the studied period. Conclusion This systematic review provides objective data showing that an increased proportion of patients with solid tumors present with opportunistic infections. We are convinced that it is a clinically relevant but still neglected problem. Selected oncologic population may be becoming eligible for antimicrobial prophylaxis against Pneumocystis or Aspergillus.
care unit of Strasbourg in France. Patients were included only if they are non-immunocompromised according to the European Organisation for Research and Treatment of Cancer (EORTC). Invasive aspergillosis was defined as an association of microbiological evidence, a radiological imaging and a clinical context. Results Eighteen patients (13 males) were identified during the study period. The median of IGS II was 60.5 (interquartile range (IRQ), 49.25-78.75). Ninety-four percent was under mechanical ventilation. Fourteen (78%) patients were suffering from liver failure. Among liver failure, twelve (86%) were beforehand suffering from cirrhosis. The median MELD score was 42 (interquartile range (IRQ), 30-53). Sixty-four percent of aspergillosis were due to Aspergillosis fumigatus. Hundred percent were pulmonary aspergillosis. Fifty-six percent of aspergillosis were associated with bacterial pneumonia. The mortality rate at the date of the latest news (an average of 2 years) was seventytwo percent. Discussion Invasive aspergillosis is not exceptional in the non-immunocompromised patient especially in patient developing liver failure. An active research of colonization/infection with aspergillus in these patients remain to be discussed. Conclusion Invasive aspergillosis in ICU has a poor prognosis. The liver failure seems to be the most important risk factor in non-immunocompromised patients according EORCT criteria.
Introduction Chest wall elastance (Ecw) has been found to increase in prone (PP) as compared to supine position (SP) in ARDS patients [1]. This makes respiratory system elastance (Ers) not reflecting lung elastance (El). Little is known about the changes of Ecw, El and lung resistance (Rl) when moving the patient from the SP to the PP via the lateral position (LP). The goal of present study was to measure Ecw, El and Rl in ARDS patients in SP, LP and PP during the proning procedure. Patients and methods It was a prospective, single-center, controlled study. ARDS patients intubated, sedated and paralyzed with PaO 2 /FiO 2 ratio < 150 mmHg, PEEP ≥ 5 cmH 2 0 and an indication of PP were included. Mechanical ventilation was delivered in volume controlled mode with constant flow inflation and end-inspiratory pause 0.100 s included into the inspiratory time. Ventilator settings were unaltered during the procedure. An esophageal balloon catheter (Nutrivent device) was used for esophageal pressure (Pes) measurement. Pressure at the airway opening (Pao) and airflow were measured by Fleish 2 pneumotachograph proximal to endotracheal tube and upstream Heat and Moisture exchanger. Pao, Pes and airflow were continuously measured during 30 min in SP, then during 1 min in LP and 30 min in PP. The side for the lateralization was that selected by routine practice (in the opposite side from central venous line). Ers and resistance of the respiratory system (Rrs) were obtained by fitting flow and Pao signals breath by breath to the first order equation. Ecw and resistance of the chest wall (Rcw) were similarly obtained by fitting flow and Pes signals breath by breath to the first order equation pertaining to the chest wall. El and lung resistance (Rl) were obtained by subtracting Ers and Rrs from Ecw and Rcw, respectively. Our ethical committee approved the protocol. Data are shown as median (first and third quartiles). Comparisons between positions were made by using paired-t-test. Results Twenty-nine patients, 19 males, of 68 (62-74) years, SAPS 2 45 (34-55) and SOFA score 7 (3-9) were included 1 (0-2) days after ARDS criteria were met. The ARDS severity was moderate in 25 cases (86%) and severe in 4 (14%). Tidal volume averaged 5.9 (5.6-6) ml/kg predicted body weight, PEEP 10 (10-10) cmH 2 O, FiO 2 60 (50-70) %, PaO 2 /FiO 2 122 (108-139) mmHg. The cause of ARDS was pulmonary in 21 cases (72%), extra pulmonary in 3 (10%) and undetermined in 6 (18%). Lateral positioning was on the right side in 14 (48.3%) and on the left side in 15 patients (51.7%). The results are shown in the Table 21. Conclusion During prone positioning in ARDS patients, as compared to SP we observed a higher RL in LP and an increased Ecw in PP.
Introduction Neuromuscular Blocking Agents (NMBA) could exert beneficial effects in Acute Respiratory Distress Syndrome (ARDS) through properties on respiratory mechanics and particularly in modifying transpulmonary pressures (PL). Patients and methods Prospective randomized control study in moderate to severe ARDS patients within the first 48 h of the onset of ARDS. All patients were monitored by an esophageal catheter and followed during 48 h. Moderate ARDS patients were randomized in two groups according to the systematic administration of a 48 h continuous infusion of cisatracurium besylate or not (control group). The severe ARDS patients group received a 48 h continuous infusion of cisatracurium besylate. The evolution during the 48 h of the study of the oxygenation and the respiratory mechanics including inspiratory and expiratory transpulmonary pressures and driving pressure were assessed and compared. Delta transpulmonary pressure (∆PL) was defined as inspiratory PL minus expiratory PL. Results Thirty patients were included, 24 in the moderate ARDS group and 6 in the severe ARDS group. NMBA infusion was associated with an improvement in oxygenation both the moderate and the severe ARDS patients group accompanied by a decrease in both the plateau pressure and the total positive end expiratory pressure. The mean inspiratory and expiratory PL were higher in the moderate ARDS patients group receiving NMBA as compared with the control group (Fig. 17). In contrast, there was no modification of both the driving pressure and the ∆PL related to NMBA administration. Conclusion NMBA could exert beneficial effects in moderate ARDS patients through higher observed inspiratory and expiratory transpulmonary pressures.

Competing interests
None. Introduction Prone position (PP) is a major treatment in management of acute respiratory distress syndrome (ARDS). The use of PP in patients with severe ARDS associated with brain injury is at high risk of intracranial hypertension. The aim of this study is to analyze the effect of PP on intracranial pressure (ICP) and cerebral perfusion pressure (CPP) in patients with ARDS and acute neurological condition requiring monitoring of ICP. Patients and methods It is a retrospective descriptive study including sixteen patients with acute brain injury (subarachnoid hemorrhage, severe head trauma, and hemorrhagic stroke) and continuous monitoring of ICP who developed a severe ARDS during ICU stay from January 2012 to December 2015 and for which PP was performed. 29 PP sessions were analyzed. Hemodynamic and respiratory parameters, blood oxygenation, PIC and PPC were studied in supine, before PP and after PP. The study was approved by FICS ethic comity. Results A significant increase in PaO 2 /FiO 2 ratio was observed in PP, from 112 ± 34 to 200 ± 84 (p < 0.01). In PP, the ICP was increased 11 ± 1.1-17 ± 1.3 mmHg (p < 0.05) while the CPP was stable 74 ± 2 versus 76 ± 2 mmHg (NS). Median duration of PP session was 16 h (12)(13)(14)(15)(16)(17)(18)(19)(20). Increasing of ICP during PP required medical treatment in 8 sessions (27%). PP session was interrupted in 6 sessions (20%). In subgroup of patients who respond to PP in terms of oxygenation, the increase of ICP was lower than in non-responders (43 vs 118%) (p < 0.05). CPP was not modified whatever the nature of the response to PP (76 ± 15-77 ± 11 in non-responders and from 73 ± 10 to 76 ± 10 in responders (NS)) (Fig. 18). Discussion Our study shows an improvement of oxygenation during PP in severe ARDS patient with acute brain injury. We observe a constant increment of PIC during PP sessions. The increment of ICP is less in responders to PP. Significant increased ICP requiring an enhancement in the medical treatment was observed in 30% of the cases, and lead in most cases to a discontinuation of the session. Our data underlined the absolute necessity to monitor ICP during PP session in patients with acute brain injury and ARDS, even if ICP is controlled previously in supine. Only 3 prospective (1, 2) and one retrospective studies evaluate the effects of PP on ICP in patients with acute brain injury and acute respiratory failure (ARF). They results are similar to ours. In all these studies, the severity of ARF was often not well specified. Roth and al.

S63
(2) had included only 10% of ARDS in a population of patient with ICP not controlled. In others studies, monitoring of ICP during PP was not systematic.  Despite the retrospective nature of the study and the small number of patients, it is the only work studying the effects of PP on intracranial pressure in patients with acute brain injury at risk for intracranial hypertension and severe ARDS according to the Berlin's definition. Conclusion Our work suggest that PP is a quite secure technique for use for the treatment of severe ARDS even patients at risk of intracranial hypertension with a benefit in terms of oxygenation without major increase of ICP particularly in PP responders.
Introduction Influenza-associated acute respiratory distress syndrome (ARDS) requiring extracorporeal membrane oxygenation (ECMO) support is known to have a good prognosis (1). However, the incidence and impact of co-infection in this setting remain unknown.

Patients and methods
We conducted a retrospective, observational analysis of data prospectively collected from all patients admitted to our medical ICU who received ECMO support for influenza-associated ARDS between 2009 and 2016. Co-infection was defined as isolation of a pathogen in the lower respiratory tract at a significant level or in the blood during the 48 h following hospital admission. When no pathogen was identified in a patient receiving antibiotics prior to bacteriological sampling, an independent adjudication committee reviewed all charts to assess if the patient had a "high probability" or "low probability" for bacterial co-infection, based on clinical, radiological and biological results available. Results are presented as median [IQR].
Results Among the 116 patients hospitalized for an influenzaassociated infection in our ICU, 77 had an ARDS requiring support by either veno-venous-(VV, n = 58), venoarterial (VA, n = 15) or venoarterio-venous-(VAV; n = 4) ECMO. 3-29.9), pre-ECMO SOFA score >14 (OR 6.2; 95% CI 1.9-19.8) as independent predictors of hospital mortality, but not co-infection (OR 2.5, 95% CI 0.8-7.9). In a second analysis, patients with proven co-infection and high probability of co-infection were grouped and compared to patients with no co-infection and low probability of co-infection; and results were similar. As compared to others co-infected patients, those co-infected with a PVL-positive S. aureus had same characteristics and similar mortality rate, but all received a treatment active against PVL production. Conclusion Co-infection is frequent in patients with influenzaassociated ARDS supported by ECMO, occurring in roughly 50% of the cases. Mortality of patients with co-infection is higher than those without, but seems mainly due to the severity of the disease. S. aureus was the most frequently identified pathogen, with a high prevalence of PVL-positive S. aureus, Infection with a PVL-positive strain was not associated with a poorer outcome as compared to other co-infections. Whether a treatment active against PVL production should be given in those patients remains to be determined.

Competing interests
None.

