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

Annals of Intensive Care, Jan 2017

Chtara Kamilia, Kais Regaieg, Najeh Baccouch, Hedi Chelly, Mabrouk Bahloul, Mounir Bouaziz, Ali Jendoubi, Ahmed Abbes, Houda Belhaouane, Oussama Nasri, et al.

A PDF file should load here. If you do not see its contents the file may be temporarily unavailable at the journal website or you do not have a PDF plug-in installed and enabled in your browser.

Alternatively, you can download the file locally and open with any standalone PDF reader:

https://link.springer.com/content/pdf/10.1186%2Fs13613-016-0224-7.pdf

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

Ann. Intensive Care Proceedings of Réanimation 2017, the French Intensive Care Society International Congress 0 Réanimation polyvalente, CHU Habib Bourguiba, Sfax, Tunisia Annals of Intensive Care 2017 , 7(Suppl 1):P144 1 Réanimation polyvalente, Faculté de médecine de Sfax , Sfax , Tunisia 2 P144 Post traumatic cerebral thrombophlebitis: prospective study about 15 cases Chtara Kamilia 3 Inserm cic 1435/urgences/samu, Centre Hospitalier Universitaire de 4 P148 Staphylococcal community‐acquired urinary tract infection in the emergency department: a sign for acute infective endocarditis? Thomas Lafon 5 Urgences/samu, Centre Hospitalier Universitaire de Limoges , Limoges , France 6 Service de réanimation polyvalente, Centre Hospitalier Universitaire de Limoges , Limoges , France 7 Limoges , Limoges , France 8 Inserm cic1435/ service de réanimation polyvalente, Centre Hospitalier Universitaire de Limoges, Limoges, France Annals of Intensive Care 2017 , 7(Suppl 1):P148 9 Bactériologiev‐irologie‐hygiène/umr‐s 1092, Centre Hospitalier Universitaire de Limoges , Limoges , France 10 Service de maladies infectieuses, Centre Hospitalier Universitaire de Limoges , Limoges , France 11 Réanimation polyvalente adulte, Centre Hospitalier Intercommunal André Grégoire, Montreuil, France Annals of Intensive Care 2017 , 7(Suppl 1):P157 12 Unité de microbiologie clinique et dosages des anti‐infectieux, Groupe Hospi‐ talier Paris SaintJ‐oseph , Paris , France 13 Réanimation, Groupe Hospitalier Paris SaintJ‐oseph , Paris , France 14 P157 Clinical impact of extended‐spectrum 15 lactamase producing Enterobacteriaceae colonization on pneumonia in ICU Caroline Schimpf 16 Department of biology, Centre d'Assistance Médicale Urgente, Tunis, Tunisia Annals of Intensive Care 2017 , 7(Suppl 1):P160 17 Department of intensive care and toxicology, Centre d'Assistance Médi‐ cale Urgente , Tunis , Tunisia 18 P160 Carbapenemase‐producing Enterobacteriaceae: experience of a Tunisian intensive care unit Amira Ben Jazia 19 Réanima‐ tion médicale, CHU Farhat Hached. Research Laboratory N° LR14ES05. Faculty of Medicine, Sousse, Tunisia com Annals of Intensive Care 2017 , 7(Suppl 1):P172 20 Réanimation médicale , CHU Farhat Hached, Sousse , Tunisia 21 P172 Effects of early use of diuretics in patients at risk of acute renal failure and oliguria Narjess Ben Aicha 22 Service de réanimation, centre d'assistance médicale‐urgente, Tunis, Tunisia Annals of Intensive Care 2017 , 7(Suppl 1):P176 23 Department of intensive care and toxicology, Centre d'Assistance Médi‐ cale Urgente , Tunis , Tunisia 24 P176 Early versus late‐onset ventilator‐associated pneumonia: causative pathogens and resistance profiles Hend Ben Lakhal 25 Neuro‐ muscular reference centrer, Cliniques universitaires Saint‐Luc, Université catholique de Louvain , Brussels , Belgium 26 Institute of neuroscience, Université catholique de Louvain , Louvain‐la‐Neuve , Belgium 27 Intensive care unit, Cliniques universitaires Saint‐Luc, Université catholique de Louvain , Bruxelles , Belgium 28 O81 Physical therapy during the early course of sepsis is safe and preserves skeletal muscle mass Cheryl Hickmann 29 Department of physical medicine and rehabilitation, Cliniques universitaires Saint‐Luc, Université catholique de Louvain , Brussels , Belgium Annals of Intensive Care 2017 , 7(Suppl 1):O81 - Introduction Head injury is a rare but possible etiology of cerebral thrombophlebitis. The diagnosis should be considered especially in front of open head injuries extended to venous sinuses. The MR angiography is the gold standard for early diagnosis. Patients and methods This is a descriptive prospective study of all trauma patients hospitalized in the intensive care unit of the University Hospital Habib Bourguiba Sfax over a period of 6  years between January 2010 and June 2016 and in whom the diagnosis of cerebral venous thrombophlebitis has been confirmed by angiography CT or MR angiography. Results During the period study, 15 patients were included. The median age of patients was 29 [17–49]  years. All patients were male, victims of poly trauma following an accident of traffic. In admission, SAPSII was 31 [24–52] and SOFA was 4 [2–8]. We have noted the presence of a serious head injury in 15 patients, extended open skull fractures of the venous sinus in 9 patients. A related chest trauma was present in 12 patients and abdominal trauma in 4 patients, trauma of the pelvis and/or members were present in 7 patients. All patients underwent mechanical ventilation. The diagnosis of cerebral venous thrombosis was confirmed by cerebral angiography CT in 9 patients and cerebral MR angiography in 6 patients. 7 patients have presented secondary pulmonary embolism. All patients did not show a contraindication against anticoagulation at diagnosis of thrombophlebitis. The thrombophilia (antithrombin III, protein C and S, homocysteine, and antiphospholipid, gene mutation factors II and V) as well as for antineutrophil cytoplasmic antibodies were negative in all patients. The outcome was favorable in 13 patients. Two patients were died due to a state of refractory septic shock. Discussion Post traumatic cerebral thrombophlebitis is a rare thrombotic vascular disease. It must be mentioned especially with presence of extensive skull fractures in open sinuses. Venous MR angiography is the gold standard. The treatment is based on anticoagulation curative dose. Its prescription can be complicated in these cases associated with traumatic intracranial hemorrhage. Conclusion Head injury is a rare but possible etiology of cerebral thrombophlebitis. Other prospective studies are needed to better understand the path physiology and the prognosis of these thromboses. P145 Pain measurement in mechanically ventilated patients with traumatic brain injury: behavioral pain tools versus analgesia/nociception index—preliminary results Ali Jendoubi1, Ahmed Abbes,1, Houda Belhaouane,1, Oussama Nasri,1, Layla Jenzri,1, Salma Ghedira2, Mohamed Houissa2 1Anesthesia and Intensive Care, Charles Nicolle Teaching Hospital, Tunis, Tunisia; 2 Intensive care, Charles Nicolle Hospital, Tunis, Tunisia Correspondence: Ali Jendoubi ‑ Annals of Intensive Care 2017, 7(Suppl 1):P145 Introduction Pain is highly prevalent in critically ill trauma patients especially those with a traumatic brain injury (TBI). Behavioral pain tools such as the Behavioral pain scale (BPS), and critical care pain observation tool (CPOT) are recommended for sedated non-communicative patients. The analgesia nociception index (ANI) assesses the relative parasympathetic tone as a surrogate for antinociception/nociception balance in sedated patients. The primary aim is to evaluate the effectiveness of ANI in detecting pain in TBI patients. The secondary aim was to evaluate the impact of Norepinephrine use on ANI effectiveness, and to determine the correlation between ANI and BPS. Patients and methods We performed a prospective observational study in 21 deeply sedated TBI patients. Exclusion criteria were nonsinus cardiac rhythm; presence of pacemaker; atropine or isoprenaline treatment; neuromuscular blocking agents and major cognitive impairment. HR, blood pressure and ANI were continuously recorded using the Physiodoloris® device at rest (T1), during (T2) and after the end (T3) of the painful stimulus (tracheal suctioning). Results In total, 100 observations were scored. Patients’ characteristics were resumed in Fig. 1. ANI was significantly lower at T2 (Med (min–max) 54.5 (22–100)) compared with T1 (90.5 (50–100), p < 0.0001) and T3 (82 (36–100), p  <  0.0001). Similar results were found in the subgroups of patients with (65 measurements) or without (35) Norepinephrine. During procedure, A negative linear relationship was observed between ANI and BPS (r2 = −0.469, p < 0.001). At the threshold of 50, the sensitivity and specificity of ANI to detect patients with BPS ≥ 5 were 73 and 62%, respectively with a negative predictive value of 86%. Discussion Conclusion ANI is effective in detecting pain in deeply sedated critically ill TBI patients, including those patients treated with Norepinephrine. © 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. Fig. 1 Baseline demographic and clinical characteristics. Values are expressed as mean ± standard deviation (SD); n (%) or median [interquartile range]. EDH extradural haemorrhage SDH subdural haemorrhage, SAH subarachnoid hemorrhage P146 The prognosis of cervical spine trauma in elderly subjects in surgical intensive care Kamal Belkadi1, Ma Bouhouri2, Youness Harti3, Afak Nsiri2, Khalid Khaleq4, Driss Hamoudi2, Rachid Harrar2 1Anesthesie reanimation, chu ibn rochd, Casablanca, Morocco; 2Reanima‑ tion des urgences chirurgicale, chu ibn rochd, Casablanca, Morocco; 3Anesthésie réanimation, CHU Ibn Rochd Casa, Casablanca, Morocco; 4Service d’accueil des urgences, Chu Ibn Rochd, Casablanca, Morocco Correspondence: Kamal Belkadi ‑ Annals of Intensive Care 2017, 7(Suppl 1):P146 Introduction The aim of our study was to assess the prognostic factors of cervical spine trauma in elderly subjects admitted in the surgical intensive care unit. Patients and methods We conducted a retrospective, and singlecenter study over 16 years (January 2000–January 2016) in Ibn Rochd hospital, we included all patients aged over 65  years with isolated cervical spine trauma, operated and non-operated, admitted in surgical intensive care, the death risk factors were searched by uni and multivariate analysis. Results 198 patients were collected, the average age was 69.4  ±  3.9  years, with a male predominance 70.2%, the main causes were road accidents (50.5%) and fall (34.3%), 68.2% had a complete form (Frankel A), 75.7% were operated. The death rate in our study was 55%. The prognostic factors in univariate analysis were: hypertension, heart disease, fall injury, surgical delay >10 h; independent factors of death in multivariate analysis: heart disease and fall injury. Conclusion The cervical spine trauma in elderly patients hospitalized in intensive care unit is poor prognosis. P147 Interest of the urine antigen testing for Legionella pneumophila in the management of severe acute pneumonia: practice survey and analysis of performance in intensive care unit Camille Thieffry1, Frédéric Wallet2, Erika Parmentier‑Decrucq 1, Raphaël Favory1, Daniel Mathieu1, Julien Poissy1 1Pôle de réanimation, hôpital salengro, C.H.R.U. ‑ Lille, Avenue Oscar Lam‑ bret, Lille, France, Lille, France; 2 Centre de biologie pathologie génétique, Centre Hospitalier Régional Universitaire de Lille, Lille, France Correspondence: Julien Poissy ‑ Annals of Intensive Care 2017, 7(Suppl 1):P147 Introduction Legionnaire’s disease is a rare but severe acute pneumonia with a difficult definitive diagnosis. Legionella’s urine antigen testing is a quick, sensitive and specific test, widely prescribed in the case of an acute and severe community-acquired pneumonia. However, its overall contribution to the diagnosis of atypical pneumonia remains unknown in daily practice. The aims of this study were to evaluate the usefulness of this test in a “real life” utilization and to identify potential clinical markers of legionnaire’s disease in order to optimize its prescription. Legionnaire’s disease is a rare but severe acute pneumonia with a difficult definitive diagnosis. Legionella’s urine antigen testing is a quick, sensitive and specific test, widely prescribed in the case of an acute and severe community-acquired pneumonia. However, its overall contribution to the diagnosis of atypical pneumonia remains unknown in daily practice. The aims of this study were to evaluate the usefulness of this test in a “real life” utilization and to identify potential clinical markers of legionnaire’s disease in order to optimize its prescription. Patients and methods We conducted a retrospective, monocentric and observational study. All the prescriptions of the urine antigen testing where monitored in the intensive care department and the emergency room admitting severe patients, in our University Teaching Hospital, from January 1st 2013 to December 31st 2015. Qualitative variables were compared by a Fischer’s exact test, and quantitative variables by a Mann–Whitney test. All tests were bilateral, and p ≤ 0.05 was considered as significant. ROC curves were determined for the variables of interest. Results During the period of the survey, 1142 urinary tests were performed in 1002 patients. Three tests were positive for 0.26% of patients. Only 569 patients suffered from an actual pneumonia. Other patients suffered mostly from acute bronchitis (137 cases), and exacerbation of chronic obstructive pulmonary disease (123 cases). The characteristics of the 3 patients suffering from legionnaire’s disease differed significantly compared to the other kinds of acute pneumonia for: the need for invasive mechanical ventilation (100 vs 34.96% of patients; p  =  0.043), PaO2/FiO2 ratio (107.5 vs 274.5; p  =  0.0107), duration of mechanical ventilation (27.5 vs 2.5  days; p  =  0.0062), natremia (129.5 vs 138  mmol/l; p  =  0.0125), Creatine Phospho-Kinase level (16,125 vs 106UI; p  =  0.0225) and Serum Glutamat Oxalacetat Transaminase (251 vs 35UI; p = 0.0157). We determined ROC curves for these last biological variables. Natremia: better threshold  =  131, Se/Sp  =  100/84.2%, AUC  =  0.92. SGOT: better threshold  =  191, Se/Sp  =  100/89.6%, AUC = 0.9. CPK: better threshold 195, Se/Sp = 100/66.5, AUC = 0.88. Discussion Urine antigen testing for Legionnaire’s disease appears to be over-prescribed in many cases with a very poor level of positivity in our cohort. However, it could be limited to patients exhibiting a pneumonia, and among them to patients with the classical biological perturbations described in this disease, which have excellent diagnostic performance. This targeted strategy would present an important benefit in term of costs. Conclusion Our results show that a better selection in the patients who could benefit of this test is mandatory. Simples markers in routine testings could help the clinician to adapt his prescription, optimizing this test’s efficiency. Introduction Urinary tract infection is a frequent cause of admission at the Emergency Department (ED). Most prevalent bacteria are usually gram-negative bacilli and Staphylococcus aureus (Sa) is rarely evidenced (2.5%) except in hospital-acquired infections due to urinary catheter [1]. Bacteriuria can be observed in Sa infective endocarditis (IE) because of the metastatic properties of Sa. We hypothesized that presence of Sa in the urine could be related to Staphylococcal bacteremia associated with unsuspected IE and not only the expression of a “usual” urinary tract infection. Patients and methods This is a descriptive single-center study conducted over a 10 year-period in the Teaching Hospital of Limoges. All patients admitted to the ED with Sa (both MSSA and MRSA) isolated from their urine cultures were retrospectively analyzed. Data were collected from the database of the microbiology department and the patient medical charts. We secondarily searched if a Sa IE had been documented in patients with Sa isolated from their blood cultures in order to establish a link between IE and presence of Sa in the urine. We used modified Dukes criteria as diagnostic criteria of IE [2]. Results Between 2005 and 2015, 420,000 patients were admitted in the ED. Out of the 204 records analyzed, 174 patients whose urine culture grew Sa were excluded because they had a urinary catheter (n  =  75) or sterile blood cultures (n  =  99). Finally, 30 patients were studied (17 men; median age: 73 years; diabetes: n = 7; mitral valvular disease: n  =  2, aortic valvular disease: n  =  2) (Table  1). Reasons for admission were markedly heterogeneous and fever accounted for 14 cases. Echocardiography was performed in 25 patients with a median delay of 6.5  days (range: 0–23  days) and IE was confirmed in 21 of them. Only three cases of IE have been diagnosed in the ED because of a fever and valvular murmur (n  =  2) but no patient was admitted to the ED for IE suspicion. Other initially suspected diagnoses were prostatitis, pneumonia or appendicitis and not related to any type Table 1 Patients characteristics Congenital heart valve defect Skin and eye petechiae of infection in 26% of the cases. The origin of Sa bacteremia was cutaneous in 67% of the cases. During the hospital stay, 70% of the patients presented secondary sites of Sa infection in addition to the urinary tract (arthritis, splenic abcess, cerebral hematoma). Among 21 patients presenting with an IE, 8 died within 7 days, and total hospital mortality reached 50%. Conclusion This case series suggests that IE should be ruled out when Sa bacteriuria is evidenced, irrespective of the clinical presentation. This could question the reality of isolated community-acquired urinary tract infections due to Sa. P149 The infectious tricuspid endocarditis in ICU: clinical features, management and outcome Amira Jamoussi1, Amira Ben Jazia1, Takoua Marhbène1, Dhouha Lakh‑ dhar1, Jalila Ben Khelil1, Mohamed Besbes1 1Medical icu, Hospital Abderrahmen Mami De Pneumo‑Phtisiologie, Ariana, Tunisia Correspondence: Amira Jamoussi ‑ Annals of Intensive Care 2017, 7(Suppl 1):P149 Introduction Since the outbreak scourge of intravenous drug addiction in Tunisia, we are witnessing the emergence of cases of infectious tricuspid endocarditis (ITE). This pathology should be studied because it requires specific medical and surgical management. The aim of the study was to describe the clinical features, management and outcome of ITE in intensive care unit. Patients and methods This was a retrospective study from January 2009 to December 2014. We enrolled patients who were hospitalized in intensive care unit and had ITE. We recorded baseline characteristics, management and outcome. Results During the study period, we collected 10 cases of ITE making an incidence rate of 3 cases for 1000 patient admissions. They were divided into 8 men and 2 women. The median age was of 37.5 years. The main reasons of ICU admission was acute respiratory failure (80%), among them 5 required mechanical ventilation. Hemodynamic failure was present in 3 cases. The median SAPS II was of 33 [19–90]. The median APACHE II was of 17 [7–53]. Different contributing factors were identified: intravenous drug abuse (6 patients), a central venous catheter (1 patient) and a pacemaker (1 patient). All patients underwent transesophageal echocardiography showing one or several vegetations on native tricuspid valve. No associated left endocarditis was found. Blood cultures were positive in 8 cases of which 5 contained 2 different micro-organisms. The identified micro-organisms were: Meticillin Resistant Staphylococcus aureus (n  =  6), Meticillin Sensitive Staphylococcus aureus (n  =  3), coagulase-negative staphylococcus (n  =  2), Enterobacter cloacae (n  =  1), and candida famata (n  =  1). Occurring complications were hospital-acquired infections (n  =  5), septic pulmonary embolism (n  =  4), withdrawal syndrome (5 cases), acute renal failure (n = 2) and atrioventricular block (n = 2). Medical treatment consisted of a double antibiotic treatment. Surgical treatment was required in 7 patients: tricuspid valve replacement by bioprosthesis (6 cases) and valvuloplasty (1 case). The average length of stay was of 31.3 days [2–56]. ITE had recurred on bioprotheses in two patients after intravenous drug resumption; they underwent surgery again and one of them died. In hospital mortality was of 30%. The outcome was favorable in 7 patients. Conclusion The ITE in ICU is a severe disease with frequent complications and in hospital mortality reaches 30%. The most frequent incriminated micro-organism is Meticillin Resistant Staphylococcus aureus. It often requires medical and surgical treatment. Intravenous drug addiction remains the most common cause and worsens the prognosis by the risk of recurrence. P150 Microbiological mapping of community‑acquired intra‑abdominal infections (IAI) and indicator of local antibiotherapy appropriateness with French national guidelines Julien Goutay1, Caroline Blazejewski2, Isabelle Joly‑Durand 3, Isabelle Pirlet4, Marie Pierre Weillaert5, Sebastien Beague2 1Interne en anesthésie réanimation, C.H. Régional Universitaire de Lille (CHRU de Lille), Lille, France; 2Réanimation polyvalente, Hospital Center De Dunkerque, Dunkerque, France; 3Equipe opérationnelle d’hygiene, Hospital Center De Dunkerque, Dunkerque, France; 4Service de chirurgie digestive, Hospital Center De Dunkerque, Dunkerque, France; 5Labora‑ toire, Hospital Center De Dunkerque, Dunkerque, France Correspondence: Julien Goutay ‑ Annals of Intensive Care 2017, 7(Suppl 1):P150 Introduction French guidelines for community-acquired IAI underline the importance of establishing antibiotherapy protocols based on regular analysis of microbiological data; and of systematic site infection cultures to determine microbial sensitivity to antibiotics. Our study describes microbial population involved in our communityacquired IAI and defines an annual follow-up indicator of probabilistic antibiotherapy inadequacy to microbial sensitivity. Materials and methods We conducted a retrospective, monocentric, observational study from January the 1st 2014 to December the 31st 2015. All community-acquired IAI in adults were included. Exclusion criteria were: cirrhosis and peritoneal dialysis. Initial probabilistic antibiotherapy and total antibiotherapy duration were left to the discretion of the physician in charge. Results of intra-abdominal cultures (IAC) were analyzed. Three microbial groups were defined: (A) A-group: wild-type bacteria strains treated with adequat antibiotherapy; (B) B-group: antibiotic resistant bacteria treated with adequat antibiotherapy; (C) C-group: multi-drug resistant bacteria treated with inadequat antibiotherapy. A Chi square analysis was performed on SPSS software (IBM). Results 98 community-acquired IAI were included: 54 (55%) had positive IAC with 133 bacteria; 34 (35%) didn’t have intra-abdominal swab; 10 (10%) had sterile cultures. Predominant strains were Gramnegative bacteria (76/133 (57%)). More represented bacteria were Escherichia Coli (50/133 (38%)) and Bacteroïdes fragilis (16/133 (12%)). 127/133 (95%) bacteria belong to A- and B-groups. B-group importance increased significantly between 2015 and 2016. C-group characteristics were comparable over the 2 years. Initial antibiotherapy was inadequate with French guidelines in 38/98 (39%) cases and with microbial antibiotic susceptibility in 6/133 (4.5%) cases (C-group). Average antibiotherapy duration was 11.3  days. Antibiotic treatment duration was too long according to French guidelines for 65/98 (67%) patients. Discussion Our microbial population in community-acquired IAI is similar to national studies with a lowest resistance rate (C-group under 10%). Probabilistic antibiotherapy proposed by French guidelines is appropriated to our microbial ecology. Antibiotherapy duration is unconformed with guidelines in 67% cases. Peroperative swabs are frequently missing (35%). Non-compliance with French guidelines highlights the importance to formalize our local procedure. This formalization at any stage (surgical, medical and biological cares) seems essential to improve our standard of care. C-group rate could be used as a real-time feedback to adapt our protocol continuously. Conclusion The goal of our study is to improve local standard of care by offering a formalization of community-acquired IAI management procedure. C-group rate seems to be a good follow-up indicator of probabilistic antibiotherapy inadequacy to microbial sensitivity, allowing an optimization of our protocol in real-time. Reference 1. Montravers et al. Recommandations Formalisées d’Experts «Prise en charge des infections intra‑abdominales», Octobre 2015. P151 The resumption of peritonitis in surgical intensive care unit Ma Bouhouri1, Kamal Belkadi2, Soufi Aziz1, Khalid Khaleq3, Afak Nsiri1, Driss Hamoudi1, Rachid Harrar1 1Reanimation des urgences chirurgicale, chu ibn rochd, Casablanca, Morocco; 2Anesthesie reanimation, chu ibn rochd, Casablanca, Morocco; 3Service d’accueil des urgences, Chu Ibn Rochd, Casablanca, Morocco Correspondence: Kamal Belkadi ‑ Annals of Intensive Care 2017, 7(Suppl 1):P151 Introduction The resumption of peritonitis is a serious complication of abdominal and pelvic surgery. It’s a medical and surgical emergency, the prognosis depends on the speed, the quality of the care, and the underlying terrain and etiology. Patients and methods We conducted a descriptive analytic retrospective study over a period of 5 years (January 2011–June 2016) 60 cases of peritonitis hospitalized in surgical intensive care unit. Results The average age of our patients was 44.36  years with a sex ratio of 1.5 (36H/24F). The most frequent risk factors were: factors relating to the ground, and factors related to the initial peritonitis. Clinical signs were dominated by fever (75%), abdominal pain (52%). The period of the average recovery was 8.2 days. The decision of the surgical revision was based on a clinical, biological and radiological criteria. 40 patients in our series, 67% of cases were taken on clinical and biological criteria while 15 patients 25% were taken on radiological criteria. In 8% of the remaining cases, the potential severity of the clinical and biological state in association with an inconclusive ultrasound, led to reoperation. The therapeutic treatment was based on a perioperative resuscitation, treatment of organ failure, empirical antibiotic therapy and by midline laparotomy surgery. Bacteriological samples performed intraoperatively allowed to have the following bacteriological profile: predominance of BGN (79%) dominated by E. coli (28%) followed by Klebsiella pneumoniae (21%), Acinetobacter and Enterococcus baumanii (12%). The multimicrobien character was found in 55%. The E. coli–Klebsiella pneumoniae association was the most frequent (37%). The anastomotic dehiscence was the direct cause of the most common surgical revision found intraoperative (62%). The average hospital stay was 8 days. The mortality rate was 61%. The main prognostic factors in our study emerged in the univariate analysis were: kidney failure, the number of organ failure, a TP <50% the needs of ventilation and the use of catecholamines. Discussion Mortality is variable depending on the studies, between 25 and 60%. Conclusion The diagnosis often difficult. Only effective and early therapeutic management reduces mortality remains high in recent years despite the various advances in the field of surgery and reanimation. P152 Prognostic factors in intra abdominal sepsis: a prospective study Reda Hafiane1, Khalid Khaleq1, Khalid Hattabi2, Mohamed Aziz Bouhouri1, Afak Nsiri1, Driss Hammoudi1, Abdelaziz Fadil2, Rachid Al Harrar1 1Service de réanimation des urgences chirurgicales, CHU IBN ROCHD de Casablanca, casablanca, Morocco; 2Service des urgences viscérales, CHU Ibn Rochd de casablanca, Casablanca, Morocco Correspondence: Reda Hafiane ‑ Annals of Intensive Care 2017, 7(Suppl 1):P152 Table 2 Main prognostic factors in intra abdominal sepsis Pesence of clammy skin Hemodynamic instability 26 Use of vasoactive drugs P153 Community acquired intra abdominal sepsis: concerning 302 cases Reda Hafiane1, Khalid Khaleq1, Khalid Hattabi2, Mohamed Aziz Bouhouri1, Afak Nsiri1, Driss Hammoudi1, Khalid Zerouali3, Abdelaziz Fadil2, Rachid Al Harrar1 1Service de réanimation des urgences chirurgicales, CHU IBN ROCHD de Casablanca, casablanca, Morocco; 2 Service des urgences viscérales, CHU Ibn Rochd de casablanca, Casablanca, Morocco; 3Service de microbiol‑ ogie, CHU Ibn Rochd de casablanca, Casablanca, Morocco Correspondence: Reda Hafiane ‑ Annals of Intensive Care 2017, 7(Suppl 1):P153 Introduction Intra abdominal sepsis is a dangerous condition causing a high mortality rate even with surgery and post operative care improvement. The aim of the study is to assess the bacteriological and epidemiological profile of this population. Patients and methods It’s a prospective observational study performed during 6 months (02/2016–08/2016) in visceral emergency operating rooms. Inclusion criteria: adults admitted with intra-abdominal infectious disease diagnosed with biological and radiological means. Exclusion criteria: post operative peritonitis and deceased patients before their admittance. Studied parameters: demographic data (gender, age…), the time management, co morbidities, number of organ failure, intraoperative incidents and postoperative evolution. Results During this period, we admitted 302 patients, the mean age: 41.14 ± 17 years. Male predominance was noticed in our population: 69.9%. Main emergencies are reported in the attached Table  3. Concerning the bacteriological profile: we had 121 positive samplings. Enterobacterias were the most frequent strain. E coli was predominant: 43%, Enterococcus faecalis: 25%, Streptococcus (viridians and acidominimus): 14%, we had 1 case of Acinetobacter baumanii resistant to imipenem. 