Introduction
The pancreaticoduodenectomy (PD) is major surgery in visceral surgery. This technique performed for the first time in 1935 by Whipple has seen much progress and development over the years that have enabled a significant reduction in mortality, while the morbidity remains high. The aim of this study was to analyze postoperative morbidity pancreaticoduodenectomies.

Patients and methods
We retrospectively studied 34 cases of cephalic duodenopancreatectomy at the department of surgical emergencies resuscitation (wing 33) spanning 6 years, between January 2010 and December 2015.

Results
The average age of patients was 58.2 years with 43% of females and 57% of males, the frequence of pancreatic resections was 5 years. The indications of cephalic duodenopancreatectomy were: tumors of pancreatic head (65%), Ampulla Vater (23%), duodenum tumors (12%). The restoration of continuity after cephalic duodenopancreatectomy was realized with a rate of 41% for pancreaticogastrostomy and 59% for pancreaticojejunostomy. The average hospital stay was 11, 7 days, with extreme lengths of 6-27 days. The postoperative course was marked by the occurrence of 8 deaths (24%), the morbidity rate was 29, 25% after PJ and 28% after PG; the most frequent complications were the pancreatic fistula (40%), the postoperative peritonitis (40%), the digestive bleeding (20%), the gastroparesis (20%). Conclusion Advances in the overall care of patients by surgical teams, anesthesiologists and intensivists, the DPC mortality is currently low in experienced centers. The multidisciplinary, involving surgeons, radiologists and especially intensive care, to manage more effectively the complications of this surgery remains burdened with high morbidity.
Introduction Severe Acute Pancreatitis (SAP) is a common but potentially lethal pathology due to the multiplicity and severity of complications that can occur at all stages of evolution. In the last decade, mini-invasive interventional treatments of infected pancreatic necrosis (IPN) have been developed. The aim of the present study was to assess the management and outcomes of SAP patients, as well as to identify the role of IPN.

Materials and methods
This was a retrospective study of prospectively collected data from all consecutive patients admitted in Intensive Care Unit (ICU) in a single French center (Hospital of Nantes) from 2012 to 2015. Using logistic regression, we evaluated the association between IPN and patients characteristics at baseline and the outcomes.  (Fig. 20), highlighting the prognostic importance of respiratory failure and acute renal failure at the time of LT, as well as complex interactions between donor and recipient features.

Discussion
Conclusion Ventilator support and/or acute renal failure at the time of LT are major predictors of mortality but complex recipients/donors relationships may moderate these associations, as demonstrated by our CART analysis.
Conclusion The results of our study are similar to those reported in medical literature. Preoperative evaluation and nutritional rehabilitation are crucial to improve patient's outcome and reduce morbidity and mortality after gastrectomy for cancer.

Introduction
The mesenteric ischemia is a condition relatively rarely. It is marked by high mortality. Mortality is primarily related to the land on which ischemia occurs and especially the time taken to diagnose. This delay is due to the low specificity of clinical signs and the absence of diagnostic laboratory test. The mesenteric ischemia remains a diagnostic and therapeutic challenge. Patients and methods Twenty cases of acute mesenteric ischemia have been collected at the surgical resuscitation (resuscitation 33) at the hospital center IBN ROCHD of Casablanca from January 2010 to December 2015.
Results The mean age of our patients is 50 year old. It is about a disease that the incidence increases these last years, particularly because of the waxing number of old patients and/or suffers from advanced cardiovascular diseases. The cardiovascular risk factor has been present in 35% of our patients.
The abdominal pain has been present in all the patients. It is a sudden, intensive pain localized the most often at the level of the epigastria, becomes diffuse in few hours or even few days. Other clinical signs have been described as the bilious vomiting that becomes fecaloid after few days. The digestive hemorrhages as the moelena and the hematemeses. A stop of the matter and the gazes was noticed in 25% of our patients. The absence of specificity of the clinical signs forced the realization of complementary examinations. The scanner becomes the reference imaging. It permits a differential diagnosis, the search of direct signs of vascular obstruction and the emphasis of intestinal pain. Four etiologies are noticed: The arterial occlusion by emboli (40%), the arterial thrombosis (20%), the venous thrombosis (25%) and the "non occlusive" form (15%). The strategy of management of the acute mesenteric ischemia is multidisciplinary, based on the equips of radiology, vascular surgery and/ or visceral surgery and resuscitation. The treatment consists in measures of general resuscitation, the techniques of endoluminal vascular disobstruction and techniques of surgical revascularization.
In spite of the improvements in the diagnosis and the therapeutic procedure of the IMA, the disease still know a rate of mortality between 60 and 97% according the studies. In our study, we noticed 8 cases of death (40%), 9 cases of good recovery (45%), 3 cases are unknown evolution (15%). Conclusion It is a vital emergency that the evolution still knows great mortality. It is very important to remind the acute mesenteric ischemia in the case of any acute abdominal symptom in order to anticipate about the natural evolution and to act in a reversible stage of the ischemia.

Competing interests
None. Introduction Emergency Departments staff are frequently exposed to many complex stressful situations and consequently burnout syndrome. Our study aimed to describe epidemiological particularities and determine the risk factors of burnout syndrome in different categories of emergency. Patients and methods We studied five academics and four regional hospitals. The level of burnout was assessed using the "Maslach Burn Out Inventory" score and the degree of depression with Major Depression Inventory (MDI) test.

Introduction
Results One hundred and forty-three correctly completed questionnaires were collected. The mean age of study population was 32 ± 7 years. Sex-ratio was at 0.63. Fifty-one per cent of the care staff were married. Physicians represented 40% and paramedical 60%. The general frequency of Burnout syndrome was 85% (n = 127). Low level Burnout was present in 32%, moderate level in 45% and high level in 23%. The depression frequency was 52%. A statistically significant correlation was found between burnout and depression firstly (p = 0.006) and between burnout and lack of equipment (p = 0.04). Their relative risk was 4.7 [1.5, 15] and 9.14 [2.2, 37.6] respectively). Main risk factors associated with high level burnout are detailed in Table 22. Conclusion Burnout syndrome frequency in our emergency departments is alarming. Helping to resolve social and psychological problems and improving work conditions may help to decrease it.

Introduction
The healthcare activity is recognized as a major polluting activity. In France, it generates 800,000 tons of waste cremated each year, and represents 12% of the tertiary energy consumptions. In the United States, it generates 7000 tons of waste per day and 7% of total CO 2 emissions in 2007 were attributed to him. Ultimately, such waste production is associated with adverse environmental and health effects. Nevertheless, near half of the hospital waste would be recyclable, particularly in our intensive care units (ICU) [1]. Furthermore, sustainable development solutions generate profits. The aim of this study is to make an overview of waste produced in a ICU and offer solutions to conserve natural resources and reduce the carbon footprint bound to the healthcare activity. Materials and methods Experimental study, single-center, concerning a period of 6 months in an ICU-high surveillance unit compound of 16 beds. We have identified all waste generated. Our packaging were given to the recycling company in connection with the hospital. Then we have studied the impact of the implementation of sustainable development solutions.
Results Firstly, we have studied the non-recycled waste and the quantity produced over a period of 1 month. Approximately 8 kg of waste is produced per patient per day with 45% of infectious waste and 55% of general waste. These results were linked with a bad distribution of garbage bags in the rooms (130 L of infectious waste versus 50 L of general waste). Secondly, we have improved our way to sort and consume and we have created recycling dies without compromising patient safety. All these measures have not increased workload. Changing bags in the rooms (20 L of infectious waste and 2 bags of 50 L of general waste) allowed to reach the normal goals of 2 sectors with a net benefit estimated at 4500 euros per year. The medical broken glass containing drugs was thrown into plastic containers of 5 L for infectious waste to prevent the risk of cuts. By creating a specific die intended to the general waste, we could quantify the production of this glass to 10 kg per week and to spare the use and the incineration of 350 containers of 5 L per year (global economy of 1000 euros). Plastic packaging represented an important proportion of the cremated waste. We have created 3 sectors of recycling including the polypropylene (80-200 kg per month), the polyethylene colorless and colored polyethylene. This plastic is sold to be recycled without additional cost for the hospital. The linerboards was cremated. We have created a recycling die (50 kg per month). This sector was subsequently extended to the entire hospital structure, particularly the pharmacy that produces 12 containers of 400 L per month. They are now sold without additional cost. Many unnecessary plastic waste is generated daily. We have removed using mild soap plastic bottles of 30 ml by using the same mild soap in pump of 500 ml (economy of 1000 euros). The use of 100 L plastic bags for the transitional deposit of linen has been deleted (economy of 450 euros). Concerning the paper: 100% of the impressions were made in simplex. Printers were parametrized on both sides by default allowing the economy of 60 reams per year (30,000 sheets), several thousand liters of water and the reduction of CO 2 emissions. Discussion Recycling is only one component of the sustainable development in health. Other avenues that could be considered to improve ICU sustainability would include examining water use (for linen), electricity use (reducing non-essential use at night…). Beyond these actions, we need to encourage our suppliers to turn to sustainable and recyclable packages to reduce the use of polluting and depletable fossil fuels such as oil. But also to develop with them circular economies where waste is returned to them to be reused. Conclusion We must ask the question also resuscitate our tons of waste. Our ICU produce large quantities of waste (over 2 tons per year per bed). However, a significant proportion, especially plastic, is recyclable with a significant environmental and financial benefit. Waste management also requires an optimal and rational use of supplies because "the best waste is that which is not produced" and that excess is not a guarantee of quality.
As already said St Exupéry in 1939: "We do not inherit the Earth from our parents, we borrow it from our children. " So do not expect tomorrow to reduce major adverse ecological impact paradoxically generated by a great profession whose ultimate goal is to cure people.  Moreover, an external consultant is rarely applied and palliative cares are insufficiently developed after «non-readmission» decisions. For providing corrective measures, this study lead to propose a «nonreadmission» process by integrating the discussion for a real «patient's care project» at the end of the ICU hospitalization. This process would lead to collect patient's opinion through advance directives, to ensure a collegial discussion including an external consultant and to allow reevaluation of global patient's clinical status and one or more organ failure(s). Then, «non-readmission» decisions would be integrated in a therapeutic project which would promote the initiation of a palliative care program if necessary. The purpose of this process is well to respect patient's autonomy and dignity as required by French law and medical ethics.

Introduction
The proportion of elderly patients is steadily increasing. Due to the growth of this part of the population who suffer from multiple pathologies, the need for hospitalization in intensive care increases. According to the simulations, the proportion of octogenarian patients in ICU will increase reaching the third of ICU patients. While chronological age is not a significant factor of poor prognosis in the ICU (1), many factors should be taken into account to evaluate the relevance of ICU admission in the senior population and withholding such intensification should be consensually discussed between clinicians and obviously as often as possible with the patient himself (2). The aim of the study was to assess the role of stakeholders (ward physicians, intensivists, family doctor and patient himself ) in the decision of withholding ICU admission for elderly patients in our internal medicine department.

Patients and methods
We made a prospective observational monocentric study, including all the elderly patients (defined as older than 75) admitted in the internal medicine department from January 2012 to June 2012. The only non-inclusion criterion was patient's refusal to participate to the survey. Collected data involve physiological (cognitive, autonomy, nutritional status), morbidities (acute and chronic diseases) and social parameters (marital status, relatives). And evaluation of quality of life by the patient himself using an analog visual scale was also obtained.
Internal medicine physicians were asked to report any ICU withholds decision for their patients. In absence of notification, every physician was questioned again the day of the concerned patient's discharge.
Results One hundred ninety-one patients were included between January and June 2012. Factors associated with a significant reduction of in hospital mortality were higher age (P = 0.002), higher lactate level (P = 0.004), chronic obstructive pulmonary disease (P = 0.04), diabetes mellitus (P = 0.03), immunodepression (P = 0.05) and respiratory failure (P = 0.02). Conclusion In patients hospitalized for VS high body mass index, low left ventricular systolic function, high white blood cell count, low creatinine clearance, high lactate level and ST-segment depression are the variables correlating significantly with high-sensitivity troponin-T concentrations. Peak of hsTnT was not significantly associated with in-hospital mortality in this setting.
Introduction Mitochondria are evolutionary endosymbionts that are derived from ancestral aerobic bacteria and so might bear and release bacterial molecular motifs supporting the role of mitochondria in danger signal regulations. Free circulating mitochondrial DNA (mtDNA) is elevated in a wild range of critical illness observed in intensive care units, and is associated with bad outcomes and mortality. The mtDNA is a molecular pattern that belongs to mitochondrial damage associated molecular patterns (mtDAMPs), and can interact with pattern recognition receptors (PRR) to induce self defense reaction. Free mtDNA activates inflammatory signaling pathways through Toll-like endosomal receptor 9 (TLR9) interactions. Nevertheless, new evidence advocates a role of the receptor for advanced glycation end-products (RAGE) in mtDNA signaling. Experimental data suggest a role of mtDNA-PRR interaction in systemic inflammation and organ dysfunctions as septic acute kidney injury or pulmonary inflammation. Impact of free circulating mtDNA on endothelial cell is not known.
The main purpose of this study was to test whether mtDAMPs and mtDNA can induce endothelial dysfunction. We also evaluated the role of mtDNA-RAGE axis in mtDAMPs induced endothelial dysfunction.

Materials and methods
Mitochondria were isolated from livers of wild type C57B6 mice. Isolated mitochondria were sonicated on ice to obtain mtDAMP preparations. Semi quantitative evaluation of mtDAMP content was tested by qPCR, with specific markers of mtDNA (cytochrome B (cytb), NADPH oxidase (ND4)). Intraperitoneal injection of 1 mg of mtDAMPs was used as experimental model in wild type and RAGE KO mice, as previously described [1]. The mtDAMPs were also administrated after ex vivo DNAse preparation. Endothelial function was assessed with a Mulvany-Halpern style myograph, 4 h after mtDAMP administrations on aorta (conductive vessel) and on 2d division of mesenteric artery (resistive vessel). Endothelial-dependent relaxation was studied by cumulative expositions of the vessels to acetylcholine (1.10-10-1.10-5 M). Endothelial-independent relaxation was studied by sodium nitroprussiate exposition. Results The mtDAMPs preparation contains a high quantity of mtDNA with a 1/cycle threshold (Ct) ratio of 0.062 for cytb expression. Intraperitoneal administrations of mtDAMPs induced a decrease of endothelial-dependent relaxation mainly on conductive vessel (p = 0.0079, n = 5 per group) and to a lesser extent on resistive vessel (p = 0.22, n = 5 per group). RAGE-KO mice were protected from mtDAMPs-induced aorta dysfunction (p = 0.53, n = 5 per group). The ex vivo exposition of mtDAMPs to a DNAse preparation decreased mtDNA content in mtDAMPs solution with a 1/Ct ratio of 0.038 for cytb expression. Eventually, the pretreatment of mtDAMPs with a DNAse preparation prevented the mtDAMPs-induced aorta dysfunction (p = 0.85, n = 5). Discussion More than prognostic markers, mtDAMPs particularly mtDNA seems implicated in endothelial dysfunction in critically ill patient. New evidence suggest RAGE interaction in endosomal TLR9 pro-inflammatory and pro-oxidant response to mtDNA [2]. Also in sepsis, physiological clearance of circulating DNA might be impaired, this results comfort the possibility of therapeutic regulation of free circulating mtDNA to prevent septic organ dysfunction related to mtDAMPs accumulations. Conclusion Exogenous mtDAMPs can induce endothelial dysfunction in mice. The mtDNA-RAGE axis is a key component of the signaling pathway involved in this dysfunction.

Introduction
The use of dynamic parameters to assess fluid responsiveness was supported by cyclic changes in stroke volume induced by mechanical ventilation. However, these parameters have several limits. Venous to arterial carbon dioxide difference inversely related to cardiac index. Consequently, fluid administration would be beneficial if carbon dioxide gap increases. Objective To investigate whether carbon dioxide gap predicts fluid responsiveness in patients with acute circulatory failure. Patients and methods We conducted a prospective study in the medical intensive care unit of Hospital Taher Sfar at Mahdia, between March 2013 and April 2016. Patients with circulatory failure and who required mechanical ventilation were included. We measured the variation of cardiac index between baseline and after volume expansion of 500 ml of saline fluid. The PiCCO2 was used to measure cardiac index. Response to fluid challenge was defined as a 15% increase in cardiac index. Before and after fluid administration, we recorded carbon dioxide difference and hemodynamic parameters. Results Among 68 included patients, 33 (49%) were responders. The causes of acute circulatory failure were septic shock (n = 45), cardiogenic shock (n = 11), and Hypovolemia (n = 12). Carbone dioxide gap was significantly higher in responders group (8 ± 7 vs 4 ± 4 mmHg, p = 0.019). The area under the ROC curve for carbon dioxide gap was 0. 68 (95% CI 0.55-0.80). The best cutoff value was 6 mmHg (Sensibility = 46%, specificity = 80%, positive predictive value = 52% and negative predictive value = 60%). The area under the ROC curve for delta carbon dioxide was 0.54 (95% CI 0.4-0.68). Conclusion In this study, baseline carbon dioxide gap was not universal indicator to predict the fluid responsiveness in patient with circulatory failure.
Introduction Supraventricular arrhythmia (SVA) is commun in intensive care unit (ICU). Its incidence seems to be higher in patients with sepstic shock. Sepsis-associated myocardial dysfunction promote the occurrence of SVA by constituting an arrythmogenic substrate or under the effect of inotropic drugs. The aim of this study is to assess the incidence and prognostic impact of SVA in patients with septic shock. Patients and methods We retrospectively studied all patients with new onset SVA suffering from septic shock in non cardiac surgical ICU. Myocardial dysfunction was evaluated by transthoracic echography (TTE) after an adequate cardiac resuscitation using intravenous fluids expansion and adjunctive vasoactive agents. SVA was detected by the electrocardiogram scope. During the study period clinical and biologic characteristics, hemodynamic tolerance (vasopressors doses, arterial pressure changes), current treatment (such as corticoid), duration of mechanical ventilation, duration of vasopressor requirement and hospital mortality were collected. Results Sixty patients were included in the study. The SVA occurred in 28 patients, with an incidence of 46%. The median time to onset was 2 days. Cardioversion was performed for 15 patients with an effectiveness of 80%. Clinical and biological characteristics were similar between the groups with and without SVA: SAPS 2 and SOFA score at the beginning of septic shock, the existence of ARDS and cardiac biomarkers (NT-proBNP, troponin). However, renal failure and the use of corticoid in septic shock were more frequent in the group with SVA. The maximum doses of vasopressor agent were not significantly different between the groups with or without SVA. Myocardial dysfunction in sepsis defined by the left ventricle ejection fraction (LVEF) less than 45% (or the need for inotropic drug for LVEF > 45%) was not associated with the occurrence of SVA (+SVA group: n = 8; −SVA group: n = 6; p: 0.57). SVA was poorly-tolerated, observed by a significant decrease in mean arterial pressure and a significant increase in norepinephrine doses within 1 h of the start of SVA. The occurrence of SVA was associated with longer duration of use of vasopressor agent and a longer duration stay in ICU (+SVA group: 21 days, −SVA group: 14 days; p = 0.03). There was no difference in duration of mechanical ventilation and hospital mortality between the two groups. Conclusion The occurrence of SVA is common in septic shock, poorly tolerated hemodynamically and associated with longer duration stay in the ICU and vasopressor need. Sepsis myocardial dysfunction isn't necessarily associated to the occurrence of SVA.
Introduction A short term beneficial effect of prone position on cardiac index has been shown in 50% of 18 ARDS patients, and was related to an increase in cardiac preload in preload responsive patients (1). The aim of this study was to evaluate the long term hemodynamic response to prone position in a larger series of ARDS patients. Patients and methods Single center retrospective observational study performed on ARDS patients hospitalized in a medical ICU between July 2012 and March 2016. Patients included were adults fulfilling the Berlin definition for ARDS, undergoing at least one prone position session, under hemodynamic monitoring by the Picco ® device, with availability of hemodynamic measurements performed before (T1), at the end (T2), and after the prone position session (T3). Prone position sessions were excluded if they were performed >7 days after ARDS onset. The following variables were recorded: demographic, SAPSII, ARDS severity and risk factor, SOFA score and cumulative fluid balance at PP onset, delay between ARDS session and PP session, hemodynamic, arterial blood gas, ventilatory settings, plateau pressure, catecholamine dose and additional treatments. Statistical analyses were performed using prone position session as statistical unit and mixed models taking into account both multiple prone position sessions by patient and multiple measurements during a prone position session. p < 0.05 was chosen for statistical significance. Data are expressed as mean ± standard deviation. Results 85 patients fulfilled the inclusion criteria over the study period, totalizing 149 prone position sessions (2 ± 1 sessions per patient). Patients' age was 65 ± 11 y, 67% were male, 75% fulfilled the criteria for severe ARDS, and SAPSII at ICU admission was 64 ± 17. ARDS risk factors were pneumonia in 63 (74%), aspiration pneumonia in 26 (31%), and sepsis in 7 (8%) patients. Duration of prone position sessions was 16 ± 3 h. Hemodynamic measurements were performed in PP 13 ± 3 h after PP session onset. At session onset, SOFA score was 15 ± 4, and cumulated fluid balance was 2.5 ± 7.1 L. Vasopressor were used in 87%, inhaled nitric oxide in 22%, and neuromuscular blocking agents in 91% of the sessions. Hemodynamic and respiratory parameters before, during and after the prone position sessions are reported in Table 24. Cardiac index increased by at least 15%, decreased by at least 15% or remained stable in 35 (24%), 36 (34%), and 78 (52%) of the sessions, respectively. As compared to both other groups, PP sessions with significant increase in cardiac index had the following significant differences at T1 by univariate analysis: lower cardiac index, lower global end-diastolic volume, lower cardiac function index, and lower vasopressor dose. Multivariate analysis is under investigation. Conclusion Prone position is associated with an increase in global end-diastolic volume, reversible after return in supine position that may explain the positive effect of PP on cardiac index observed in ¼ of the PP sessions. Introduction Make sure that our patient have a good circulatory condition is a daily challenge for the intensivist. One of the therapeutics is fluid and one of his purpose is to increase venous return and then cardiac output. In order to examine that, There are several tools as the transthoracic echocardiogram wich allows the visualisation and the study of the respiratory variability from the inferior vena cava (IVC). Unfortunately there are some situations where the IVC visualisation is difficult (obesity, gut surgery, emphysema). The IVC is easily seen by a transhepatic ultrasound in her retrohepatic section. We make the hypothesis that the shape of the IVC could be predictive of fluid responsiveness.