6 cases of yeasts were found (Candida albicans). Antibiotic use was: Ampicilline: 44%, Ceftriaxone 41%, metronidazole: 76% and tazocilline: 4%. Discussion In our context, community acquired intra abdominal sepsis leads to a high death rate. We noticed relevant parameters: a late time management, high gravity scores, some surgical procedures not directed by supervisors. Therefore, we have to establish therapeutic protocols tailored to each disease in order to improve patients’ management and help to reduce the mortality rate. Conclusion Early diagnosis and care for intra abdominal sepsis represent a major way to prevent complications. Bacteriological proof is necessary to adjust post operative antibiotherapy. Introduction Intra- abdominal sepsis represents a life threatening condition. Its manifestations are non specific and can quickly lead to multi organ failure if not treated correctly. Patient’s assessment is essential in order to adjust the therapy. The aim of our study is to highlight the prognostic factors in this situation. Patients and methods It’s a prospective observational study performed during 6 months (02/2016–08/2016) in visceral emergency operating rooms. Inclusion criteria: adults admitted with intra-abdominal infectious disease diagnosed with biological and radiological means. Studied parameters: demographic data, co morbidities, number of organ failure, type of anesthesia, intraoperative incidents and evolution. Results were analyzed using SPSS software, prognostic factors were extracted with univariate then multivariate analysis. Significant results were noted. Results During this period, we admitted 302 patients, the mean age: 41.14 ± 17 years. Male predominance was noticed in our population: 69.9%. The mortality rate was: 13.2%. The main prognostic factors were reported in the attached Table 2. Discussion Many significant prognostic factors were identified: Age, existence of hemodynamic failure with renal involvement, long operative time and the use of vasoactive drugs. A high hemoglobin level at the admission was a protective factor. The presence of respiratory distress, the sex and the presence of yeasts were not significant factors in our study. Conclusion Intra abdominal sepsis is causing quickly a multi organ dysfunction syndrome leading to death. Therefore, our priority is to stop this sepsis with the help of the surgeon and the efficient use of antibiotics. Table 3 Main emergencies with epidemiological profile Peptic perforation peritonitis intestinal perforation peritonitis P154 Conformity of antibiotic prescribing in emergency room Fatma Kaaniche Medhioub1, Rania Allela2, Najla Ben Algia3, Samar Cherif4 1Faculté de médecine de Sfax, Sfax, Tunisia; 2Hopital régional mahres, Fac‑ ulté de médecine de Sfax, Sfax, Tunisia; 3Intensive care, hopital régional Gafsa, Sfax, Tunisia; 4Intensive care, hopital régional mahres, Sfax, Tunisia Correspondence: Fatma Kaaniche Medhioub ‑ Annals of Intensive Care 2017, 7(Suppl 1):P154 Introduction The development of bacterial resistance is a major public health problem due to unreasonable use of antibiotics. The introduction of appropriate antibiotic therapy has a positive impact on patient survival and a significant economic impact. The objective of this study is to evaluate the compliance of antibiotics prescribed in emergencies. Patients and methods Prospective study conducted on 1  year (01/01/2015–31/12/2015). We have included patients admitted to the emergency with hyperthermia (>38°), hypothermia (<36°) or two criteria of systemic inflammatory response syndrome. Were collected in the emergency department: history, presence of prior antibiotic therapy, demographic and clinical characteristics at admission, prescription of antibiotics in emergencies and its modalities and the discharge diagnosis. During hospitalization were collected: the introduction, modification or discontinuation of the antibiotic, the reasons for this change and certainty diagnosis. The compliance analysis of antibiotic therapy was performed by an expert group (two infectiologists, a bacteriologist and an emergency doctor) with regard to the current recommendations. Two groups were defined and compared: group of patients receiving complies antibiotic therapy and group with antibiotics considered improper. The criteria associated with non-compliance were sought. Results Four hundred and twenty-two patients were enrolled. The final diagnosis retained an infectious etiology in 356 patients (84%). The mean age was 62.4 ± 18 years. Blood cultures were taken in 370 cases (87.7%). Infectious sites were most often lung (52%) and urine (32%). Severe sepsis was diagnosed in 14 patients (3.3%). A complies prescription was found in 335 patients (79.4%). Antibiotic therapy was started in 302 patients (71.5%) at the emergency and classified complies with 234 (55.4%). Among the 68 patients (16.1%) with an illegal antibiotic, it was continued in 24 (35.3%) during hospitalization. Among 120 patients (28.4%) did not receive antibiotics, this attitude was classified complies in 115 patients (95.8%). Non-compliance was related to the presence of antibiotics in the last 3  months and the presence of renal failure. Conclusion Particular attention should be paid to the antibiotic prescription in patients subject to prior exposure to these. Dose adjustments should be respected in cases of renal failure. Regular evaluation of the antibiotic prescription in the emergency is necessary. P155 Pulmonary resections’ bacterial cartography Mohamed Taoufik Slaoui1, Souhail Boubia2, Y. Hafiani1, A. Khaoudi1, R. Cherkab1, W. Elallam1, C. Elkettani1, L. Barrou.1, M. Ridaii2 1Anesthesia service surgical resuscitation, chu ibn rochd, Casablanca, Morocco; 2 Thoracic surgery, chu ibn rochd, Casablanca, Morocco Correspondence: Mohamed Taoufik Slaoui ‑ Annals of Intensive Care 2017, 7(Suppl 1):P155 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 Infection is a major cause of morbidity and mortality in burned. The bacterial ecology varies among centers. Despite the progress in the management of severe burned, mortality remains very high. The aim of this study is to establish the pathogenic profile of AB in this population. Materials and methods Single-center retrospective study of 7  months, including any serious burned hospitalized for more than 48 h in intensive care, and who benefited from bacteriological samples during his stay. Infectivity was retained on a range of clinical and biological arguments (CDC criteria) changed). They excluded all burned died for noninfectious causes, and patients with isolated settlement. Results Sixty-two (72) patients were infected by the AB during our study period. The sex ratio (M/F) was 1.7 and the mean age was 39 ± 23 years. Nosocomial pneumonia was present in 61.11% of cases. Urinary tract infection was present in 18.05% of cases. Bacteremia was present in 12.5% of cases. Skin infection was present 8.33% of cases. The resistance profile was marked by 100% of cases of resistance to third-generation cephalosporins (C3G), 88% of cases of resistance to fluoro-quinolones (FQ), 74% of cases of resistance to imipenem and 64, 28% of cases of resistance to tigecycline. Conclusion The incidence of infection with Acinetobacter baumannii in our unit remains high compared to that of intensive care units. Colonization and infection by the AB are significantly associated with increased length of stay, and mortality, and given the gravity of hospitalized patients, failure to comply with hygiene and abusive use of antibiotic prophylaxis. Introduction ESBL are enzymes mostly found in Enterobacteriaceae and confer resistance to all beta lactams antibiotics except cefoxitin and carbapenems. Recently, a significant increase in the rate of ESBLrelated infections in ICU makes difficult the choice of empiric antibiotic therapy, especially in patients colonized by extended-spectrum β-lactamase producing Enterobacteriaceae (ESBLe) [1]. Notably, very few data are currently available regarding the role of ESBLe colonization on further pneumonia involving the same bacteria [2]. The aim of our study was to describe the incidence of ESBLe infections among ESBLe-colonized ICU patients. Patients and methods This study was conducted retrospectively from January 1st 2011 to May 1st 2016, in our intensive care department. All admitted ESBLe-colonized patients who develop an infection during their ICU stay have been included in the study. The only exclusion criterion was an antibiotic treatment for an ESBLe infection at ICU admission. Results During the period of the study, 386 stays were associated with an ESBLe colonization in 384 patients. 148 infections were diagnosed in patients colonized by ESBLe, among which 78 pneumonias. In 18 cases (23%) the ESBLe was involved in the pulmonary infection (PN-ESBLe+) and was the only responsible bacterium in 66% of cases. The ESBLe was the same in screening and pneumonia in 15 cases (83%). The PN-ESBLe+ was associated with septic shock in 9 (50%) cases, acute respiratory distress syndrome in 2 (11%) cases and neurologic failure in 7 (39%) cases. Episodes were ventilator-associated pneumonia in 56% (10 cases) of PN-ESBLe+ and 52% (31 cases) of PN-ESLBe-. The most common pathogens involved were Escherichia coli, Klebsiella pneumoniae, and Enterobacter cloacae in both groups. Comparing groups (PN-ESBLe+ or PN-ESLBe−), only the notion of prior antibiotic therapy within 30 days (OR 3.9 [1.07–18.3]; p = 0.03) and colonization by ESBL Klebsiella pneumoniae (OR 4.04 [1.02–16.1]; p  =  0.04) were more frequent in PN-ESBLe+. At least one empiric antibiotic was effective on the ESBLe in 83% of cases. In vitro antibiotic susceptibility tests demonstrate 100% efficiency of the association of piperacillin/ tazobactam and amikacin on ESBLe involved in pneumonia. Mortality at day 28 was 24% for PN-ESBLe+ and 44% for PN-ESBLe−. Hospital mortality was 53 and 38% respectively (p = NS). Among 70 extra-pulmonary infections, ESLBe take part in 30 (43%) cases. The involvement of ESBLe was significantly lower in pneumonia than in other infections (p = 0.01). Discussion Due to the single center character of our study, results cannot be extrapolated to the whole ICU population. Nevertheless, the observed incidence of colonizing ESBLe in our study is close enough from others studies. This point consolidates reflection about ICU pneumonia empiric treatment. Conclusion The involvement of colonizing-ESBLe in ICU pneumonia is rare in our population and significantly lower than in other infections. Identified risk factors for PN-ESBLe  +  are a prior antibiotic therapy within 30  days and colonization with K. pneumoniae. Alternative associations to carbapenem remain efficient in all cases of pneumonia in our ICU and should probably be kept in mind. Introduction Immediate adequate treatment of ICU-acquired Gram negative bacilli (GNB) bloodstream infections (BSI) improves patients’ prognosis. Risk factors of resistance of GNB-BSIs should be better assessed. Materials and methods Data from a large French national ICU network were explored during a 10-year period (2005–2014). Patients with a GNB-BSI were included and were divided into two groups according to the resistance (R) profile (BSI due to a R isolate or not). The following three groups were considered: (1) all GNB-BSI including Pseudomonas spp., Acinetobacter spp., Stenotrophomonas spp. and Enterobacteriacae (Eb) for which the following R were considered: ticarcillin (Pseudomonas spp., Acinetobacter spp., Stenotrophomonas spp.); ceftazidime (cefta) (P. aeruginosa (PA), Acinetobacter spp., Stenotrophomonas spp.), third generation cephalosporin (3GC) (Eb) and imipenem (all GNB, during the period 2011–2014 only), (2) PA cefta R from 2005 to 2014 and (3) Eb species resistant to 3GC from 2005 to 2014. Univariable hierarchical logistic models with two levels (random center and region effects) were used to select variables associated with resistance using a p value threshold of 0.2. Selected variables were further introduced in multivariable analyses using a hierarchical model with two random effects. Results From 265,035 patients admitted in an annual median of 158 French ICUs, 9553 experienced an ICU-acquired (>48  h.) BSI, 5062 (53%) BSI due to GNB, including 1764 (35%) BSI due to R isolates. PA was identified in 1167 (23%) BSIs (480 (41%) R) and Eb in 3298 (65%) BSIs (1226 (34%) R). The median annual incidences of R GNB BSIs/10,000 ICU patients were: 68 for all R GNB BSI, 41 for Eb 3GC-R BSIs and 9.2 for PA cefta-R BSI. There was a significant increase of annual incidence for all GNB R and Eb 3GC-R BSI. Independent factors associated with all R GNB BSI were: 1) ICU variables: percentage of patients with an immunosuppression other than neutropenia (7.9–14%: OR 1.23; 95% CI, [1.04–1.46]; >14%: 1.31 [1.09–1.57]); percentage of resistant GNB the previous year (55–66%: 1.87 [1.59–2.2]; >66%: 2.93 [2.43–3.53]) and 2) patient-variables: antimicrobial therapy at ICU admission (1.79 [1.55–2.08]); presence of an invasive device (CVC or intubation) (1.99 [1.25–3.16]) before infection; and one protective factor: trauma at ICU admission (0.76 [0.65–0.89]). The year effect was significant both for all R GNB and 3GC-R Eb but not for R PA. This effect was more pronounced for 3GC-R Eb, with an increase in the risk of R from 2005 to 2014 (Fig. 2). The duration from ICU admission to infection was the main risk factor of R for all BGN and sub-groups (Eb and PA): the probability of having a BSI due to a R strain increased with the time in ICU before infection (Fig.  3). ICUbased random effect remains significant indicating major impact of local epidemiology. Fig. 2 Evolution of the risk to have a BSI due to a resistant strain according to the year of ICU admission Fig. 3 Days from ICU admission to infection Limitation The absence of information about antibiotic consumption may partly explain the remaining significant center random effect in the final models. Conclusion The duration from ICU admission to BSI was a main risk factor for a resistant isolate in GNB BSI. Resistance rates increased over time, especially for 3GC-R Eb and were highly dependent of local previous epidemiology. P159 Sepsis at ICU admission due to extended‑spectrum β‑lactamase producing enterobacteriaceae among colonized patients: prevalence, risk factors and prognosis Keyvan Razazi1, Jérémy Rosman1, Nicolas de Prost1, Guillaume Carteaux1, Chloe Jansen2, Jean Winoc Decousser3, Christian Brun‑Buisson 1, Armand Mekontso Dessap1 1Réanimation Médicale, Hôpital Henri Mondor, Créteil, France; 2Cepi, Hos‑ pital Henri Mondor, Créteil, France; 3Microbiologie, Hôpital Henri Mondor, Créteil, France Correspondence: Jérémy Rosman ‑ Annals of Intensive Care 2017, 7(Suppl 1):P159 Introduction Prevalence of Extended-spectrum beta-lactamaseproducing Enterobacteriaceae (ESBL-PE) carriers dramatically increases all over the world with a spread to the community. The increasing prevalence of ESBL-PE carriage at Intensive Care Unit (ICU) admission raises important questions on empiric therapy strategies in patients presenting with infection, which may include the use of a carbapenem as first-line therapy. Data on ESBL-PE sepsis at ICU admission among colonized patients are lacking. Patients and methods We prospectively assessed between 2009 and 2015 the prevalence, risk factors and prognosis of ESBL-PE sepsis among ESBL-PE carriers at ICU admission. The following data were collected: demographic characteristics, which included sex, age, simplified acute physiology score (SAPS II), location before ICU admission, antibiotic exposure, surgery during the previous year, presence of underlying disease, Charlson comorbidity index, presence of indwelling devices and outcomes. Results A total of 597 patients had ESBL-PE carriage detected at admission, corresponding to 9.5% of admitted patients. Among these patients, 325 patients had sepsis at ICU admission. Fifty patients (15.4%) had ESBL-PE related sepsis at ICU admission. ESBL-PE infection included 23 (46%) urinary tract infections, 14 (28%) pulmonary infections, 9 (18%) abdominal infections and 4 (8%) other infections. All but two ESBL-PE pneumonia cases were hospital-acquired (86%) while community-acquired ESBL-PE urinary tract infection was not uncommon (12/36 = 33%). By multivariable analysis, prior urinary tract disease [OR 3.0 (1.1–8.0)], hospital-acquired sepsis at admission [OR 2.9 (1.4–5.7)], treatment with fluoroquinolone within the past 3  months [OR 2.8 (1.2–6.4)] past ESBL-PE infection [OR 2.8 (1.2–6.5)] were independent predictive factors for ESBL-PE sepsis at admission, whereas a pulmonary source of sepsis [OR 0.30 (0.15–0.61)] was protective. The final model showed a good calibration (chi2 = 3.7, p = 0.45) and discrimination (area under the curve = 0.85). Patients with ESBL-PE related sepsis had more often septic shock and bacteraemia at admission. ESBL-PE related sepsis was also more often associated with more frequent inadequate empirical therapy (68 vs 87%, p  <  0.001). However, mortality did not differ significantly between patients with ESBL-PE infection and others (20 vs 22%). Conclusion At ICU admission, ESBL-PE related sepsis was relatively infrequent among colonized patients. Our predictive factors for ESBL-PE may help choosing empiric therapy for sepsis among ESBL-PE carriers at ICU admission. The study did not show a significant association between ESBL-PE infection at admission and mortality. 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. Fig. 4 CPEc increasing from 2011 to 2015 This work aims to focus on the impact of CPEc increasing in our intensive care unit. Patients and methods A retrospective and descriptive study conducted in a Tunisian intensive care unit, including all hospitalized patients infected by an Enterobacteriaceae. We have determined annual incidence of CPEc from January 2008 to December 2015. Results One thousand two hundred and three episodes of Enterobacteriaceae infections were eligible in 748 patients (324 male/424 female) aged between 14 and 88 years. The global prevalence of CPEc across study period was 4.48% (54/1203).The prevalence of CPEc in each site was respectively of 23% in hemocultures, 18% in coproculture, 12% in catheters, 3.3% in respiratory tract, and 3% in urinary tract. The overall incidence of (CPEc) increased from 0.69% (1/144) in 2008 to 7.77% (15/193) in 2015 (Fig. 4). Conclusion Our study confirms the rapid spread of CPEc in Tunisian hospital and the urgent need for a well-structured and coordinated national surveillance plan in order to limit their dissemination. P162 Extended spectrum beta lactamase producing enterobacteriacae (ESBL‑PE) infections in ICU Achille Kouatchet1, Rafael Mahieu2, Emmanuel Weiss3, David Schnell4, Jean‑Ralph Zahar 5 1Service de Réanimation médicale et Médecine hyperbare, Centre Hospitalier Universitaire d’Angers, Angers, France; 2Réanimation médicale, Centre Hospitalier Universitaire d’Angers, Angers, France; 3Département d’anesthésie‑réanimation, Hôpital Beaujon, Boulevard du Général Leclerc, Clichy, France, Clichy, France; 4Réanimation médicale, CHU de Strasbourg, Strasbourg, France; 5Laboratoire de bacteriologiev‑irologie‑hygiene, Hôpital Avicenne, Bobigny, France Correspondence: Achille Kouatchet ‑ ackouatchet@chu‑angers.fr Annals of Intensive Care 2017, 7(Suppl 1):P162 Introduction 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. Results During the study period 146 patients with ESBL-PE infection met eligibility criteria with respectively a median age and SAPS II score of 63 (51–74) and 50 (38–70). The median SOFA Score at first day of antibiotic therapy and ICU admission were 7 (4–11) and 7 (5–11) respectively. The most frequent site of infection were respiratory tract (45%), urinary tract (20%) and abdominal (17%). The most frequent isolated species were: Escherichia coli (43%), Klebsiella sp (37%) and Enterobacter sp (18%). Respectively 50, 23 and 27% patients had septic shock, severe sepsis and sepsis according to Bone criteria. Among ESBL-PE, 98.6% were carbapenem and 46.5 were BLBI sensitive. Among the whole population, 47 (48%) patients received a carbapenems as ET. 66 (68%) received a DT with carbapenems and 31 (32%) patients received an alternative DT. The most frequent reasons for maintaining carbapenems as DT were: Antibiotic susceptibility tests (38% of cases), severity level (33% of cases) immunosuppression (8% of cases). The Median length of ICU stay after infection was respectively 12 (6–27) and 11 (7–16) days for carbapenems and alternatives DT (p = 0.1). The D28 mortality was 24% for patients with carbapenems DT and 24% for patients with alternatives DT (p = 0.02). Surprisingly, there were no differences between the 2 groups (carbapenems vs alternatives) in term of severity. Conclusion Alternatives are frequently used for ESBL-PE infections in ICU. In our cohort 31 (32%) patients received antibiotics other than carbapenems regardless of the severity. P163 Extended‑spectrum beta‑lactamase‑producing enterobacteriaceae cross‑transmission in the absence of private room in intensive care unit Margaux Artiguenave1, Paktoris‑Papine Sophie 1, Florence Espinasse2, Faten El Sayed3, Aurélien Dinh4, Cyril Charron1, Guillaume Geri5, Antoine Vieillard‑Baron 1, Xavier Repessé1 1Réanimation médico‑chirurgicale, Assistance Publique ‑ Hôpitaux de Paris, Hôpital Ambroise Paré, Boulogne‑Billancourt, France; 2Equipe opérationnelle d’hygiène hospitalière, Assistance Publique ‑ Hôpitaux de Paris, Hôpital Ambroise Paré, Boulogne‑Billancourt, France; 3Service de microbiologie, Assistance Publique ‑ Hôpitaux de Paris, Hôpital Ambroise Paré, Boulogne‑Billancourt, France; 4Equipe mobile de microbiol‑ ogie, Assistance Publique ‑ Hôpitaux de Paris, Hôpital Ambroise Paré, Boulogne‑Billancourt, France; 5Réanimation Médicale, Hôpital Cochin, Paris, France Correspondence: Xavier Repessé ‑ Annals of Intensive Care 2017, 7(Suppl 1):P163 Introduction Multidrug micro-organisms are responsible for longer hospitalisations and poorer outcomes in intensive care unit (ICU). The transmission of extended-spectrum beta-lactamase producing enterobacteriaceae (ESBL-PE) is prevented by the application of additional contact precautions, mainly relying on isolation in a private room and hand hygiene with waterless alcohol-based solution. Contact isolation cannot be achieved in our 12-bed ICU only composed of two twin bedrooms. We aimed at reporting the ESBL-PE acquisition in this peculiar architectural form of ICU and at studying the impact of twin bedrooms on ESBL-PE cross-transmission. Patients and methods An observational and non-interventional study was prospectively conducted in the 12-bed ICU of a university hospital Ambroise Paré (Boulogne-Billancourt, France). Inclusion criteria were: (1) adult patients and (2) a period of hospitalisation allowing the patient to be nursed by at least two paramedical teams. Characteristics of patients at admission (age, sex, SAPSII) and clinical data during hospital stay (duration of mechanical ventilation, duration of ICU stay, outcome) were prospectively collected. Microbiological data concerning ESBL-PE imported and acquired carriage were monitored by rectal swabs collected at admission and once weekly every Monday for the whole duration of the ICU stay. ESBL imported carriage was defined as a first screening positive for ESBL whereas ESBL acquired carriage as a negative first screening at admission followed by at least one positive rectal swan. Mechanistic of a potential cross-transmission was studied following a three-step process consisting in (1) identifying patients considered as possible ESBL sources (index patients) for transmission, (2) classifying each ESBL strain according to the CTXm 1 and 9 groups and (3) diagnosing potential cross-transmission by gene sequencing of remaining cases of possible transmission. Results From June 2014 to April 2015, 550 patients were admitted in the ICU, among which 470 followed the inclusion criteria. The rate of ESBL colonization at admission was 13.2% (n  =  62), mainly with Escherichia coli. Two hundred and twenty-one non-colonized patients were screened at least twice. The incidence of ESBL acquisition was 4.1% (9 patients on 221), also mainly with Escherichia coli. Mortality did not differ between ESBL carriers and non-carriers. In univariate analysis, ESBL acquisition was associated with the Injury Global Score II (IGSII) and the Sequential Organ Failure Assessment (SOFA) at admission, the need for catecholamine and the ICU length of stay (LOS). In multivariate analysis, ICU LOS and IGSII at admission were the strongest risk factor for ESBL acquisition. The nine ESBLacquired carriers had one to three index patients defined as a patient hospitalized who shared at least 1  day hospitalization. The CTXm grouping of the ESBL strains excluded a cross-transmission for 4 patients. The gene sequencing did it for 3 others and confirmed a cross-transmission in only two patients (0.8%). The cross-transmission emanated from the same source of a CTXm 1 ESBL-producing E. coli. This patient shared 1  day in a different unit with the first acquired carrier and 2  days in the same unit with the other. No case of crosstransmission in the same room was observed. Conclusion The rate of 13.2% of ESBL carriage on admission was comparable to other rates in French ICUs (15%). Despite the absence of contact isolation, the incidence of ESBL acquisition was 4.1% which is actually lower than transmission rates previously published in other ICUs. A cross-transmission concerned two ESBL-acquired carriers only and resulted from the same index patient during short shared hospitalizations of 1 day in a different unit and 2 days in the same unit. Our results question whether the contact isolation in private rooms plays a major role for the prevention of ESBL cross-transmission in ICU, although the external validity of our results could be questionable. 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. Materials and methods A 24  months (January 2014 to December 2015) observational study in a 14 beds ICU in a general hospital. Our unit, is the sole and the referral one of all French Guiana department. Results Over the study period, 670 patients were admitted to ICU and 603 of them (90%) were hospitalized more than 48  h. The mean occupancy rate was 82.5  ±  20.6% and the mean colonization index (with ESBL-PE) was de 37 ± 18.1%. The mean age was 43.4 ± 21.1 years. The sex-Ratio (M/F) was 1.3. The mean IGS II calculated at admission to ICU was 44.6 ± 24.2. The most recorded organ failures at admission to ICU were respiratory and hemodynamic ones (56.7 and 37.2% respectively). At admission to ICU, 44.2% of patients presented active infection and 57.3% received antibiotics. Multidrug resistant (MDR) bacteria carriage was found in 88 patients (13.4%) at ICU admission and was acquired in ICU in 89 other patients (13.4%). The most isolated MDR bacteria at admission were ESBL producing E coli and K. pneumoniae. However, the most isolated MDR bacteria during ICU stay were ESBL producing K. pneumoniae and E cloacae. During the ICU stay, 98 patients (14.6%) had presented 147 episodes of ICU acquired infections (ICU-AI). Over the 177 patients carrying MDR bacteria, 159 (89.8%) carried ESBL-PE and 66 developed ICU-AI. ESBL-PE caused 21.2, 37.5, 20, and 66.7% of 1st, 2nd, 3rd and 4th ICU-AI episodes respectively. Statistical analysis didn’t show any link between ESBL-PE carriage and a first episode of ICU-AI caused by ESBL-PE. Conclusion Our study show a high prevalence of ESBL-PE bacteria carriage at admission in our ICU. ESBL-PE carriage was not associated to higher prevalence of ICU-AI caused by the same microorganism. This finding can help to reduce the inappropriate use of carbapenems in such conditions. P165 On‑line hemofiltration versus conventional hemofiltration in septic shock patients: clinical safety and effectiveness Kaouther Dhifaoui1, Zied Hajjej1, Amira Fatnassi1, Walid Sellami1, Iheb Lab‑ bene1, Mustapha Ferjani1 1Department of critical care medicine and anesthesiology, Military Hospi‑ tal of Tunis, Tunisia, Tunis, Tunisia Correspondence: Zied Hajjej ‑ Annals of Intensive Care 2017, 7(Suppl 1):P165 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). P166 Usefulness of biological testing during renal replacement therapy in ICU patients Fahmi Dachraoui1, Sabrine Nakkaa1, Abdelwaheb M’ghirbi1, Ali Adhieb1, Dhouha Ben Braiek1, Kmar Hraiech1, Ali Ousji1, Islem Ouanes1, Hammouda Zaineb1, Saousen Ben Abdallah1, Lamia Ouanes‑Besbes 1, Fekri Abroug1 1Réanimation polyvalente, CHU Fatouma Bourguiba, Monastir, Tunisia Correspondence: Fahmi Dachraoui ‑ Annals of Intensive Care 2017, 7(Suppl 1):P166 Introduction Clinical and biological monitoring of efficacy and safety of RRT sessions is thought useful and in many ICUs biological testing at mid RRT session and at its end is routinely performed. The aim of the present study is to evaluate the impact of laboratory tests performed during RRT session on clinical decision making and treatment alteration. Patients and methods Retrospective study including all consecutive patients hospitalized in the medical ICU of the University Hospital Monastir, Tunisia between January 2015 and September 2016, requiring intermittent hemodialysis performed in the ICU. For each patient we collected demographic characteristics (age, sex, SAPS III, diagnosis, comorbidities), indication of the RRT, clinical and biological parameters before, during and at the end of RRT session. Based on the patient monitoring records during each RRT session we identified therapeutic interventions started before the end of the RRT session in the light of the results of laboratory tests performed during the session (usually in the middle the session): infusion of glucose, potassium, transfusion, extension of the session. Results During the study period, 370 patients were admitted to the ICU. Of these 24 patients required acute hemodialysis. The median age of these patients were 54 years (IQR = 31), 53% of them were female. The main comorbidities were hypertension, diabetes, chronic renal failure (CRF) with preserved diuresis, respectively in 55, 38 and 36%. ¾ of the patients included were in septic shock and had median SAPS III score of 92 (IQR  =  34). Anuria, pulmonary oedema, hyperkalemia, and acidosis indicated RRT sessions respectively in 47.1, 24.5, 8.8, 9.8 and 9.8%. Results of laboratory tests performed during RRT sessions prompted a specific attitude in the following rates: infusion of glucose in 35.7%, addition of potassium in 22%, and extension of the session duration 15.2%. Conclusion The practice of laboratory tests during the RRT sessions seems useful since it could impact clinical decision making in more than one-third of sessions. P167 Feasibility of regional citrate anticoagulation for membrane‑based therapeutic plasma exchange in ICU Simon Klein1, Mattéo Miquet1, Jean‑Marc Thouret 1, Vincent Peigne1 1Réanimation, Centre Hospitalier Métropole‑Savoie, Chambéry, France Correspondence: Vincent Peigne ‑ Annals of Intensive Care 2017, 7(Suppl 1):P167 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. P168 Modelization of the cost‑effectiveness of anti‑thrombin to reduce the incidence of membrane thrombosis during continuous hemofiltration Vincent Peigne1, Jean‑Louis Daban 2, Mathieu Boutonnet2, Bernard Lenoir3 1Réanimation, Centre hospitalier Métropole Savoie, Chambéry, France; 2Réanimation, Hôpital d’Instruction des Armées Percy, Clamart, France; 3Département d’anesthésie‑réanimation, Hôpital d’Instruction des Armées Percy, Clamart, France Correspondence: Vincent Peigne ‑ Annals of Intensive Care 2017, 7(Suppl 1):P168 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. P169 Vascular access sites for acute renal replacement in intensive care unit Amira Ben Jazia1, Amira Jamoussi2, Takoua Merhbene3, Dhouha Lakh‑ dhar4, Jalila Ben Khelil2, Mohamed Besbes2 1Medical ICU, Hospital Abderrahmen Mami De Pneumo‑Phtisiologie, Ariana, Tunisia; 2Réanimation médicale, Hôpital Abderrahmen Mami, Ariana, Tunisia; 3Réanimation respiratoire, Hôpital Abderrahmen Mami de pneumo‑phtisiologie, Ariana, Tunisia; 4Service de réanimation médicale, Centre d’assistance médicale‑urgente, Tunis, Tunisia Correspondence: Amira Ben Jazia ‑ Annals of Intensive Care 2017, 7(Suppl 1):P169 Introduction Several temporary venous catheterizations are sometimes required for acute renal replacement therapy (RRT) in the intensive care unit (ICU). This study compares catheterizations in the femoral and jugular veins in terms of patient safety. Materials and methods This was a descriptive retrospective review of dialysis sessions (DS) records monitoring performed in patients older than 17 years hospitalized in medical intensive care unit between April 2011 and December 2015. A study of dialysis catheter, was conducted in critically ill adults requiring RRT was performed. Catheter insertion site, catheter age and urea reduction ratio (URR) were analyzed. Results URRs were analyzed from 330 dialysis sessions (n  =  64 patients). The mean rate of URRs was 52.8  ±  12.4. Only 31.4% of dialysis sessions (DS) were efficient with URR ≥ 60. This study analyzed 64 patients who underwent two different sites of catheterization: the femoral and jugular site. The mean age of cathéters was 1.613 days. No significant difference (P  =  0.18) in the efficiency of (DS) was detected between sessions performed through femoral (n = 225; 68%) and jugular (n = 105; 32%) dialysis catheters. Conclusion Femoral and internal jugular acute vascular access sites are both acceptable for RRT therapy in the ICU. The effectiveness of (DS) in the ICU is low (31.4%). An analysis of predictive factors of inefficiency of (DS) is expected to improve our results. P170 Renal replacement therapy protocol with regional citrate anticoagulation: observational study of efficacy with a new post‑filter ionized calcemia target Celine Derreumaux1, Thierry Seguin1, Jean‑Marie Conil 1 1Réanimation polyvalente, Hopital Rangueil, Toulouse, France Correspondence: Celine Derreumaux ‑ Annals of Intensive Care 2017, 7(Suppl 1):P170 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. P171 Impact of the use of an oXiris filter versus an AN69ST filter on the duration of hemofiltration in intensive care Charlotte Kelway1, Valery Blasco1, Cyril Nafati1, Karim Harti1, Laurent Reydellet1, Jacques Albanese1 1RPPF, Hopital de la Timone, Marseille, France Correspondence: Charlotte Kelway ‑ Annals of Intensive Care 2017, 7(Suppl 1):P171 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. They were 63 [24, 87] median (IQR) aged, with diabetes mellitus, 22%; hypertension, 47.8%; cardiac failure, 32% and chronic respiratory failure, 43%. Chronic and obstructive kidney diseases were excluded. Median SAPS II was 37 [23, 75]. 