Patients and methods
We have performed fluid challenge in patients under mechanical ventilation. The need for fluid therapy is the intensivist in charge decision. We performed a echocardiogram and we take two measures of the ICV: major axis and minor axis, the ICV is measured avec the sus hepatic vena. A elastometry index (EI) is determined which is the ratio of minor axis to minor axis. The fluid challenge is 250 ml of isotonic saline then we perform a new echocardiogram. A Tag is written on the patient to take the same ultrasound slice. We retain one increase of 15% of the cardiac index (IC) as a success of the filling. We exclude the presenting patients A right cardiac insufficiency, an arrhythmia and/or a HTAP. The statistical analysis is realized with the software R. Results Between August, 2015 and January, 2016 we included 49 patients. The average age is of 59 years (24-81), IGS2 of 48 (20-84), Ejectionnal Fraction of 48% 35-75) and the S wave tricuspid is 14 (9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21). The causes of the filling were an oliguria (31%), a low blood pressure (20%), a low cardiac output (30%), a hyperlactatémia (9%) and an other cause in 10% of the cases. We find a positive correlation between the EI and the increase of the IC, also for the area of the VCI and the respiratory variations of the VCI (p0.05) the other variables are not predictive (BP, E/e' , E/A). The data are summarized in the picture 1. ROC curves has been established (  Only 16% of the journals studied required authors to use STARD. A high Impact factor and the year of the study were the items associated with a better SQS the presence of a conflict of interest was associated with a lower SQS in univariate analysis. A higher impact factor (> 5), was the only independent factors statistically significantly (p = 0.037) associated with higher SQS in a multivariate regression model. Discussion Our study showed that the SQS were very low. Assessment of a study depends on quality of reporting. Blindness and participant sampling are the cornerstone to evaluate such bias as spectrum, verification, review and selection bias of a study, and were unfortunately scarcely reported compared to existing data in diagnosis accuracy reporting. One of the limitation is the 5 years sample of the study. We have planned to continue the analysis for a 10-year review starting just after the 2003 STARD publication. Conclusion Our study showed that several items remain poorly reported. We recommend systematic use of STARD criteria in the elaboration and reporting of future studies that evaluates the preload dependence.
Introduction Neurological impairment, i.e. encephalopathy, is commonly observed in patients with decompensated cirrhosis and/or portosystemic shunts admitted in ICU. Often ascribed to high plasmatic levels of ammonia, encephalopathy could also be induced by drugs or infection, due to altered blood-brain barrier (BBB) permeability. This latter setting is often underdiagnosed and encephalopathy related to hyperammonemia (so called hepatic encephalopathy-HE) being pointed out as the culpit of all neurological symptoms in cirrhotic patients. Quinolones and betalactamins were recently found in the cerebrospinal fluid of HE patients and it has been shown that the expression of efflux pumps, responsible for drugs passing through the BBB, was altered in animal models of HE. The purpose of this study was to assess the incidence of neurological impairment, i.e. encephalopathy, in cirrhotic patients hospitalized in  Discussion Overall, we reported a higher rate of lumbar puncture than those reporting in others studies concerning status epilepticus. Furthermore the rate of 16% of pleocytosis directly linked to status epilepticus is slightly higher than in most studies. Unfortunately we didn't realize a second lumbar puncture to assess the pleocytosis normalization during the days following the first lumbar puncture. The pathophysiological hypothesis of this phenomenon may be that prolonged/repeated seizures during status epilepticus would induce a blood-brain barrier dysfunction thereby favoring a cerebrospinal pleocytosis. Conclusion In our study, 16% of status epilepticus without infectious or neoplastic origin had a cerebrospinal pleocytosis directly linked to status epilepticus. This pleocytosis was significantly associated with myoclonic seizures and blood leukocytosis. These data may help to interpretation of cerebrospinal fluid pleocytosis during status epilepticus.
Introduction Neurological prognostication from cardiac arrest survivor is a current concern. EEG patterns and NSE dosage are two important prognostic factors. NSE threshold for prediction of poor outcome appear controversial, in part, because of variability in dosage timing and measurement techniques. Synek Score is routinely used in our center to classify comatose patients in post cardiac arrest. The aim of this study was to assess the prognostic value of NSE and Synek classification to predict poor neurological outcome. Introduction Traumatic Brain Injury (TBI) is a major public health problem. It is the leading cause of death and disability in young subjects. One of the principles of the TBI management is prevention of secondary cerebral insults including maintaining perfusion and cerebral oxygenation, control of intracranial pressure (ICP). An increase in ICP above 20 mmHg is associated with poor outcome. Cerebral hypoxia can occur with normal level of ICP and cerebral perfusion pressure (CPP).Monitoring of regional partial pressure of brain tissue oxygen (PbtO2) is a safe and reliable method for measuring cerebral oxygenation.

Patients and methods
A retrospective single-center observational study was conducted between January 2012 and December 2013, aimed to study the influence of PbtO2 with severe TBI patients outcome at 6 months through Glasgow Outcome Scale (GOS). The hourly values of ICP, PbtO2 and CPP were recovered on daily monitoring sheets. We compared two groups according to their GOS. During the study period, 66 patients underwent a monitoring ICP and PbtO2. Results The mean age was 33.7 ± 14.1 years. 78.8% were men. The initial Glasgow score was 6.6 ± 3.7. The mean Simplified Acute Physiology Score (SAPS II) was 43.4 ± 11.2 and Injury Severity Score (ISS) 30.5 ± 9.7. At 6 months, 7 patients had died (GOS1). Forty patients had a good outcome: GOS 4-5 (Group 1). Sixteen patients had poor outcome: GOS 2-3 (Group 2). In Group 2, there are significantly more PbtO2 hourly values below 10 mmHg at Day0 (5.5 ± 6.4 vs 2.3 ± 3, 2 in Group 1, p = 0.025); and more PbtO2 hourly values greater than 20 mmHg at Day0 (8.4 ± 7.0 vs 4.8 ± 4.9, p = 0.045). Conclusion PbtO2 less than 10 mmHg or greater than 20 mmHg at Day0 is associated with poor outcome at 6 months in the severe TBI. The PbtO2 allows a more individual approach of monitored TBI.

Competing interests
None.
Introduction Organ donation in patients after a decision to withdraw life-supportive therapies (WLST) (Maastricht 3 condition: M3) have been performed in our hospital since May 2015. We report here main characteristics of donors, data on M3 procedure and results on renal transplant recipients. Patients and methods All potential donors were included in a survey from May 2015 to June 2016, according to the French national M3 protocol defined by the French Organ Procurement Agency (Agence de la Biomédecine:ABM) [1].The demographical, clinical and biological characteristics of the donors, the different deadlines and times of the protocol and data of renal transplantation were collected and analyzed.
Results 28 patients had inclusion criteria. Patients were admitted in intensive care unit for cardiac arrest (68%), strokes (14%), traumatic brain injury (14%), ARDS (4%). Of them, 15 procedures (54%) were stopped (6 refusals of organ donation, 4 medical contra-indications discovered with additional exams, 1 failure of vessel cannulation, 4 deaths more than 3 h after extubation). 28 kidneys were harvested and 26 transplantations performed (1 renal cancer discovered during procurement surgery).The characteristics of the donors, deadlines of the protocol and transplant recipients are reported in the Table 26. Conclusion The French programm Maastricht 3 offered a new possibility of organ donation in our hospital. Thanks to these donors, the number of renal grafts increases and the preliminary results on transplant recipients are encouraging in line with the preliminary report of the ABM. Nevertheless, it is necessary to follow the transplant recipients and extend the procedure to new centres.

Discussion
In this study, we found some relevant risk factors for microaspiration (age, low score at GCS) consistent with literature on the subject. Patients with paralytic agents had less GAM which may be due to higher PEEP, higher cuff pressure and less enteral nutrition because of the severity of the underlying diseases. Conclusion This study did not show any increased risk of microaspiration in intubated COPD patients, whatever stage of COPD.
Introduction Protected specimen brush (PSB) is considered to be one of the standard methods for the diagnosis of ventilator-associated pneumonia (VAP). To our knowledge, there is no study assessing effect of prior antibiotherapy on direct examination, bacteriological culture and concordance of direct microscopy and culture. Patients and methods All consecutive episodes of suspected VAP were retrospectively evaluated between January 2014 and December 2014 in a 20-bed intensive care unit. Patient's characteristics and preexisting conditions were abstracted from the medical charts. After assessment of VAP probability using the clinical pulmonary infection score (CPIS), PSB were performed in patients with a CPIS of 6 or more. Based on antibiotic treatment in patients when bacteriological specimens were obtained, two groups were defined: no antibiotic group and antibiotic treatment started before PSB group. Two independent bacteriologists retrospectively reviewed direct examination and culture of PSB to assess bacteriological concordance, defined as non-concordant when direct examination and culture were different, concordant when direct examination and culture were similar and partially concordant when either direct examination or culture were comparable but with other microorganisms lacking in one or the other method. Results During this 12-months period, among 73 mechanically ventilated patients, 116 episodes of suspected VAP with PSB were evaluated. We found 60% of PSB (n = 70) performed without antibiotic treatment and 40% of PSB (n = 46) performed under antibiotherapy. We found no significant differences in patient's demographics, characteristics, and severity between both groups. Patients received antibiotics for the following reasons: aspiration pneumonia (n = 12), peritonitis (n = 8), VAP (n = 7), community-acquired pneumonia (n = 4), septic shock of unknown origin (n = 4), pyelonephritis (n = 3), meningitis (n = 2), acute pancreatitis (n = 2) and others (n = 4). The median duration of mechanical ventilation in the antibiotic receiving group and in the group without antibiotics was 7.5 days (IQR; 5-12 days) and 9 days (IQR: 5-22), respectively. When PSB was performed under antibiotic treatment, direct examination was positive in 26% (n = 12), culture was positive in 28% (n = 13) and those methods were concordant, non concordant and partially concordant in 93% (n = 43), 4% (n = 2) and 3% (n = 1), respectively. On the other hand, when PSB was performed without antibiotics, direct examination was positive in 65% (n = 46), culture was positive in 62% (n = 43) and those methods were concordant, non concordant and partially concordant in 71% (n = 50), 18% (n = 12) and 11% (n = 8), respectively. In univariate analysis, we found a significantly higher proportion of negative direct examination and negative culture in the antibiotic group (p > 0.001). Moreover, these methods were significantly more frequently concordant (p = 0.01), with a higher rate of both negative microscopic exam and culture when compared to the no antibiotic group (76%, n = 33 vs 38%, n = 18). Surprisingly, among the 13 patients previously treated with antibiotics with positive culture, 38% (n = 5) of the microorganisms showed antibiotics sensitivity. Discussion Whether prior antibiotic treatment may induce false negative of false positive treatment is a well-recognized phenomenon, the precise effect of antibiotics on direct examination and quantitative culture is not well assessed in VAP. Moreover, despite recent development of clinico-radiological score, diagnosis of VAP remains difficult, with no gold-standard. Therefore, bacteriological guided therapy is of particular importance. We found PSB realization under antibiotic treatment is associated with a lower rate of positive direct examination and culture and suggest performing these bacteriological samples without antibiotherapy. Some authors have suggested lowering the diagnostic threshold point of this bacteriological technique in order to preserve its accuracy. However, we can postulate that microorganisms responsible of superinfection in mechanically ventilated patients treated with antibiotics may be resistant and therefore the PSB could be positive. Conclusion In patients with a high pre-test probability of ventilatoracquired pneumonia, recent introduction of antibiotics significantly reduced the diagnostic accuracy of protected brush specimen by reducing rates of positive direct examination and culture. Further studies should evaluate if antibiotic discontinuation may revert this effect. Ann. Intensive Care 2017, 7(Suppl 1):7 We have had non conflict of interest in this study. Results We included 44 patients in the phase1 and 58 patients in the phase2. Baseline characteristics of patients were similar in both groups. Compliance with all the measures has been improved between the two period from 0 to 32.75%. The incidence density decreased from 33.74 to 18.05 VAP per 1000 ventilator days between observational and interventional period, but the all-cause mortality was almost equal in the 2 groups (88.6 vs. 86%). Discussion With the implementation of our bundle, observance of the team were improved in the second group, compared to the first and the incidence density decreased from 33.74 to 18.05 VAP per 1000 ventilator days between both period. This result is consistent with the littérature. Sure enough, many studies show the same effect of VAP prevention with a decrease of nearly 50% of the incidence density of VAP, after implementation of a «ventilator -bundle [2]. Conclusion The implementation of a "ventilator bundle, " has significantly reduced the incidence of VAP in our service. In the contrary, our study failed to demonstrate a reduction in mortality.
Introduction With an increasing incidence and high mortality rates, sepsis is a public health issue. There is growing evidence that sepsis induces long lasting alterations of transcriptional programs through epigenetic mechanisms that may lead to protracted inflammation, organ failure, sepsis-induced immune suppression (SIIS), secondary infections and death. We hypothesized that epigenetic changes contribute to the pathophysiology of SIIS. To test this hypothesis, we studied the effects of histone deacetylases (HDAC) inhibition with trichostatin A (TSA) in a double-hit murine model of SIIS and secondary pneumonia. Materials and methods C57BL/6 mice were treated with TSA (2 mg/ kg ip) or saline serum (CTL) 30 min before induction of sepsis by cecal ligation and puncture (CLP). Surviving mice underwent intratracheal instillation of 1.5 × 10 6 CFU of Pseudomonas aeruginosa 8 days after CLP. We evaluated the effect of TSA on survival and cellular responses to the primary and secondary infections. Cellular responses in the blood, spleen and BAL were assessed by flow cytometry after CLP (Days 1, 3 & 8) and after pneumonia (4 & 12 h). We also studied lymphocyte apoptosis and dendritic cells (DC) expression of CD40, CD86, and MHCII. Bacterial clearance was assessed in the BAL and in the blood 4 and 12 h after pneumonia. Continuous variables represented as mean ± SD were compared using Student t test. Kaplan-Meier curves were compared by the log rank test. P < 0.05 indicated statistically significant differences. Results Whereas treatment with TSA did not change survival after CLP, TSA improved survival after tracheal instillation of P. aeruginosa (P = 0.009, Fig. 22). TSA-treated mice had significantly higher absolute DC, T and B-lymphocytes counts with reduced lymphocyte apoptosis after CLP. Four hours after secondary pneumonia, TSA-treated mice had significantly higher DC counts and improved bacterial clearance in the BAL, with reduced systemic dissemination of P. aeruginosa. Conclusion HDAC inhibition with TSA improves survival in our murine model of secondary pneumonia, improves bacterial clearance and attenuate cellular features of SIIS. These results suggest that sepsisinduced epigenetic changes contribute to the advent of SIIS. Comprehensive characterization of epigenetic changes associated with SIIS might allow us to identify new therapeutic targets to reprogram immune cells in sepsis and avoid SIIS. length of ICU stay was 16 ± 14 days. 47 patients acquired 58 NIs (36.2% BSI, 20.5% pneumonia, 10.3% CRI and 10.3% UTI. There was no bacteriological documentation of NI in 22.5% of cases. NIs occured 4 days post burns. The most three isolated pathogens were: Acinetobacter spp. (31%), P. aeruginosa (22.5%) and extended spectrum betalactamase-producing Enterobacteriaceae (17%). The most frequently administered antibiotics were polymyxin/carbapenem/teicoplanin combination (21%), polymyxin/carbapenem combination (15%) and carbapenem/tigecycline combination (15%). In our study, mortality rate was 50%. Conclusion Nosocomial infection occured in 24.7% of cases in burn patients, caused by Acinetobacter spp, P. aeruginosa and Enterobacteriaceae BLSE. So, eradication of infection in burn patients require effective surveillance and infection control in order to reduce mortality rates, length of hospitalization and associated costs.
Introduction Infection of the lower respiratory tract is the most common cause of infection in intensive care unit (ICU) (1). Although the attributable mortality of ventilator associated pneumonia remains debated, the recurrence of these infections is always associated with a significant morbidity (2). Staphylococcus aureus methicillin-sensitive (SAMS) is one of the most frequently germs involved in ICU pneumonia especially in trauma patients. The aim of the study was to establish the risk factors associated with microbiological treatment failure of pneumonia, caused by SAMS. Materials and methods we retrospectively identified 185 patients who developed a first episode of ventilator associated pneumonia caused by SAMS during a 6 years-period (2009-2014). The primary end point was the microbiological treatment failure defined as a second episode of pneumonia caused by SAMS corresponding to either a persistent or a recurrence of the pneumonia (Fig. 23). The primary aim of the study was to identify factors associated with a treatment failure, the secondary objective was to identify factors associated with the occurrence of second episode (i.e. persistent, recurrence, superinfection and/or relapse of pneumonia caused by any bacteria) during or after treatment of the first episode caused by SAMS. Definition of outcomes was based after analysis of current concepts available in the literature. Factors associated with primary and secondary objectives in univariate analysis (p-value < 0.20), or clinically relevant ones, were entered in a multivariate logistic regression. The final selection was performed using the stepwise selection based on the Akaike criterion. Results Fifty-nine patients (33.3%) developed a second episode of pneumonia and among them, 30 (16.6%) were considered as a microbiological failure. In a multivariate analysis, the association of oropharyngeal flora (FOP) with the SAMS (OR, 3.86; 95% CI, 1.33-11.11; p = 0.011) and the need of emergency surgery (OR, 5.26; 95% CI, 1.39-35.2; p = 0.035) were predictive of a microbiological failure. Empirical antibiotic therapy with amoxicillin-clavulanic acid (OR, 2.28; 95% CI, 1.15-4.51; p = 0.017) and performing emergency surgery (OR, 3.60; 95% CI, 1.34-9.65; p = .011) were predictors of a second episode of pneumonia caused by any bacteria. Conclusion In this retrospective, monocentric study, the co presence of orophryngeal flora and the need of emergency surgery were associated with microbiological failure of pneumonia caused by SAMS in ICU.
Introduction Ventilator-associated pneumonia is a major iatrogenic problem since it is a cause of hospital morbidity, mortality and increase of health care costs. It has been studied many times, but data's revision is always necessary. Our study aimed to describe epidemiology of ventilator-associated pneumonia and identify local causative pathogens.