80% were on mechanical ventilation. RIFLE score before diuretics administration was assessed at oliguria onset as (patients without risk, zero; R, 69%; I, 17%; F, 10%; L, 4%; E, zero). Fluid resuscitation after oliguria onset was administered in 77% and vasopressors in 80%. Median (IQR) delay from oliguria onset to diuretic administration was 5 [0.5, 22] h while optimization of prerenal hemodynamic disturbances was already achieved. The median (IQR) MBP before and after therapeutic intervention was respectively, 74 [46, 100] and 95 [69, 110] mmHg. Median (IQR) delay from initiation of therapeutic intervention and MBP improvement was 1.5 [0, 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 Fig. 5 Progression of RIFLE score classes respectively from baseline to after therapeutic intervention 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. P173 Epidemiology and risk factors of Acinetobacter baumannii ventilator associated pneumonia Walid Sellami1, Zied Hajjej1, Soumaya Ben Yedder2, Walid Samoud1, Bous‑ selmi Radhouene1, Bousselmi Mariem3, Iheb Labbene1, Mustapha Ferjani1 1Department of critical care medicine and anesthesiology, Military Hos‑ pital of Tunis, Tunisia, Tunis, Tunisia; 2Department of critical care medicine and anesthesiology, Military hospital of tunis, tunisia, Tunis, Tunisia, Tuni‑ sia; 3Department of critical care medicine and anesthesiology, Military Hospital of Tunis, Tunisia, tunis, Tunisia Correspondence: Walid Sellami ‑ Annals of Intensive Care 2017, 7(Suppl 1):P173 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 18  years and ventilated more than 48  h and developing AB VAP. 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. P174 Incidence and risk factors of central line associated bloodstream infections and its risk factors in a Tunisian medical intensive care unit Nesrine Sma1, Asma Ammar2, Khaoula Meddeb1, Asma Ben Cheikh2, Hend Ben Lakhal1, Jihene Ayachi1, Ahmed Khedher1, Nesrine Fraj1, Mes‑ saouda Khelfa1, Yamina Hamdaoui1, Imed Chouchene1, Nabiha Bouafia2, Mohamed Boussarsar3 1Réanimation médicale, CHU Farhat Hached, Sousse, Tunisia; 2Hospital hygiene unit, Farhat Hached Hospital, Sousse, Tunisia; 3Medical intensive care unit, Farhat Hached Hospital. Research Laboratory N° LR14ES05. Faculty of Medicine, Sousse, Tunisia Correspondence: Mohamed Boussarsar ‑ hamadi.boussarsar@gmail. com Annals of Intensive Care 2017, 7(Suppl 1):P174 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. For all subjects, age, sex, underlying diseases, SAPS II score, ICU length of stay, invasive procedures and their durations (mechanical ventilation, central catheterization, urinary catheterization) were recorded. Risk factors were evaluated by a multivariate logistic regression model. Results Among a total of 237 admissions from September 15th 2015 to September 15th 2016, 163 (68.7) patients were eligible. One hundred twenty-five (76) patients had a central line. A total of 27 episodes of CLABSI were assessed in 23 (18.4) patients. The mean SPASII of patients with CLABSI was 33 ± 15.4. Their mean CHARLSON index was 1.8  ±  1.7, median duration of catheterization was 4 [1.5– 7]  days and 8 (34.8) had more than one catheterization attempt. The rate of CLABSI was 19.2/1000 catheter.days. Gram positive bacteremia was determined in 13% of BSI patients. Of these isolates, 3 were Staphylococci. Gram negative bacteremia was determined in 35% of these isolates, 4 were Acinetobacter baumannii, 3 were Klebsiella pneumonia and 1 was Proteus mirabilis and in 56% of cases BSI was diagnoses clinically. A univariate analysis identified ventilatorassociated pneumonia, sedation, and longer interval between onset of CLABSIs and the duration of catheterization as risk factors of CLABSIs. In multivariate analysis, the independent factors of CLABSI which are the duration of catheterization (OR, 1.06; 95% CI, [1.003–1.139]; p  =  0.042) and catheterization attempt number (OR, 1.99; 95% CI, [1.18–3.37]; p = 0.01). Thirteen (56.5) patients developed septic shock and they all died. Discussion The rate of CLABSI in our ICU (19.2/1000  catheter.days) 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. P178 Ventilator acquired pneumonia: diagnosis treatment and bacterial ecology in a Moroccan intensive care unit Hanane Ezzouine1, Mahmoud Kerrous1, Saad El Haoui1, Soufiane Ahdil1, Abdellatif Benslama1 1Anesthesiology and intensive care department, University Teaching Hospital IBN Rushd‑Casablanca, Casablanca, Morocco Correspondence: Hanane Ezzouine ‑ Annals of Intensive Care 2017, 7(Suppl 1):P178 Introduction The management of ventilator acquired pneumonia is a diagnostic and therapeutic challenge. Antibiotic therapy is a key link. The objective of this work is to study the epidemiological, clinical patients who developed VAP during their stay in the Medical Intensive Care Unit of the CHU Ibn Rushd in Casablanca and are features of the bacterial ecology of VAP during the 2015. Patients and methods It is a retrospective descriptive study and analytical spread over 1  year, from January 2015 to December 2015. Were included all patients with pneumonia Ventilator, with or without bacteraemia, after hospitalization for more than 48  h. Were collected clinical, biological, radiological, bacteriological and scalable patients included. Results The average age of the patients was 42.19  years with a sex ratio (M/F) 1.47 APACHE II score average was 16.4, the average SAPSII was 31.12 and the average was 2.39 SOFA. 86.5% of patients were intubated on admission in intensive care. 3.5% were intubated within less than 5 days notice after admission and the average time to onset of VAP was 3.76 days. The VAP were early in 68.4% of cases. The average hospital stay of patients being 19.63 days. 73.68% in the sample used in our patients was bronchial aspiration. The most offending germs are 27.2% Acinetobacter baumannii, Pseudomonas aeruginosa 16.3%. 100% of patients received empirical antibiotic therapy chosen according to the ecology of the service. 77.4% died. 43.8% of deaths were directly related to VAP. Conclusion Ventilator acquired pneumonia is main problem in our ICU. The bacteriological ecology must be usually known. In our unit Acinetobacter baumanii is the main germ associated. P179 Determinants and prognostic factors of Acinetobacter baumannii ventilator‑associated pneumonia Khalid Abidi1, Tarek Dendane1, Ssouni Oussama1, Jihane Belayachi2, Naou‑ fal Madani2, Redouane Abouqal2, Amine Ali Zeggwagh1 1Medical intensive care unit, Mohamed V University Hopital Ibn Sina, Rabat, Morocco; 2Service des urgences médicales hospitalières ‑ ibn sina – université mohamed v – rabat, Hopital Ibn Sina, Rabat, Morocco Correspondence: Tarek Dendane ‑ Annals of Intensive Care 2017, 7(Suppl 1):P179 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%). P180 Ventilator‑associated pneumonia in the elderly: a study of the prognosis Hatem Ghadhoune1, Anis Chaari2, Guissouma Jihene1, Hend Allouche1, Insaf Trabelsi1, Habib Brahmi1, Mohamed Samet1, Hatem El Ghord1 1Réanimation médicale bizerte, Faculté de médecine de Tunis, Bizerte, Tunisia; 2Intensive care unit, King Hamad University Hospital, Muharraq, Bahrain Correspondence: Hatem Ghadhoune ‑ Annals of Intensive Care 2017, 7(Suppl 1):P180 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). Results During the study period, 86 patients were included. The causes of admission in the intensive care unit (ICU) were shock (n  =  30), acute respiratory failure (n  =  41) and disturbed level of consciousness (n  =  15). Diabetes mellitus, hypertension and chronic obstructive pulmonary disease were the most common comorbidities (44.2, 32.6 and 40.7% respectively). Mean age was 73.3  ±  6.5  years. Sex-ratio (M/F) was 1.8. Mean APACHE(II) score was 24 ± 9. The mean duration of mechanical ventilation was 10  ±  15  days. Thirty patients (34.9%) developed VAP. ICU-mortality was significantly higher in the VAP (+) group (90 vs 55.4%; p = 0.001). Multivariate analysis identified two independent factors predicting ICU mortality: Shock on admission (OR = 22.7, CI 95% [5.1–101.2], p < 0.001) and VAP (OR = 5.7, CI 95% [1.2–28.1], p = 0.033). Conclusion VAP is common in critically-ill elderly patients and is associated with worse outcome. Therefore, preventing its onset is of paramount importance. 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, piperacillin– tazobactam, 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: 11–25) 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. P182 Admission in intensive care unit for severe adverse drug event: what finding? Julien Arcizet1, Bertrand Leroy1, Caroline Abdulmalack2, Catherine Ren‑ zullo1, Maël Hamet2, Jean‑Marc Doise 2, Jérôme Coutet1 1Pharmacy unit, C.H. Chalon sur Saône William Morey, Chalon‑sur‑Saône, France; 2Intensive care unit, C.H. Chalon sur Saône William Morey, Chalon‑sur‑Saône, France Correspondence: Julien Arcizet ‑ julien.arcizet@ch‑chalon71.fr Annals of Intensive Care 2017, 7(Suppl 1):P182 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. Results On 1545 admissions, 154 patients were hospitalized for unintended ADE. The average age was 70  years old [26; 95], with a men/women ratio equal to 1.8. The clinical severity IGSII score found was 51 [13; 120]. Average length of stay in ICU was 5.5 days [1; 28] on average in this unit. The main reasons of admission were: hematologic failure (in particular hemorrhagic) (29.9%), metabolic failure (19.5%), renal failure (11.0%), neurological failure (11.0%) and sepsis (10.4%). Respiratory, hepatic, hemodynamic failures and hypersensitivity reactions represented respectively less than 10% of cases. 34 patients (22%) included died during their stays in ICU. On average, 7.6 drugs were found in the usual treatment of the patient. 32.5% of this population had a known cognitive disorder and 62% of them selfmanagement of their treatment. The main drugs involved were: furosemide (16.9%), metformin (13.0%), perindopril (9.1%), lysine acetylsalicylate (8.4%), warfarin (8.4%) and fluindione (7.8%). The most common drug families involved were: drugs of the cardiovascular system (33.8% of cases), anticoagulants and antiplatelet agents (31.8%), antidiabetics (16.9%) and psychotropic (13.6%). 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. Chaigar Mohammed Cheikh1, Zakaria Quechar1, Hanane Ezzouine1, 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 days in medical ICU of the UH Ibn Rochd of Casablanca. Statistical analysis was performed using SPSS 21.0. Results We enrolled 151 patients witch correspond to 8.2% of all admissions. The sex ratio was 1.6, the average age was (43.2  ±  18.5  years). The majority of patients (75%) were transferred from the emergency department. Medical pathology was the main motive of admission (82.1% of cases). Means of severity scores were calculated as follows: APACHE III (56.2  ±  22.9), IGS III (40  ±  9.6), OSF (0.6  ±  0.7). The average length of stay was (42.2  ±  59.7  days). The incidence of nosocomial infection was 82.8%, the average day of onset was 9.34  ±  8.7  days. Pneumonia, bacteremia and vascular catheter infections were the main sites, Gram-Negative Bacilli were the most frequently identified, dominated by Acinetobacter baumanii (27.9%). Mechanical ventilation (91.4%) and vascular catheterization (84.1%) are the most used invasive devices. Antibiotics (92.1%), sedation (91.4%) and vasopressors (60.9%) were the main administrated treatments. The outcome was favorable in 37.7% of cases. Hemodynamic instability (64.2%) and respiratory complications (61.6%) were the complications most frequently observed, septic shock occurred in 49% of cases. Mortality rate was 55%. In univariate analysis, the variables that have emerged as risk factors of mortality were: sex, length of hospitalization, severity scores (APACHE III, IGS III, OSF), the Charlson comorbidity score adjusted to age, traumatic pathology, the occurrence of nosocomial infection, septic shock, hemodynamic instability, neurological worsening, use of vasopressors, and tracheostomy. In multivariate analysis: nosocomial infection (p  =  0.04), hemodynamic worsening (p  =  0.03), use of vasopressors (p  <  0.01) and antibiotics (p  <  0.01) appeared to be risk factors of mortality. Conclusion Although patients hospitalized in ICU for more than 15  days are few, they represent a serious problem of care and an important part of the activity of intensive care (bed occupancy, care costs). P184 Admissions and readmissions to the intensive care unit of patients with hematologic malignancies: a 5 years retrospective study Magalie Joris1, Dimitri Titeca Beauport1, Loay Kontar1, Delphine Lebon2, Bérengère Gruson2, Michel Slama1, Jean‑Pierre Marolleau 2, Julien Maizel1 1Réanimation médicale, Centre Hospitalier Universitaire, Amiens, France; 2Hématologie clinique et thérapie cellulaire, Centre Hospitalier Universi‑ taire, Amiens, France Correspondence: Magalie Joris ‑ joris.magalie@chu‑amiens.fr Annals of Intensive Care 2017, 7(Suppl 1):P184 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. Results The mean age was 58.6  ±  14.8  years, with a male–female ratio of 1.55. The most represented malignancies were acute leukemia (40.8%) and non-Hodgkin lymphomas (26.8%); 16.2% were hematopoietic stem cell transplant recipients. 54% of patients had newly diagnosed malignancy, 20.8% were in complete remission (CR), 11.7% had stable disease or partial remission and 13.6% had progressive disease. 46.4% of patients presented with severe neutropenia at the time of ICU admission. The main lifethreatening complications precipitating ICU admission were acute respiratory failure for 43.8%, sepsis for 51.7%, acute kidney injury for 14%, neurological disturbance for 18.1% and preventing tumor lysis syndrome for 15.8%. 11.3% presented with hemophagocytic lymphohistiocytosis (HLH). 34.3% of patients received non-invasive ventilation, 46.8% mechanical ventilation (MV), 54.3% needed vasoactive drugs administration and 40.8% had renal replacement therapy. ICU, day 28 and 6  months mortality were 39.8, 46.4 and 63.4% respectively. By 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). 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. Reference 1. Azoulay E. Outcomes of critically ill patients with hematologic malignancies: prospective multicenter data from France and Belgium—a groupe de recherche respiratoire en réanimation onco‑hématologique study. J Clin Oncol. 2013;31:2810–18. P185 Prognosis of lung cancer patients admitted to ICU Julie Gorham1, Lieveke Ameye,2, Thierry Berghmans,1, Marianne Paes‑ mans2, Jean‑Paul Sculier, 1, Anne‑Pascale Meert 1 1Intensive Care and Thoracic Oncology, Institute Jules Bordet, Brussel, Belgium; 2Data centre, Institute Jules Bordet, Brussel, Belgium Correspondence: Anne‑Pascale Meert ‑ Annals of Intensive Care 2017, 7(Suppl 1):P185 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. Patients and methods We conducted a retrospective study including all patients with lung cancer admitted for a medical or surgical complication in the intensive care unit of a cancer hospital between September 1, 2008 and December 31, 2013. mbol’  >  [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. Results During this period, 212 ICU admissions occurred in 180 patients with lung cancer. The majority of them were men (64%), had non small cell lung cancer (80%) and metastases at the time of admission (81%). 54% received an antineoplastic therapy during the month preceding the ICU admission. 47 patients (26%) died during hospitalisation with 36 deaths in the intensive care unit and 11 in the hospital ward after ICU discharge. The three main reasons of admissions were: cardiovascular problems (32%), respiratory failure (29%) and neurological (16%) complications. The SAPS II score (OR 1.07; 95% CI 1.04–1.11) as continuous covariate and the presence of respiratory complications (OR 4.00; 95% CI 1.76–9.07) were the 2 factors independently affecting hospital mortality in multivariate analysis. Median overall survival since ICU admission was 3.7 months (95% CI, 2.8–4.4). Median overall survival for patients discharged alive from ICU was 4.8 months (95% CI, 3.9–5.6 months). Considering patients discharged alive from the hospital, only the presence of metastases was a statistically independent prognostic factor in multivariate analysis (HR 2.3; 95% CI 1.44–3.65). Discussion The prognosis of patients with lung cancer admitted in ICU improved probably due to a better selection of these patients eligible for intensive care. As already observed in general cancer patients’ population, prognosis factors for hospital mortality are related to the acute complications but survival after hospital discharge is dependent from the cancer stage. Conclusion Lung cancer patients are admitted in critical care in over half of the cases for cardiovascular and respiratory complications. The hospital mortality rate of these patients admitted in ICU was 26%. However survival after hospital discharge remains low and dependent of the cancer metastatic status. ICU admission should be considered for patients with lung cancer. Introduction Admission of cancer patients with poor prognosis in intensive care units (ICU), like acute myeloid leukemia (AML) resistant to the first course of induction chemotherapy, continue to be controversial. The ICU trial may be an alternative in this case to ICU refusal. The ICU trial is a full-code ICU admission followed by reappraisal of the level of care. The objective of this study was to find variables available at ICU admission and at day 3 in order to predict prognosis of critically ill medical patients with relapsed acute myeloid leukemia. Patients and methods Retrospective monocentric study of consecutive patients with a relapsed AML admitted to the 30 beds medical ICU of an academic hospital. We evaluated hematological treatments, organs supports and mortality in ICU. Results Between 2002 and 2014, 24 patients with relapsed AML were admitted in the ICU. At admission, patients were 54  years old, IGS 2: 64 ± 24, Lactates: 4.9 mmol/L (±4.7). Eight patients underwent bone marrow transplant (BMT). Five patients had graft-versus-host disease (GVHD). 12 patients were admitted for septic shock, 7 patients for acute respiratory failure, 2 for cardiac arrests, 1 for coma, 1 for acute kidney injury and 1 for hemorrhagic shock. BMT was significantly associated with higher mortality [Odds ratio (0R) 13.0 (95% confidence interval (95% CI) 1.7–99, 43)—p: 0.02]. 7 BMT patients died in ICU. Neutropenia [OR 0.33 (95% CI 0.05–1.87)—p: 0.4] and GVHD (OR 2.0 [95% CI 0.07– 51.6)—p: 1.0] were not able to predict mortality in ICU. The first day in ICU: 15 patients were under mechanical ventilation, 17 patients need vasopressor perfusion and 3 patients dialysis. Mortality in ICU was 37%. 4 patients died from acute illness before day 3. Among the 24 patients admitted in ICU, none of the life-sustaining interventions at admission were associated with mortality: invasive mechanical ventilation [OR 9.14 (95% CI 0.9–92.4) - p: 0.08], vasopressor perfusion [OR 6.2 (95% CI 0.6–62.2)—p: 0.18] and renal replacement therapy [OR 4.8 (95% CI 0.3–65.8)—p: 0.27]. On day 3, life supports were not associated with higher mortality: invasive mechanical ventilation [OR 3.7 (95% CI 0.32–41.1)—p: 0.35], vasopressor perfusion [OR 2.0 (95% CI 0.27–14.7)—p: 0.64] and renal replacement therapy [OR 1.2 (95% CI 0.08–16.45)—p: 1.0]. Conclusion Mortality in ICU was 37% in patients with relapsed AML. 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. In case of relapsed AML with BMT, ICU management is still challenging. P187 Mortality analysis of the chronically critically ill patients: an epidemiological prospective study Severine Couffin1, David Lobo2, Nicolas de Prost3, Nicolas Mongardon4, Gilles Dhonneur5, Roman Mounier2 1Surgical intensive care, Hospital Henri Mondor, Créteil, France; 2Anes‑ thesia and surgical intensive care, Hospital Henri Mondor, Créteil, France; 3Réanimation Médicale, Hôpital Henri Mondor, Créteil, France; 4Service de réanimation médicale, Hôpital Cochin, Paris, France; 5Anesthesia and intensive care medicine, CHU Henri Mondor, Créteil, France Correspondence: Severine Couffin ‑ Annals of Intensive Care 2017, 7(Suppl 1):P187 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 HLADR, pointing to the value of studying the immune system of these patients. 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]. P189 The hemorrhage postpartum: inventory Setti‑Aouicha Zelmat 1, Djamila‑Djahida Batouche 2, Fatima Mazour3, Belkacem Chaffi4, Nadia Benatta5 1Réanimation, etablissement hospitalier spécialisé 1er novembre, oran, Algeria; 2Réanimation pédiatrique, Centre Hospitalier et Universitaire d’Oran, Oran, Algeria; 3 Anesthesie ‑réanimation chirurgicale, EHS 1er Novembre, oran, Algeria; 4Service de gynéco‑obstétrique, EHS 1er Novembre, oran, Algeria; 5Cardiologie, Centre Hospitalier et Universitaire d’Oran, Oran, Algeria Correspondence: Setti‑Aouicha Zelmat ‑ Annals of Intensive Care 2017, 7(Suppl 1):P189 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 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. S85 Incidence and prognostic impact of new onset supraventricular arrhythmia in patient with septic shock Walid Sellami1, Zied Hajjej1, Soumaya Ben Yedder1, Walid Samoud1, Iheb Labbene1, Mustapha Ferjani1 1Department of critical care medicine and anesthesiology, Military Hospi‑ tal of Tunis, Tunisia, Tunis, Tunisia Correspondence: Walid Sellami ‑ Annals of Intensive Care 2017, 7(Suppl 1):S85 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. Before PP (T1) End of PP (T2) After PP (T3) dPmax (mmHg s−1) Lactates (mmol L−1) VT (ml kg−1 PBW) † p < 0.05 versus before PP; ‡ p < 0.05 versus end of PP; * p < 0.05 versus after PP 1. Jozwiak M. Beneficial hemodynamic effects of prone positioning in patients with acute respiratory distress syndrome. Am J Respir Crit Care Med 2013;188:1428–33. S87 Is the shape of the inferior vena cava a good tool to assess fluid responsiveness? A pilot study Hélène Cochet1, Jean Pierre Ponthus1, Virginie Amilien1, Martial Tchir1, Elise Barsam1, Mohsen Ayoub1, Jean Francois Georger1 1Réanimation polyvalente ‑ surveillance continue, Ctre Hospitalier Intercommunal de Villeneuve Saint Georges, Lucie et Raymond AUBRAC, Villeneuve‑Saint‑Georges, France Correspondence: Hélène Cochet ‑ Annals of Intensive Care 2017, 7(Suppl 1):S87 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 (9–21). 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 (Table 25; Fig. 21). Conclusion It is an alternative and simple technique once acquired. The limits are essentially established by technical conditions such as Table 25 Relevant parameters ICV respiratory variation (%) Fig. 21 See text for description the necessity of being very perpendicular at the risk of falsifying the EI. For now it is just a static index but an ovoïd vena cava (EI: 0.85) may be predicative of fluid responsiveness. The strangely big size of the VCI can be attributed to a deformation of the in-depth image because of the probe but this should not modify in theory the report of EI. It is a question in the future of testing this parameter at the spontaneous breathing patient, study his variability with respiration and strengthening its external validity. S88 Quality of reporting of fluid responsiveness evaluation studies: a five year systematic review Izaute Guillame1, Matthias Jacquet‑Lagrèze 1, Jean‑Luc Fellahi 1 1Anesthésie réanimation, Hopital Louis Pradel, Bron, France Correspondence: Matthias Jacquet‑Lagrèze ‑ Annals of Intensive Care 2017, 7(Suppl 1):S88 Introduction The Standards for the Reporting of Diagnostic Accuracy Studies (STARD) statement was developed in 2003 to improve the quality of reporting of diagnosis studies (1). STARD is a list of 25 essential items, it was updated in 2015 (2). Fluid responsiveness evaluation is a cornerstone of hemodynamic management of patient with circulatory failure. Numerous studies have evaluated tools for assessing fluid responsiveness. The purpose of this work was to evaluate the quality of reporting of studies assessing fluid responsiveness, using the STARD criteria. Materials and methods We used MEDLINE via PubMed to search all publication of studies assessing the ability to predict or diagnose fluid responsiveness in a perioperative or circulatory failure context between the 1st of January 2004 and the 1st of January 2014. We have presented herein the last 5 years (2010–2014). Paediatric studies, publication with no abstract available or in another language than English, review, meta-analysis, and journals with impact factor less than 2 were excluded. We have checked these studies for all the 25 STARD criteria. The rating methodology of each item was discussed and double checked (GI and MJL). The 2003 and 2015 STARD quality score (SQS) was the integer of the present item. Univariate and multivariate analysis were conducted to find out if some characteristic of the studies were linked with a better SQS. Following items were studied: year of publication, positive or negative study, location of study, expert team, funding source, conflict of interest, the use of a new device, instruction to the author including STARD statement, impact factor of the journal. Every covariate associated with a p-value lower than 0.20 in the univariate analysis were used in the multivariate analysis. Statistical analysis was performed with R software. All the test were two-sided, p < 0.05 was considered statistically significant. Results After a double screening (GI, MJL) of 430 articles (2010–2014), 97 were selected, then 46 were excluded because of an IF < 2. Fifty-one studies were selected and reviewed. The mean 2003 overall quality score was 11.85 on a scale of 0–25, whereas the mean 2015 overall quality score was 15.34 on a scale of 0–30. The 2003 SQS remained stable between 2010 and 2014 (p = 0.173). Correlation between 2003 and 2015 SQS was good (r = 0.83; p < 0.001). Some items are insufficiently reported, especially: participant sampling (adequately reported in 19.6% of studies), executing and reading the tests (11.7%), blindness (11.8%) and, the description of test reproducibility (19.1%). Some items are better reported: technical specifications and/or references to tests (65.1%), methods for calculating or comparing (90.2%), clinical and demographic characteristics (82.3%), diagnostic accuracy (90.2%). 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. S89 Neurological impairment in cirrhotic patients admitted to ICU: hepatic versus drug‑induced encephalopathy Julie Assaraf1, Simona Tripon2, Maxime Mallet3, Marika Rudler4, Julien Mayaux5, Dominique Thabut6, Nicolas Weiss7 1Unité de soins intensifs d’hépatogastroentérologie, Pitié‑Salpêtrière Hos‑ pital, Paris, France; 2Unité de soins intensifs d’hépatogastroentérologie, Hôpital Pitié‑Salpêtrière, Paris, France; 3Unité de Soins Intensifs. Service d’Hépato‑gastro‑entérologie du Pr Poynard, CHU Pitié Salpétrière, Paris, France; 4Hepatology and gastroenterology, Pitié‑Salpêtrière Hospital, Paris, France; 5Réanimation médicale, Hôpital Pitié‑Salpêtrière, Paris, France; 6Brain Liver Pitié‑Salpêtrière Study Group (BLIPS), Hôpital Pitié‑Salpêtrière, Paris, France; 7Unité de réanimation neurologique, Hôpi‑ tal Pitié‑Salpêtrière, Paris, France Correspondence: Julie Assaraf ‑ Annals of Intensive Care 2017, 7(Suppl 1):S89 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 ICU and to analyse its promoting factors, especially ammonia levels, infection and drug intake. Patients and methods Between February 2015 and September 2016, all cirrhotic patients admitted in ICU were prospectively included. We collected patient’s clinical and biological parameters, especially ammonia levels, all the ongoing treatments, and seeked for the presence of evolutive infection defined as an infection considered as clinically relevant. Cirrhotic patients having encephalopathy and ammonia levels above 75  μmol/L were considered as having HE. We performed univariate and multivariate analysis to estimate the association between ammonia level, infection or drugs intake and the presence of encephalopathy. Results 202 cirrhotic patients were included, they were 58 ± 11 yearsold, 71% were men. Etiology of cirrhosis was as followed: alcohol 110 (54%), virus 27 (13%), NASH 19 (10%), mixed 30 (15%), others 16 (8%). 