Materials and methods
We carried out a prospective study in an intensive care unit. Were included patients intubated for more than 48 h, from April 2015 to May 2016, and presenting signs of ventilator-associated pneumonia (fever, abundant and purulent secretion, increase of FiO 2 greater than 0.2, signs on chest-X ray) with positive culture of endotracheal aspirate. Were excluded patients with germ colonization. Results A total of 268 patients were ventilated for more than 48 h. Among them thirty-four patients aged of 39 ± 18.5 years presented 54 episodes of ventilator-associated pneumonia (that is 1.58 ± 0.85 episodes per patient). The mean SOFA score was 5.2 ± 1.8. The main reasons of mechanical ventilation were loss of consciousness secondary to poisoning (20%), respiratory distress (20%) and status epilepticus (12%). The mean duration of stay was 62.5 days with extremes at 7 and 180 days. The average time between hospitalization and suspicion of ventilator-associated pneumonia was 5.9 ± 2.7 days. The average value of the Clinical Pulmonary Infection Score at suspicion was 5 ± 1.26. The average time between recurrences was 16.8 days with extremes at 4 and 39 days. The culture of endotracheal aspirate identified two pathogens in 11%. It reveled Acinetobacter baumanii in 35% in which 63% were imipenem resistant, Pseudomonas aeroginosa in 33%, Klebsielle pneumoniae in 22%, Staphylococcus  Fig. 23 See text for description aureus methicillin resistant in 7%. Extended spectrum β-lactamases bacteria were found in 10% and carbapenemases producers in 12%. Empirical antibiotherapy was always association of imipenem and colistin. It was necessary to adapt it to antibiograms in 33/54. Ventilator-associated pneumonia was complicated by septic shock in 28% and acute respiratory distress syndrome in 24%. Patients evolved to healing in 63% of episodes (n = 34), to superinfection in 26% (n = 14) and to death in 9% (n = 5). Pseudomonas aeruginosa was the most frequent germ in superinfection (7/14), Acinétobacter baumanii was the most pathogen associated to death (3/5). Conclusion Ventilator-associated pneumonia is an iatrogenic disease that threatens lives. It's in part avoidable. Preventive measures have to be implemented to reduce its frequency, consequences and costs.
Introduction During mechanical ventilation, mismatch between respiratory muscles activity and the assistance delivered by the ventilator results in dyspnea and asynchrony and is commonly observed in intensive care unit (ICU) patients. Proportional Assisted Ventilation (PAV) is a ventilatory mode that adjusts the level of ventilator assistance to the activity of respiratory muscles estimated by an algorithm. To date, PAV has been mostly studied in patients without severe dyspnea or asynchrony. We hypothesized that, compared to pressure support ventilation (PSV), PAV will prevent severe dyspnea or asynchrony. Patients and methods Were included ICU mechanically ventilated patient exhibiting severe dyspnea or asynchrony with PSV. Three conditions were successively studied: 1) PSV on inclusion (Baseline), 2) PSV after optimisation of ventilator settings in order to minimize dyspnoea and asynchrony (Optimisation), and 3) PAV. Ten-minutes recording were performed with each condition. The intensity of dyspnea was assessed by the Visual Analogic State (VAS, only in patients able to communicate) and by the Intensive Care Respiratory Distress Operating Scale (IC-RDOS) for all the patients. The electrical activity (EMG) of extradiaphragmatic inspiratory muscles was measured. The  Fig. 24 Bayesian NMA with random effect prevalence of asynchrony was quantified by the visual inspection of the airway flow and pressure traces. Results 34 patients were included, 74% male, aged 66 [58-78] years, SAPS2 57 [39-66], mechanically ventilated for 6 [4][5][6][7][8][9] days. The tidal volume (Tv) was higher in the Optimisation and PAV than in the Basal condition (Table 27). The respiratory rate(RR) was lower with PAV than in the other conditions. The dyspnea-VAS was lower with Optimisation and PAV than with the Basal conditions. The IC-RDOS was lower with PAV than with the two other conditions. The asynchrony index was lower with PAV than with the two other conditions. Parasternal EMG activity was lower with PAV and Optimisation (Fig. 25). Conclusion In ICU patients receiving mechanical ventilation with PSV and exhibiting severe dyspnea or asynchrony, the optimisation of ventilator settings with PSV and the PAV mode decrease in the simiar way the severity of dyspnea and the prevalence of patient-ventilator asynchrony.
Introduction In spite of recent research and progress in weaning protocols, extubation failure still occurs in 10-20% of patients and is associated with poor outcomes, with a mortality rate of 25-50%. Many risk factors for planned extubation failure have been suggested, including hypercapnia at end of spontaneous breathing trial (SBT). However, performing arterial blood gases at the end of SBT is not routinely recommended whereas EtCO2may be routinely monitored during a low pressure support SBT. The aim of this prospective observational study was to determine the clinical usefulness of EtCO2 to predict extubation failure. Patients and methods We recorded clinical data and EtCO2 during a successful 1 h low level pressure support SBT (at the beginning, after 5 min and at the end of the trial). Patients ventilated through tracheostomy and unplanned extubations were excluded. Extubation failure was defined as death or the need for reintubation within 72 h (1) after extubation; this delay was prolonged to 7 days (2) in case of noninvasive ventilation after extubation, which was systematic in older patients or those with cardiorespiratory disease, as per our weaning protocol. Multivariable logistic regression analysis was performed to identify independent variables associated with extubation failure. Results One hundred and fifteen ventilated patients were enrolled in our study from July 2015 to June 2016. The median age of these patients was 63 [52-75] years, their median Simplified Acute Physiology Score (SAPS) II was 48 [38-61] points and 42.5% (n = 49) were female. Seventeen (15%) patients had chronic obstructive pulmonary disease. Reintubation rate was 15% (n = 18). EtCO2 at other time points as well as its changes during the SBT were also similar between groups. The three variables predicting extubation failure in the multivariable logistic regression model were a past medical history of cirrhosis, acute respiratory distress syndrome before weaning and lower minute ventilation at the end of SBT. Conclusion EtCO2 during a successful SBT seems useless to predict outcome of extubation.
Introduction Airway management in intensive care unit (ICU) patients is challenging [1]. "Airway failure", defined as the inability to breathe without endotracheal tube, differs from "weaning failure", defined as the inability to breathe without an invasive mechanical ventilation. However, most of the studies assessing predictive factors of extubation failure did not separate airway from weaning failure. We aimed to describe incidence of extubation failure in critically ill patients, separating for the first time airway from weaning failure, in a prospective multicenter observational study. Patients and methods A prospective, observational, multicenter study was conducted in 26 French ICUs. All adult patients consecutively extubated in ICU were included. An ethics committee approved the study design (code UF: 9242, register: 2013-A01402-43). The study was registered on clinicaltrials.gov (identifier no.NCT 02450669). Clinical parameters were prospectively assessed before, during and after extubation procedure. Extubation failure was defined as the need to reintubate less than 48 h after extubation. Extubation failure could be due to airway failure, weaning failure or mixed airway and weaning failure. Results From December 2013 to May 2015, 1514 intubation-procedures were studied in 1465 patients from 26 centers. 49 patients (3.2%) were intubated twice. The median number of intubation-procedures included by center was 27 . The flow chart of the study is shown in Fig. 26. Incidence of extubation failure was 10.4% (157 of 1514 intubation-procedures). Incidence of airway failure, weaning failure and mixed failure were respectively 4.6% (70 of 1514), 5.2% (78 of 1514) and 0.6% (9 of 1514).
Conclusion Extubation failure at 48 h occurred in 10.4% of the 1514 extubation procedures recorded, 44% due to airway failure, 50% to weaning failure and 6% to mixed airway and weaning failure. Specific risk factors will be determined using this multicenter database.
Introduction Acute on chronic liver failure (ACLF) have been recently defined by an acute decompensation of a chronic liver disease associated to organ failure and a high mortality rate. Few authors reported on the use of total plasma exchange (TPE) in patients with the current definition of ACLF. The aim of this pilot study was to evaluate the efficiency and safety of TPE in critically ill cirrhotic patients admitted with ACLF in the ICU. Patients and methods A prospective cohort of cirrhotic patients admitted to the ICU between February 2015 and February 2016. TPE was performed using a plasma filter (TPE2000, Hospal ® ) on a CVVHDF machine (Prismaflex ® , Baxter ® ) connected to the patient with a femoral double lumen 13F catheter. The plasma volume exchanged per session was 1.3-1.5 of the total plasma volume. Ratio and type of fluid replacement were 50% with 5% albumin solution followed by 50% with fresh frozen plasma. Clinical and biological parameters, and the following scores MELD, SOFA, CLIF-SOFA, CLIF-OF and Child Pugh were evaluated prior, after TPE session and 7 days distant of treatment.
Results Seven male patients with a mean age of 50.6 ± 7.7 years comprised the study and had a total of 20 TPE sessions. The etiology of cirrhosis was alcoholic (n = 6) or post-HCV (n = 1). The reasons of ACLF were acute alcoholic hepatitis (n = 2), variceal bleeding (n = 2) and sepsis (n = 3). Prior to TPE, the mean scores of SOFA, CLIF-SOFA, CLIF-OF, MELD and Child-Pugh were respectively 9.4, 12, 11.7, 37.6 and C12.5. Mean total bilirubin prior and after TPE sessions was reduced from 392.3 ± 117.8 µmol/l to 244.9 ± 93.2 µmol/l (reduction of 37.5%; p = 3.81E−6); at day 7, mean total bilirubin was still lower at 333 ± 132 µmol/l (p = 0.1). Mean INR prior and after TPE improved from 5.54 ± 3.38 to 2.32 ± 0.80 (reduction of INR of 58.1%, p = 4.76E−5) and at day 7 of treatment at 4 ± 1.64 (reduction of 43%, p = 0.125). Mean GGT levels reduced by 28.5% (p = 0.01). Mean platelet counts (50.8 ± 19.5 G/L) reduced by 10.2% (p = NS). The probability of survival at 10, 28 and 90 days was 42.8, 28.6 and 14.3%. One patient was transplanted and still alive. Tolerance during sessions was good similar to CVVHDF. Two side effects related to the femoral catheter were observed (bacteremia and hemorrhagic shock post catheter ablation). Conclusion This preliminary study of TPE in ACLF showed a marked reduction of liver enzymes and improvement in coagulation parameters with a relative good safety. A specific caution should be undertaken regarding catheter related complications. TPE worth to be Fig. 26 Flow chart of the FREE-REA study Introduction Extubation is a key moment for the patient on his way to recovery. Extubation failure concerns 10-20% of ICU patients and is closely linked to nosocomial pneumonia. The practice concerning enteral feeding interruption at time of extubation has not been investigated. Fasting before extubation may prevent aspiration and development of nosocomial pneumonia. Thus, fasting and gastric content suctioning before extubation may be reasonably considered as a mean to reduce this burden. Fasting before extubation may prevent aspiration and development of nosocomial pneumonia. Thus, fasting and gastric content suctioning before extubation may be reasonably considered as a mean to reduce this burden. However, fasting, as recommended before elective general anesthesia is likely to be ineffective in the setting of extubation in the ICU, due to patients' gastroparesis and prolonged gastric stasis. Beyond the potentially unnecessary burden in terms of paramedical workload, fasting may have some side effects such as caloric deficit, hypoglycemia, or delayed extubation. Given the current lack of objective data concerning the clinical practice of feeding/fasting and gastric tube suctioning before extubation in the ICU, we undertook this descriptive study to assess current practice. Materials and methods We conducted a retrospective, multicenter study in eleven intensive care units in the west of France over a 1 month timespan. All patients extubated were included and data about enteral feeding during the peri-extubation period as well as extubation failure and nosocomial that pneumonia occured within 7 days were recorded. Data observed in the eleven participating centers were completed with a short email survey concerning declarative practice performed among 43 intensive care units. Results During the study period, 162 patients were included. Overall, 25 patients (15%) failed extubation and needed reintubation within the 7 days following planned extubation. Pneumonia was significantly more frequent reintubated patients than the other (36 vs. 4%, p < 0.01). Hundred patients (62%) received enteral feeding at the time of extubation. Compared to patients who did not receive enteral feeding, those patients had a higher disease severity (SAPSII score 50, [41; 63] vs. 45 [35; 52], p < 0.01; longer duration of mechanical ventilation 7 [4; 13] vs. 1.5 [1; 3] days, p < 0.01). Accordingly, those patients had a higher rate of extubation failure (21 vs. 7%, p = 0.01) and pneumonia (13 vs. 3%, p = 0.05). Among the 100 patients receiving enteral feeding, fasting was implemented before extubation for 64 patients (64%). Similarly, the incidence of pneumonia was not different between groups (n = 9 (14%) vs. n = 4 (11%), p = 0.76). After extubation, the fasting patients experienced a longer delay until feeding resumption as compared to non-fasting patients (21 h [6; 42] vs. 8 [5; 22]), but this difference did not reach statistical significance. Overall gastric content suctioning before extubation was not commonly performed; before extubation: 30% of the fasting patients and 26% of the non fasting patients. Among the 11 participating centers, while some centers imposed a fasting period before extubation to all their patients, some did it infrequently. However, no center never imposed fasting, illustrating between and within center heterogeneity. This heterogeneity was confirmed on the larger scale declarative email survey (88% response rate amont 43 units) which showed that only 44% of the units had a written standardized operational procedure for extubation. Survey respondents reported to practice fasting before extubation "Always", "Frequently" and "Never or Rarely" in respectively 71, 21 and 8% of cases. Conclusion Both practices, fasting as well as pursued nutrition until extubation are commonly performed in ICUs, with little standardization of practice. Safety seems equivalent, as no clinically significant difference in terms of reintubation rate and pneumonia were observed. Thus, the equipoise condition appears met to undertake a trial evaluating feeding strategies in the peri-extubation period.
Introduction Noninvasive ventilation (NIV) has become a cornerstone for the supportive therapy of acute respiratory failure (ARF). Survival benefits in chronic obstructive pulmonary disease (COPD) and cardiac patients have been demonstrated. Although ARF and COPD patients are at risk of malnutrition that adversely affects patient outcomes, few data are available regarding the management of nutritional support in non-invasively ventilated patients. We sought to describe nutritional management in patients receiving NIV as the first line therapy for ARF. Secondary objectives were to assess the impact of early nutrition use on the need for invasive mechanical ventilation, occurrence of ICUacquired pneumonia, length of stay, and death. Patients and methods We conducted an observational study from the multicenter French database fed by 20 French ICUs. Our institutional review board approved this study. Adult medical patients admitted to the ICU and receiving NIV for more than 2 days were included. Exclusion criteria were patients admitted after surgery, readmitted in ICU, patients with neuromuscular disease and treatment-limitation decisions on admission. Four groups of patients were defined according to nutrition received during the first 2 days of NIV: (1) No nutrition; (2) Enteral nutrition: patients who received enteral nutrition with or without parenteral nutrition; (3) Parenteral nutrition only (3) Oral nutrition only. The impact of nutrition on day-28 mortality was assessed through the use of a Cox model adjusted on clinically relevant covariates. The impact of nutrition on other secondary end-point i.e. ICU-acquired pneumonia occurrence, need for invasive mechanical ventilation were assessed using a Fine & Gray models. Patients were censored after 28 days of follow-up. Choice among collinear variables was performed considering clinical relevance, rate of missing variables and reproducibility of definitions. Results were given as hazard ratio (HR) for Cox models and subdistribution hazard ratios (sHR) and 95% confidence intervals (CI). The impact on duration of stay was estimated by a multivariate Poisson regression. P values less than 0.05 were considered as significant. Statistical analysis was performed using SAS 9.4 (Cary, NC). Results During the study period, 16,734 patients were included in the database and 1075 met inclusion criteria. Among them, 622 received no nutrition; 28 received enteral nutrition, 74 received parenteral nutrition only, and 351 received oral nutrition only. Overall, 86 patients developed ICU-acquired pneumonia (8%), 158 required invasive mechanical ventilation (14.7%) and 161 died before day-28 (15%). Median length of stay was 6 days [4; 9]. After adjustment for confounders, type of nutrition support was associated with an increase day-28 mortality (P = 0.02). Compared to oral nutrition, enteral nutrition was associated with an increase day-28 mortality [sHR 2.91, 95% CI 1.44-5.89; P = 0.003] whereas parenteral nutrition and no nutrition did not influence this outcome. The type of nutrition was not associated with the occurrence of ICU-acquired pneumonia (P = 0.18). However, patients who received enteral nutrition experienced more frequently ICU-acquired pneumonia [sHR = 3.00, 95% CI 1.08-8.37; P = 0.036] as compared to oral nutrition patients. Ventilator free days within the 28 days were negatively associated with the type of nutrition (P < 0.0001). Compared to oral nutrition, parenteral and enteral nutrition were negatively associated with ventilator free days within the 28 days [RR per day = 1.48, 95% CI 1.22-1.78; P < 0.0001 and RR per day = 1.77, 95% CI 1.37-2.30; P < 0.0001]. Delta PaCO 2 measured between the first 2 days was not associated with any type of nutrition. Conclusion More than half the patients receiving NIV were fasting within the first two NIV days. Oral nutrition was prescribed for onethird of them and was well tolerated. Lack of feeding or underfeeding had no impact on mortality and ventilator free days within the 28 days. However, enteral nutrition was associated with an increased occurrence of ICU-aquired pneumonia and a higher mortality rate. was high, caloric debt during temporary ECLS was low in comparison with previous results [1]. Overnutrition was frequent in the NEC group and would justify implementation of nutrition protocol. Incidence of GI intolerance remains frequent and could justify systematic used of motility agents with introduction of EN. Conclusion Enteral nutrition in patients treated with temporary extracorporeal life support is feasible and may be improve with systematic motility agents and implementation of nutritional protocol.
Introduction Cardiac surgery with cardiopulmonary bypass (CPB) is associated with a generalized inflammatory response with concomitant immune paresis which predisposes to the development of postoperative infections and sepsis (1). Lymphocytes are essential agents of innate and adaptive immune responses during infections or inflammation processes. Lymphopenia has been associated with immune dysfunction during septic shock, and it has been shown that low absolute lymphocyte count was predictive of postoperative sepsis (2). Furthermore, impaired lymphocyte function probably occurs after CPB. Thus, we investigated mechanisms involved in postoperative lymphopenia and impaired lymphocyte function after CPB. The aims of this study were: 1) To describe a potential relationship between lymphopenia and occurrence of postoperative infections. 2) To demonstrate that CPB induces lymphocytes apoptosis. 3) To demonstrate that CPB impaired lymphocyte function (ability to proliferate). 4) To demonstrate that IL-10, PD-L1 (programmed cell death 1 ligand 1) and Indoleamine 2,3-dioxygenase (IDO) could be interesting targets to restore lymphocyte ability to proliferate after CPB. Patients and methods Blood cell counts with differentials obtained within the first postoperative week were analyzed in 828 patients undergoing cardiac surgery in 2015. Postoperative lymphopenia was defined as a lymphocyte count <1.3 × 109 cells L −1 . Postoperative infections were defined following CDC criteria. Study procedures: The following analysis were performed before (T0) and 24 h after (T1) cardiac surgery with CPB: Lymphocyte apoptosis; T-cell proliferation ability following polyclonal stimulation; HLA-DR and PD-L1 expression on monocytes; plasma IDO activity and IL-10 levels; and the ability of lymphocytes to undergo a clonal proliferation when stimulated using specific inhibitors of IL-10 and IDO. The study was approved by our local ethics committee. Patients were informed of the observational nature of the study and gave their consent.

Results
1. Early lymphopenia after CPB was associated with the occurrence of postoperative infection: Postoperative infections occured with a median delay of 6 days. Patients who developed postoperative infections had a significantly lower lymphocyte count at Day 4, Day 5 and Day 6 than patients without postoperative infections. 2. CPB induced lymphocyte apoptosis and decreased T-cell proliferation ability. 3. CPB during cardiac surgery decreased mHLA-DR expression. 4. CPB increased IDO activity, PD-L1 expression and IL-10 plasma levels. 5. IL-10 or PD-L1 inhibition of inhibition could restore ability of lymphocytes to proliferate, although IDO inhibitors did not show any effect.

Conclusion
We provided new evidences that CPB induces immunosuppression. We also demonstrated that IL-10 and PD-L1 could be interesting targets to restore ability of lymphocytes to proliferate. As maintaining MV during CPB decreased plasmatic levels of IL-10, our study brings new evidences that ventilator strategies could be of interest to decrease postoperative infections.
respectively 50.7% (n = 3729), 41.1% (n = 3024) and 16.1% (n = 1174) of the included patients. Mortality was of 47.4% in the overall population (n = 3483) and was higher in neutropenic patients (60.3 vs. 43.6% in non-neutropenic patients; P < 0.001). Neutropenia was independently associated with poor outcome when adjusted for underlying malignancy, allogeneic stem cell transplantation and severity as assessed by organ support (OR 1.45; 95% CI 1.27-1.65). Mortality decreased progressively over time in both non-neutropenic (from 54 to 44%; P < 0.0001) and in neutropenic patients (from 72 to 57%; P < 0.0001). When adjusted for confounders, admission during a more recent period was independently associated with favourable outcome and did not change the final model. Conclusion This preliminary analysis suggests a meaningful survival in neutropenic critically ill cancer patients despite an independent association between neutropenia and mortality. Additional analyses are on-going in order to adjust for study weight, heterogeneity across studies, assess the influence of neutropenia duration or G-CSF use, and confirm the influence of neutropenia in a predefined subgroup of patients.
Introduction Candida bloodstream infections (CBI) are frequent and increasing in hospitalized patients, especially in intensive care units.
Considering the results of some experimental in vitro and animal studies, it seems that yeasts belonging to Candida genus are able, so as to survive, to modulate the immune response of the host by guiding T cells polarization to Th2 profile. Th1 and Th17 cytokines are known to be involved in host defense against CBI. However, these data are mainly experimental or collected after candidemia. The aim of this study is to precise kinetic of cytokines network during human CBI.