122 patients (60%) displayed neurological impairment, i.e. encephalopathy at admission, 84 (42%) had ammonia levels above 75  μmol/L, 80 (40%) had evolutive infection and the following drugs were found: furosemide, hydrochlorothiazide, betablockers, lactulose, rifaximine, betalactamins, proton pump inhibitors (PPIs), quinolones, benzodiazepines, morphine derivatives, insulin, anti-epileptic drugs, corticosteroids and immunosuppressants. In univariate analysis, patients with encephalopathy had significantly higher Child and MELD scores, higher ammonia, bilirubin and leukocytes levels, more evolutive infections, and received more frequently PPIs, lactulose and quinolones as compared to those who had no encephalopathy. Patients with encephalopathy took also more frequently an association of agonist of efflux pumps (e.g. PPIs, betalactamins or quinolones) (p = 0.0005). In the group of patients displaying low ammonia levels (less than 75  μmol/L), encephalopathy was significantly increased in patients having infection (p < 0.0001), taking quinolones (p = 0.0086), PPIs (p  =  0.0196), and betalactamins (p  =  0.0438) whereas ursodeoxycholic acid treatment had a protective effect (p  =  0.0299). Multivariate analysis showed that high ammonia levels (p  <  0.0001), presence of infection (p < 0.0001), and taking quinolones (p = 0.0403) were independently associated with the presence of encephalopathy. Conclusion These results highlight the involvement of drugs in the physiopathology of neurological impairment in cirrhotic patients. Several drugs that were implicated can modify the activity of efflux pumps expressed on the BBB. Besides HE, cirrhotic patients are susceptible to develop drugs-induced encephalopathy regardless of ammonia level, underlying the necessity to carefully assess every treatment indication, given their potential neurotoxicity. S90 Functional outcome of prolonged hypoglycemic encephalopathy Guilaume Barbara1, Bruno Megarbane2, Laurent Argaud3, Guillaume Louis4, Nicolas Lerolle5, Francis Schneider6, Stéphane Gaudry7, Nico‑ las Barbarot8, Angéline Jamet,9, Hervé Outin10, Sébastien Gibot1, Pierre‑Edouard Bollaert 1 1Réanimation médicale, hôpital central, C.H.U. de Nancy, Nancy, France; 2Service de Réanimation Médicale et Toxicologique, CHU Lariboisière, Paris, France; 3Réanimation Médicale, Hospices Civils de Lyon ‑ Groupe ‑ ment Hospitalier Edouard Herriot, Lyon, France; 4Réanimation polyva‑ lente, Hôpital de Mercy, Ars Laquenexy, France; 5Réanimation médicale, Centre Hospitalier Universitaire d’Angers, Angers, France; 6Réanimation médicale, C.H.R.U. Hôpitaux Universitaires Strasbourg, Strasbourg, France; 7Service de réanimation médico‑chirurgicale, CHU Louis Mourier, Colombes, Colombes, France; 8Service de réanimation polyvalente, Centre hospitalier, Saint Brieuc, France; 9Réanimation médicale, CHU de Poitiers, Poitiers, France; 10Service de Réanimation Médico‑Chirurgicale, CHI de Poissy/Saint‑Germain‑en‑Laye ‑ Site Poissy, Poissy, France Correspondence: Pierre‑Edouard Bollaert ‑ pe.bollaert@chu‑nancy.fr Annals of Intensive Care 2017, 7(Suppl 1):S90 Introduction Little is known about the causes, clinical course and longterm outcome of prolonged hypoglycemic encephalopathy. We investigated functional long-term prognosis and identify prognosis factors of patients admitted in an Intensive Care Unit (ICU) with prolonged neurological manifestations related to obvious hypoglycemia. Patients and methods This was a retrospective study conducted in patients hospitalized from July 1, 2004, to July 1, 2014, in 9 polyvalent intensive care units in French university and general hospitals. Participants were adults admitted to the ICU with a Glasgow Coma Score <8 due to hypoglycemia <0.5  g/l and persistent consciousness disorders after normalizing blood glucose levels. Patients with possible other causes of consciousness disorders, previous cognitive disorders, hypothermia <35 °C or circulatory arrest within 24 h after ICU admission, were excluded. Follow-up phone call was used to determine the functional outcome using modified Rankin Scale at 1-year minimum (mRS) with mRS 0–3 defining good and mRS 4–6 poor outcomes. Results Forty-nine patients were included. Median age was 55 years. Causes of hypoglycemia were various, mainly including insulin or oral antidiabetic drugs abuse (65%) and neuroendocrine carcinoma (16%). Twenty (41%) patients died in the ICU, and a total of thirty-one (63%) patients had a poor outcome at 1-year follow-up. Twenty-two (45%) patients underwent therapeutic limitation, primarily due to no hope expected for improvement (86%). Among survivors with poor outcome at ICU discharge (n = 15), one died 10 months later and six improved their outcome at 1-year follow-up with a median decrease in mRS of −2 [IQR −2.5 to −1.75], including five patients who finally reached the good outcome group. Among the patients with good outcome on ICU discharge, one died, seven further improved their outcome at 1  year with a median decrease in mRS of −2 [IQR −2.0 to −1.0], whereas one patient deteriorated while remaining in the good outcome group (from mRS 1–2). On multivariate analysis, only low mRS prior to ICU admission (OR 2.6; 95% CI  =  1.1–6.3; P  =  .03) and normal brain imaging (OR 7.1; 95% CI 1.1–44; P  =  .03) were significantly predictive of outcome. All patients (n  =  15) who stayed hypoglycemic 480 min or more evolved poorly. Conclusion Poor outcome was observed in about 60% of this population of prolonged hypoglycemic encephalopathy. Some patients can recover satisfactorily over time though we did not identify any robust factor of good outcome. S91 Cerebrospinal fluid characteristics during status epilepticus: a descriptive and prognostic study Mathilde Holleville1, Stéphane Legriel1, Anne Laure Chateauneuf1, Sébast‑ ien Cavelot1, Gwenaëlle Jacq,1, Jean‑Denis Moyer 1, Jean Pierre Bedos1, Fabrice Bruneel1 1Réanimation médico‑chirurgicale, C.H. de Versailles, Le Chesnay, France Correspondence: Mathilde Holleville ‑ Annals of Intensive Care 2017, 7(Suppl 1):S91 Introduction The lumbar puncture is an important tool of the etiologic investigations during status epilepticus. Scarce data are available regarding cerebrospinal fluid pleocytosis directly induced by seizures. We aimed to describe the cerebrospinal fluid characteristics of the patients admitted to the intensive care unit for status epilepticus, and to report the clinical, biological and radiological data of these patients. Furthermore, to determine whether status epilepticus per se increases the cerebrospinal fluid pleocytosis, we specifically studied cerebrospinal fluid parameters after status epilepticus in adult patients after exclusion of the cases of status epilepticus induced by acute infection or neoplastic disease of the central nervous system. Patients and methods This monocentric retrospective study (2005– 2012) included patients of more than 18  years old admitted to the intensive care unit for status epilepticus and who had a lumbar puncture. Status epilepticus was defined by a single seizure lasting more than 5  min or multiple seizures without neurological recovery in between. A cerebrospinal fluid pleocytosis was defined by a cerebrospinal fluid cell count of more than 4/mm3. We analyzed demographic, medical, biological data and outcomes at 3  months and 1  year. The groups were compared using univariate and multivariate analysis. Cerebrospinal fluid pleocytosis was considered directly linked to seizures when etiologic investigations identified causes of status epilepticus which were not known to induce cerebrospinal fluid pleocytosis. Results Among 191 patients admitted to the intensive care unit for status epilepticus, 124 (65%) had a lumbar puncture and were included. Eight of them had a cerebrospinal fluid pleocytosis of infectious or neoplastic origin. Among the 116 remaining patients, 97 (84%  =  group B2) had normal cerebrospinal fluid parameters, and 19 (16%  =  group B1) had a cerebrospinal fluid pleocytosis considered as directly linked to seizures, with a median [Q1–Q3] cerebrospinal fluid cell count of 8/mm3 [7–20]. No other cerebrospinal fluid parameters changes were observed in the two groups, unless a slight increase in the cerebrospinal fluid lactate (mmol/l), but without significant difference between the two groups (3.1 [2.7–3.4] group B1 vs 2.9 [2.5–3.5] group B2). In the multivariate analysis, only myoclonic status epilepticus (OR 55.1; CI 95% 6.02–1103.88; p  =  0.001) and leukocytosis at the admission (OR 1.1 per 103/mm3 increment; CI95% 1.01–1.21; p  =  0.01) were associated to pleocytosis directly linked to status epilepticus. The intensive care unit mortality was significantly increased in patients with pleocytosis directly linked to status epilepticus (5 (26%) vs 7 (7%), p = 0.02), but hospital mortality was not significantly different between the 2 groups. 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. S92 Synek score and NSE to predict poor neurological outcome after cerebral anoxia and therapeutic hypothermia Dimitri Titeca Beauport1, Magalie Joris1, Philippe Merle2, Loay Kontar1, Antoine Riviere1, Bertand De Cagny1, Thierry Soupison1, Michel Slama1, Julien Maizel1 1Réanimation médicale, Centre Hospitalier Universitaire, Amiens, France; 2Explorations fonctionelles du système nerveux adulte, Centre Hospitalier Universitaire, Amiens, France Correspondence: Dimitri Titeca Beauport ‑ titeca.dimitri@chu‑amiens.fr Annals of Intensive Care 2017, 7(Suppl 1):S92 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. Materials and methods We conducted a retrospective monocentric study in our medical intensive care unit between November 01, 2013 and August 31, 2016. All patient having at least one EEG and H48-72 NSE dosage available were included. Patients dead from post cardiac arrest shock, stroke, moribund, spontaneous awaking, or missing data were excluded. Electroencephalograms were prospectively classified by electrophysiologist according to the Synek score and NSE were performed within H48-72 post cardiac arrest. Results Among 162 cardiac arrest survivors, 70 were retrospectively analyzed. All patients received sedation with midazolam and sufentanil and mild therapeutic hypothermia for 24 h. Forty-eight were classified as poor outcome (CPC ≥ 3) and 22 as good outcome (CPC 1–2). Serum NSE levels above 46 µg/l was associated with a poor neurological outcome with sensitivity of 75%, specificity of 100% and AUROC at 0.95. The Synek score ≥3 was also strongly associated with poor neurological outcome with sensitivity of 91.5%, specificity of 90.9% and AUROC at 0.95. Conclusion In the limits of this study, Synek score ≥3 on first EEG and H48-72 NSE  >  46  µg/l are two strong predictor of poor neurologic outcome. The prognostic value of the association of the NSE and the Synek score to predict a poor neurological outcome remain to be determined. 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. 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. Table 26 Characteristics of  the donors, deadlines of  the protocol and transplant recipients Characteristics of potential donors (n = 28) Age in years, mean [range] CrCl by Cockroft‑ Gault, mean (ml/mn) [range] Length of stay in ICU, mean (day) [range] Deadlines of the protocol (n = 13) Time from extubation to death, mean (min) [range] Warm ischemia time, mean ‘min) [range] Duration of vessel cannulation, mean (min) [range] Duration of normothermic regional perfusion, mean (min) [range] Duration of cold storage, mean (min) Characteristics of transplant recipients (n = 26) Age in years, mean [range] Length of stay in transplant unit, mean (day) [range] Duration before creatinine <250 μmol/L after transplantation, mean (day) [range] Serum creatinine (μmol/L) at the hospital discharge, mean [range] Delayed Graft Function, n (%) 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. S96 Is chronic obstructive pulmonary disease a risk factor for microaspiration in intubated critically‑ill patients? Thècle Degroote1, Emmanuelle Jaillette1, Farid Zerimech2, Balduyck Malika2, Saad Nseir1 1Centre de réanimation, Centre Hospitalier Régional Universitaire de Lille, Lille, France; 2Laboratoire de biochimie et biologie moléculaire, Centre Hospitalier Régional Universitaire de Lille, Lille, France Correspondence: Thècle Degroote ‑ Annals of Intensive Care 2017, 7(Suppl 1):S96 Introduction Microaspiration of gastric and oropharyngeal contaminated secretions occurs frequently in intubated critically-ill patients, and plays a major role in the pathogenesis of ventilator-associated pneumonia (VAP). At basic state, patients with chronic obstructive pulmonary disease (COPD) have an increased risk of microaspiration (due to gastro-esophageal reflux disease, pharyngo-laryngeal dysfunction…), this risk may even be more important under mechanical ventilation. The main purpose of this study is to determine if COPD is a risk factor for global abundant microaspiration (GAM) in intubated critically-ill patients. Materials and methods We gathered data about two prospective multicentric randomized trials focused on microaspiration in intubated patients. Data about COPD were retrospectively collected in order to complete previous data. Microaspiration of gastric and oropharyngeal secretions was respectively determined by quantitative measurements of pepsin and salivary amylase in all tracheal aspirates during the first 48 h after intubation. GAM was defined as the presence at significant level of pepsin (>200 ng/ml) and/or salivary amylase (>1685 UI/L) in at least 30% of the tracheal aspirates. In order to find GAM independent risk factors, we realized an univariate and multivariate analysis of the variables collected. Results Out of 448 patients included in the studies, 415 were analyzed among which 95 patients with COPD. 360 patients (87%) had GAM. Neither COPD diagnosis, nor spirometric severity nor specific therapeutics were associated with GAM. Risk factors for GAM in univariate analysis were the age, diabetes, low score in Glasgow Coma Scale (GCS), and no recourse to paralytic agents or vasopressors. After multivariate analysis, age was identified as independant risk factor for GAM (OR [IC 95%] =  1.03 [1.01–1.05], p  <  0.001); whereas high score in GCS (OR [IC 95%] = 0.93 [0.86–0.99], p = 0.03) and use of paralytic agents (OR [IC95%] = 0.46 [0.23–0.90], p = 0.02) were associated with less occurrence in GAM. GAM was not associated with any increase of length of mechanical ventilation, length of ICU stay, VAP incidence or surmortailty. 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. S97 Microscopic examination and quantitative culture of protected brush specimen in suspected episodes of ventilator acquired pneumonia: effect of prior antimicrobial treatment Jean‑François Llitjos 1, Marlène Amara2, Guillaume Lacave3, Jean Pierre Bedos3, Béatrice Pangon2 1Réanimation Médicale, Hôpital Cochin, Paris, France; 278150, Hospital Center De Versailles, Le Chesnay, France; 3Réanimation médico‑chirurgi‑ cale, Centre Hospitalier de Versailles, Le Chesnay, France Correspondence: Jean‑François Llitjos ‑ Annals of Intensive Care 2017, 7(Suppl 1):S97 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. S98 Implementation and impact assessment of a “ventilator‑bundle” at the university clinics of Kinshasa: before and after study José Mavinga1, Joseph Nsiala Makunza2, M E Mafuta3, Yves Yanga1, Amisi Eric1, Jp Ilunga1, Ma Kilembe1 1Anesthesia and Reanimation, Cliniques universitaires de kinshasa, Kinshasa, Democratic Republic of the Congo; 2Anesthésie‑Réanimation, Hôpital Privé d’Athis‑Mons ‑ Site Caron, Athis‑Mons, France; 3Scool of public health, Université de Kinshasa, Batiment administratif, Kinshasa, Democratic Republic of the Congo Correspondence: José Mavinga ‑ Annals of Intensive Care 2017, 7(Suppl 1):S98 Introduction In 2005, an international consensus conference took stock of the various measures to be implemented for the prevention of ventilator acquired pneumonia (VAP) [1]. These measures are often gathered in groups of 3 or 5 under the term of “ventilator-bundle.” The effectiveness of these “bundles” was poorly evaluated in African environment. Objective To establish a VAP prevention program and assess its impact on morbidity and mortality of patients under mechanical ventilation in our service. Patients and methods Prospective, mono centric, quasi-experimental before-after study. It took place in the intensive care unit of the University Clinics of Kinshasa in the Democratic Republic of Congo (DRC). This service is equipped with 8 beds and a respirator for two beds. The observational period (phase1) was carried out from February 1st to December 31st, 2014 and the intervention period (phase2) from February 15st, 2015 to February 15st, 2016. All consecutive patients intubated and mechanically ventilated for more than 48  h were included. Five preventive measures were held: hand hygiene, the elevation of the head of the bed at 30°–45°, the daily lifting of sedation, oral decontamination with chlorhexidine and control cuff pressure of the endotracheal tube. Compliance with this bundle was assessed by direct observation without the knowledge of caregivers. The diagnosis of “VAP” was held before a clinically modified sore (m CPIS) >6. The main outcomes were the incidence of VAP and mortality. The protocol for this study was approved by the Ethics Committee of the School of Public Health of the University of Kinshasa, under the approval number: ESP/EC/015/2015. 