Patients and methods
This was an ancillary study of an institutional project dedicated to pathophysiology of candidiasis. We have included 32 patients with candidemia and 54 controls (27 matched hospitalized controls and 27 healthy subjects). The sera of cases were gathered before (almost 5 days before), during and after the isolation of yeasts from blood culture, defined as day 0 (D0). Quantitative analysis of 28 cytokines by Luminex ® technology and of (1,3)-β-d-glucans by Fungitell ® test were performed on 132 samples. The amplitude of Th profile response was expressed by summing the amount of the most relevant cytokines for Th1, Th2 and Th17 profiles, in pg/ mL. For each patient, the highest level of response was considered as 100%. Results are expressed for the population by means of the results. We then performed univariate analysis (Fischer exact test for qualitative variables, Mann-Whitney and Wilcoxon test for quantitative variables, Spearman for correlation; GraphPad Prism V6 software) and a multidimensional analysis by principal component analysis (PCA; IgorPro software). Results Patients with candidemia exhibited an increase in proinflammatory cytokines (IFNγ, TNFα and IL-12), in comparison with the anti-inflammatory cytokines (IL-4 and IL-10) before D0 (p = 0.034) in univariate analysis. The ratio between mean values reverses at D2 and D3 (p = 0.02) and the increase of Th2 response level from D0 to D4 is correlated to the decrease of Th1 response (r = −0.885; p = 0.033) in univariate analysis and PCA. A pro-inflammatory response (Th1) is associated with a reduced mortality (RR = 0.58 [0.34; 0.98]) and with a lower β-D-glucans levels (p < 0.0001). Discussion We describe here a dynamic cytokine profiles in response to candidemia. Pro-inflammatory response predominates before D0 and reverses after. This is contradictory to the postulate that an antiinflammatory background could predispose to invasive candidiasis in ICU patients and exhibiting a "Post-infectious immune suppression conditions". But the relative deficiency in Th1 response compared to simultaneous anti-inflammatory cytokines secretion observed after CBI is in accordance with experimental data, suggesting the modulation of the immune response by Candida. The link between cytokinic profile and mortality can also raise the hypothesis of an influence by genetic factors on the regulation and direction of the immune response and so, the existence of a high-risk population.
Conclusion These data suggest a relation between Candida and the orientation of the immune response towards a pattern deleterious for the infected host. This could allow to determine the most relevant cytokines varying during CBI. They could be used as biomarkers to identify the patients who could benefit from an early treatment in a preemptive targeted therapeutic strategy. These data will be paralleled to genetic background and to circulating Candida derived molecules to precise the relative part of the host and the pathogen in this complex interaction. Introduction Lung ultrasound is widely used in intensive care, ermergency and pneumology medicine, for assessing acute respiratory pathologies. It is noninvasive, radiation free and rapidly available at the patient's bedside and provides an excellent accuracy. So, lung ultrasound may be an interesting tool for the physiotherapist as it allows to assess with more accuracy the patient improving the chest physiotherapy indication and monitoring (1).
As far as we are aware, no study has evaluated the impact of lung ultrasound on clinical-decision making by physiotherapists in the use of chest physiotherapy. This case report highlights the lung ultrasound interest in chest physiotherapy in patient with lung consolidation. Patients and methods This was a case report written following the recommendations of the CARE guideline (2). The case was a 68-years-old female patient, non intubated, hospitalized in a respiratory ICU. She was hypoxemic (PaO 2 = 59 mmHg and SaO 2 = 89%), with dyspnoea at rest and an increasing radiological opacity at the right lung base. Hypoxemia was the indication for physiotherapist referral. At the clinical examination, the physiotherapist's findings were: decreased mobility, dullness and abolished vesicular sound at the base of right hemithorax. This clinical examination and chest x-rays analysis allowed the physiotherapist to propose several clinical hypotheses: pleural effusion, obstructive atelectasis or pneumonia. The chest physiotherapy treatment differs according to the type of lung deficiencies. For example, the physiotherapist must to refer the patient to the medical staff in case of pleural effusion or may implement hyperinflation technique in case of obstructive atelectasis. Determining the nature of lung deficiencies is essential to provide the more suitable therapeutic strategy. So, the physiotherapist decided to perform a lung ultrasound examination to retain the more likely hypothesis. Results Ultrasound examination performed by the physiotherapist highlighted the presence of a lung consolidation at the infero-lateral and posterior parts of the right lung with a pneumonia pattern: presence of tissue-like sign, shred sign, dynamic air bronchogram and fluid bronchogram. The medical staff implemented antibiotic treatment. The ultrasound findings guided the physiotherapist to choose chest physiotherapy technique improving the alveolar recruitment: nearly prone position (left side down) and continuous positive airway pressure during 45 min. The patient response to the treatment was monitored by ultrasound and showed a decrease of the lung consolidation size and apparition of B lines, meaning a gain of lung aeration. These findings were associated with SpO 2 improvement but without decrease of dyspnoea. Discussion Lung ultrasound allowed the physiotherapist to precise the nature of the radiological lung opacity. As it is more accurate than clinical examination or chest x-ray, this suggests a more suitable choice of chest physiotherapy techniques than conventional clinical decision-making process. Ultrasound findings suggested a positive response to the chest physiotherapy treatment. The apparition of re-aeration signs (B lines, decreased consolidation size) showed a short-term efficacy of the chest physiotherapy treatment. This allowed the physiotherapist to continue the treatment during 1 week and obtain a substantial clinical improvement. Conclusion The use of lung ultrasound in the clinical decision-making process may help the physiotherapist to choose with more accuracy the therapeutic strategy. Moreover, it allows to monitor the treatment in real-time and assess the patient's response. The use of this tool may allow the physiotherapist to determine the optimal indications for chest physiotherapy and thus avoid unnecessary or inappropriate treatments.
Introduction Critical illness together with immobilization have deleterious effects on patients outcome, especially in the presence of sepsis. Increased muscle catabolism and membrane inexcitability reduce muscular mass and impair function within the first days after sepsis onset (1). Early mobilization could potentially limit muscle wasting and functional impairment in this population. The purpose of this study was to test whether exercise during the early phase of sepsis is safe and beneficial and to which extent it can limit skeletal muscle protein catabolism and preserve function. Patients and methods Adult patients admitted with the diagnosis of severe sepsis were included and randomly allocated to two groups; 1) Control group (Ctrl-G): manual passive/active manual mobilization twice a day or 2) Experimental group (Exp-G): additional two times 30 min of passive/active cycling exercise. Both groups benefited from a reduced sedation, adjusted nutritional intake and bed to chair transfer as soon as possible. Skeletal muscle biopsy and electrophysiological testing were realized at day-1 and day-7. Muscle histology, biochemical and molecular analyses of anabolic/catabolic and inflammatory signalling pathways were performed. A group of four healthy subjects was used to obtain non pathological values. Hemodynamic parameters and patients perception were collected during each session. Results Twenty-one patients were included, however 3 died before the second muscle biopsy. Ten patients in Ctrl-G and nine in Exp-G were finally analysed. Muscle fibre cross sectional area (µm 2 ) was significantly preserved by exercise (relative changes were Ctrl-G: −45 ± 41% vs Exp-G:12 ± 19%, p = 0.001). Markers of catabolic systems were highly increased during sepsis compared to healthy subjects and reduced in both groups 7 days after admission. However the reduction in mRNA (relative change) tended to be more important in Exp-G: MURF-1 (Ctrl-G: −31 ± 67% vs Exp-G: −63 ± 45%, p = 0.15), MAFbx (Ctrl-G: −7 ± 138% vs Exp-G: −56 ± 37%, p = 0.23), LC3b (Ctrl-G: 5 ± 47% vs Exp-G: −21 ± 18%, p = 0.18) and Bnip3 (Ctrl-G: 27 ± 198% vs Exp-G: −59 ± 23%, p = 0.02). Anabolic and inflammatory markers were not affected by exercise. Electrophysiological testing, including direct muscular stimulation, was abnormal on Day-1 in 10 of 13 evaluated patients. Since only a limited number of patients could be reassessed a second time, comparison between groups was not possible.
In general, all activities were well tolerated by patients with no adverse events.

Introduction
The pulmonary auscultation is used by respiratory therapist (RT) to evaluate the efficiency of a treatment. Listen to the noises coming from the primary bronchi (PB) is important because it is the place where secretions can be accumulated. Therefore, it is crucial to know exactly where to place the stethoscope's chestpiece on the chest. Few studies have analyzed the chest area where the PB were located. Our hypothesis is that PB are localized on a line that joins axillary fossa (Bi-Axillary line: BAL). The aim of our study is to evaluate the probability to find the primary bronchi by analysis of chest radiography.
Patients and methods A retrospective study was performed by analysis of chest X-Ray using the software: TM reception ® , which allows precise measures to the tenth of millimeter. All the X-Rays were made on confined to bed patients hospitalized within intensive care unit, internal medicine and abdominal surgery rooms.
The following measures (in mm) were made between: The exclusion criteria were: BMI < 18.5 kg/m 2 and BMI > 30 kg/m 2 , scoliosis, minor patient, lack of visibility of one of the axillary fossa, lack of visibility of PB, clavicular asymmetry, kyphosis, lack of symmetry in the shot, atelectasis and pneumothorax. Statistics: Normality test: KS. Mean values are expressed with their SD and 95% CI.

Results
Discussion In this study, we performed analysis of chest x-Rays of bedridden patients and we demonstrated that it is possible to localize easily, on either side of the BS, the right and left PB at ± 25 mm distance (LP) above a line joining axillary fossa. This study constitutes a new tool for the RT who, by using stethoscope with a chestpiece of 10 cm 2 surface area, will be able to listen to noise coming from PB. Conclusion The data presented herein (Fig. 31) show that right and left PB are located at a mean distance of 25 (±5) mm and 27 (±6) mm above the BAL, on both sides of the BS. The BAL represents thus an easy and precise mode to detect right and left PB by bedridden.
Finally, the distance between the hyoid bone and the SC is about 12 cm. As the PB are located after the bifurcation, this information constitutes another useful way for the localization the right and left PB by bedridden patient.
Introduction Critically ill patients frequently develop muscle weakness, which is associated with prolonged intensive care unit and hospital stay (1). This randomized controlled trial (Clinical Trials NCT02047617) was designed to investigate whether a daily training session using a tilt table, started early in stable critically ill patients with an expected prolonged ICU stay, could improve strength at ICU and hospital discharge compared to a standard physiotherapy program.

Patients and methods
The study protocol was approved by an ethics committee and informed consent was obtained from all patients. Patients admitted in adult ICU of Marie Lannelongue hospital, France, who were mechanically ventilated for at least 3 days were included. Exclusion criteria were cerebral or spinal injury, pelvic or lower limb fracture. Patients were assessed each day for temporary contraindications for mobilization out of bed (RASS score <−2 or >1; hemodynamic instability; a continuous intravenous dose of epinephrine/  . No significant difference was observed in terms of MRC score or in terms of pts with or without weakness (MRC > 48) at ICU or hospital discharge. However, the number of pts with weakness was significantly higher in the group before Tilt mobilization, suggesting a more rapid improvement in the tilt group. The ICU and hospital lengths of stay were not different between groups. Discussion The prevalence of muscle weakness in our population is high before mobilization (90.6%, 95% CI 85.2-95.6), is still 48.6% at ICU discharge but represents only ~ 1% at hospital discharge. This low hospital discharge prevalence is probably related to the early and intense physiotherapy in both groups, which may explain our inability to demonstrate superiority of the addition of tilt table positioning, although a faster recovery is suggested. Conclusion Training sessions using a tilt table, in addition to early and intense physiotherapy did not improve muscle strength evaluated using MRC score in surgical ICU patients with muscle weakness.