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  ×  106  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). Fig. 22 See text for description 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. S100 Nosocomial infections in intensive burn care unit Amel Mokline1, Achraf Laajili1, Helmi Amri1, Imene Rahmani1, Nidhal Mensi1, Lazheri Gharsallah1, Sofiene Tlaili1, Bahija Gasri1, Rym Ham‑ mouda1, Amen Allah Messadi1 1Burn care department, Trauma and burn center, Tunis, Tunisia Correspondence: Amel Mokline ‑ Annals of Intensive Care 2017, 7(Suppl 1):S100 Introduction Nosocomial infections (NI) are common in burn patients due to the loss of the first line of defense against microbial invasion, immunocompromising effects of burn injury, and invasive diagnostic and therapeutic procedures. The objective of this study was to identify the incidence of nosocomial infection (NI), the pathogens and their antibacterial patterns, and prognosis of these burn patients. Patients and methods A retrospective study was conducted in a 20 bed intensive burn care unit during 6  months. Patients were eligible for the study, if they met the following criteria: total burn surface area (TBSA)  >  10%, length of ICU stay ≥48  h, and infected in accordance with the criteria of the National Nosocomial Infections Surveillance (NNIS) and the criteria of the SFETB [1–2]. In this study, NIs were classified into four main groups: pneumonias, bloodstream infections (BSI), catheter related infections (CRI), and urinary tract infections (UTI). For included patients, skin levy, blood cultures, urine and sputum cultures were drawn during fever or clinical features of sepsis. Results During the 6-month study period, 190 patients were admitted to the ICU, 47 patients were included (24.7%). 32 were male and 15 female. The mean age was 41 ± 19 yr. The mean TBSA was 41 ± 24%. 63% were admitted from another hospital. Burn injuries were due to domestic accidents in 53% and self immolation in 25%. The mean 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. S101 Predictors of treatment failure for patients with pneumonia caused by Staphylococcus aureus methicillin‑sensitive in a intensive care unit: a retrospective study Pierre‑Antoine Allain 1, Nathallie Gault2, Catherine Paugam‑Burtz 1, Arnaud Foucrier1 1Anesthesiology and critical care, Hospital Center University Beaujon (AP‑HP), Clichy, France; 2Unité d’épidémiologie et recherche clinique, Hospital Center University Beaujon (AP‑HP), Clichy, France Correspondence: Arnaud Foucrier ‑ Annals of Intensive Care 2017, 7(Suppl 1):S101 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 Fig. 23 See text for description 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. S102 Ventilator‑associated pneumonia: never enough, never give up! Sahar Habacha1, Bassem Chatbri1, Aymen M’rad1, Youssef Blel1, Nozha Brahmi1, Yousra Bourbiaa2, Lamia Thabet2 1Departement of intensive care and toxicology, Centre d’Assistance Médicale Urgente, Tunis, Tunisia; 2Biology departement, Trauma Center, Ben Arous, Tunisia Correspondence: Sahar Habacha ‑ Annals of Intensive Care 2017, 7(Suppl 1):S102 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 FiO2 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 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. S103 Comparative performance of different automated weaning modes: a network meta‑analysis Arthur Neuschwander1, Looten Vincent2, Chhor Vibol1, Yavchitz Amelie1, Matthieu Resche‑Rigon 2, Jean Mantz1, Romain Pirracchio1 1Anesthésie réanimation, Hôpital Européen Georges‑Pompidou (AP‑HP), Paris, France; 2Biostatistiques, Hôpital Saint‑Louis (AP‑HP), Paris, France Correspondence: Arthur Neuschwander ‑ arthur.neuschwander@gmail. com Annals of Intensive Care 2017, 7(Suppl 1):S103 Introduction Mechanical ventilation (MV) weaning is a crucial step in critically ill patients. MV duration is associated with an increased risk of ventilator associated events, even though its specific impact on mortality has never been clearly demonstrated (1). Automated closed loop systems might help the weaning process. A recently published meta-analysis has reported a reduction in MV duration when using an automated weaning mode as compared to non-automated mode (2). However, the different automated modes have not been compared to each other. The objective of this network meta-analysis was to compare the performance of the three major automated weaning modes, i.e. the Automode°, the Smartcare° and the Adaptative Support Ventilation (ASV°) for MV weaning in critically ill and postoperative adult patients. Materials and methods We included all randomised control trials that compared automated closed loop weaning applications either to another automated application or standard care, including weaning according to a written weaning protocol or nurse driven protocols. The three modes of automated modes included in the study were ASV°, Smartcare° and Automode°. The primary outcome was the duration of MV weaning, defined as the time between randomization and a successful extubation. We also planned subgroup analyses in the ICU and the post-operative populations. The quality of the studies was assessed independently by two blinded investigators, using the evaluation recommended by the Cochrane Collaboration. A network Bayesian meta-analysis using random effect models and based on aggregate data from the included studies was performed using the gemtc package (R project, Vienna). This trial was declared in PROSPERO in August 2015 (CRD42015024742). Results Search of databased identified 776 articles; 233 were screened for eligibility after removal of duplicates. Abstract analysis led to the exclusion of 191 articles with a final full text analysis of 42 randomised Fig. 24 Bayesian NMA with random effect control trials. Ultimately, 15 trials were included in the analysis, representing 1424 ventilated patients. Nine studies included patients in the post-operative period while six were conducted in ICU. The automated mode was ASV° (A) in 9 studies, Smartcare° (C) in 4 studies and Automode° (B) in 2 studies. All studies reported the duration of MV weaning as defined in our protocol. In all studies, the control group was standard care with a weaning process driven either by nurses or physicians. In 12 studies (75%) a written weaning protocol was used in the control group. All ICU studies used sedation protocols based on sedation scores, none of them including systematic daily sedation interruption. Each one of the automated application was associated with a significant reduction in the duration of MV as compared to the control. When comparing all different modes using the network meta-analysis framework, ASV° appeared to be the best automated mode when it pertains to reducing the duration of mechanical ventilation weaning (Fig.  24). Subgroup analysis showed similar results in the post-operative and the ICU populations. Conclusion Compared to standard weaning practice, the 3 major automated weaning modes significantly reduced the duration of MV weaning in critically ill and post-operative adult patients. ASV° was associated with the most significant effect when compared to the two other automated modes (Smartcare°, Automode°). Further physiological respiratory studies would help to understand the underlying mechanisms accounting for the superiority of ASV. S104 Impact of proportional assisted ventilation on dyspnea and asynchrony in mechanically ventilated patients Côme Bureau1, Maxens Decavèle1, Sébastien Campion2, Roukia Ainsouya2, Marie‑Cécile Niérat 2, Julien Mayaux3, Hélène Prodanovic3, Mathieu Raux4, Thomas Similowski1, Alexandre Demoule1 1Réanimation médicale, inserm umr_s 1158 neurophysiologie respiratoire expérimentale et clinique, Pitié‑Salpêtrière Hospital, Paris, France; 2Inserm umr_s 1158 “neurophysiologie respiratoire expérimentale et clinique”, Pitié‑Salpêtrière Hospital, Paris, France; 3Unité de réanimation et de surveillance continue, service de pneumologie et réanimation médicale, Pitié‑Salpêtrière Hospital, Paris, France; 4Département d’anesthésie‑réani‑ mation et urm_s 1158, Hôpital Universitaire Pitié‑Salpêtrière, Paris, France Correspondence: Côme Bureau ‑ Annals of Intensive Care 2017, 7(Suppl 1):S104 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 Table 27 Breathing pattern, dyspnea score and  preva‑ lence of asynchrony according to condition RR (min−1) Fig. 25 Electromyographic of activity parasternal muscles. *p < 0. 05 compared to basal condition 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 [4–9]  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. Thomas Similowski1, Martin Dres1 Annals of Intensive Care 2017, 7(Suppl 1):S105 Introduction In intensive care unit (ICU) patients, diaphragm dysfunction is associated with adverse clinical outcomes. Ultrasound measurements of diaphragm thickness (Tdi), excursion (EXdi) and thickening fraction (TFdi) have been proposed as estimators of diaphragm function, but have never been compared to phrenic nerve stimulation. Our aim was to describe the relationship between Tdi, EXdi, TFdi and diaphragm function evaluated using the change in endotracheal pressure after phrenic nerve stimulation (Ptr,stim), and to compare their prognostic value. Patients and methods Ptr,stim and ultrasound variables were measured in mechanically ventilated (MV) patients <24  h after intubation (“initiation of MV”, under assist-control ventilation, ACV) and at the time of switch to pressure-support ventilation (“switch to PSV”). Diaphragm dysfunction was defined as Ptr,stim <11 cmH2O. Results 112 patients were included. At initiation of MV, Ptr,stim was not correlated to Tdi (rho  =  −0·13, p  =  0·28), EXdi (rho  =  0·10, p = 0·66) or TFdi (rho = −0·03, p = 0·80). At switch to PSV, TFdi and EXdi were correlated to Ptr,stim, (rho  =  0·87, p  <  0.001 and 0·45, p  =  0·001, respectively), but Tdi was not (rho =  −0·09, p =  0·45). At switch to PSV, a TFdi <29% could reliably identify diaphragm dysfunction (sensitivity and specificity of 85 and 88%, respectively), but Tdi and EXdi could not. This value was associated with increased duration of ICU stay and MV, and mortality. Conclusion Under ACV, neither Tdi, EXdi nor TFdi were related to Ptr,stim. Under PSV, TFdi was strongly correlated to diaphragm strength and, when decreased, was associated with poorer outcome. S107 End‑tidal carbon dioxide during spontaneous breathing trial to predict extubation failure: a single center prospective observational study Faten May1, Keyvan Razazi1, François Bagate1, Guillaume Carteaux1, Nico‑ las de Prost1, Armand Mekontso Dessap1 1Réanimation médicale, Hôpital Henri Mondor, Créteil, France Correspondence: Faten May ‑ Annals of Intensive Care 2017, 7(Suppl 1):S107 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 the end of SBT was similar between patients with failed and successful extubation: 38 [29–41] vs. 35 [30–40] mmHg, p = 0.9. 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. included by center was 27 (11–72). 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. S109 A multicenter prospective observational study of 1514 extubation procedures in 26 intensive care units: the FREE‑REA study Samir Jaber1, Hervé Quintard2, Audrey De Jong1, for the FREE‑REA study group 1DAR B, Hôpital Saint Eloi, Montpellier, France; 2Réanimation polyvalente, Hôpital Saint‑Roch, Nice, France Correspondence: Audrey De Jong ‑ Annals of Intensive Care 2017, 7(Suppl 1):S109 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 Fig. 26 Flow chart of the FREE‑REA study S110 Efficiency and safety of total plasma exchange in critically ill cirrhotic patients with acute on chronic liver failure admitted to the ICU Ilias Kounis1, Faouzi Saliba1, Stephane André1, Marc Boudon,1, Philippe Ichai1, Aline Younes1, Lionel Nakad1, Audrey Coilly1, Teresa Antonini1, Rodolphe Sobesky1, Eleonora De Martin1, Didier Samuel1 1Centre hépato‑billiaire, APHP Hôpital Paul‑Brousse, Villejuif, France Correspondence: Faouzi Saliba ‑ Annals of Intensive Care 2017, 7(Suppl 1):S110 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 evaluated in large trials in ACLF’ patients, with a liver transplant project, and a lesser degree of organ failure. S111 Management of enteral feeding during extubation in the intensive care unit: a multi‑center retrospective study in 11 French intensive care units Mickael Landais1, Noemie Hubert2, Mai‑Anh Nay 3, Johann Auchabie4, Bruno Giraudeau5, Reignier Jean6, Arnaud W Thille7, Stephan Ehrmann8 1Service de réanimation médicale, CHU Hôtel‑Dieu Nantes, Nantes, France; 2Réanimation, C.H.U de Caen, Caen, France; 3Réanimation médicale polyvalente, Hôpital de La Source, CHR Orléans, Orléans, France; 4Réanimation médicale, Centre Hospitalier Universitaire d’Angers, Angers, France; 5Unité de biostatistiques, Hôpital Bretonneau, Tours, France; 6Réanimation médicale, CHU Hôtel‑Dieu Nantes, Nantes, France; 7Réanimation Médicale, CHU de Poitiers, Poitiers, France; 8Réanimation polyvalente, CHRU Hôpitaux de Tours, Tours, France Correspondence: Mickael Landais ‑ Annals of Intensive Care 2017, 7(Suppl 1):S111 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%). Median fasting duration before extubation was 6  h [2; 13]. Despite a higher severity of disease of patients undergoing fasting before extubation (SAPS II of 50 [42; 66] vs. 48 [31; 60], p = 0.05). The rate of extubation failure was similar between the fasting patients and the others: 15 of 64 patients (23%) versus 6 of 36 (17%) of the patient (p = 0.60). 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. S112 More than half the patients receiving non‑invasive ventilation are fasting Nicolas Terzi1, Michaël Darmon2, Jean Reignier3, Stephane Ruckly4, Maïté Garrouste‑Orgeas 5, Elisabeth Gratia1, Alexandre Lautrette6, Elie Azoulay7, Bruno Mourvillier8, Laurent Argaud9, Laurent Papazian10, Marc Gainnier11, Dany GoldgranT‑oledano 12, Samir Jamali13, Anne Sylvie Dumenil14, Carole Schwebel15, Jean‑François Timsit 16, OUTCOMEREA study group 1Service de réanimation médicale, Clinique de Réanimation Médicale, Grenoble, France; 2Réanimation Médicale, CHU Saint‑Etienne ‑ Hôpital Nord, Saint‑Étienne, France; 3Réanimation médicale, CHU Hôtel‑Dieu Nantes, Nantes, France; 4Reanimation, Hôpital Bichat‑Claude Ber ‑ nard (AP‑HP), Paris, France; 5Réanimation, Fondation Hopital Saint Joseph, Paris, France; 6Réanimation médicale, CHU Gabriel‑Montpied, Clermont‑Ferrand, France; 7Réanimation médicale, Hôpital Saint‑Louis, Paris, France; 8Réanimation Médicale et Infectieuse, GH Bichat Claude Bernard, Paris, France; 9Réanimation Médicale, Hospices Civils de Lyon ‑ Groupement Hospitalier Edouard Herriot, Lyon, France; 10Service de réanimation‑détresses respiratoires et infections sévères, Hôpital Nord, Marseille, France; 11Réanimation des urgences médicales, Hôpital de la Timone, Marseille, France; 12Réanimation polyvalente, Centre Hospitalier Général, Gonesse, France; 13Réanimation médicale, Centre Hospitalier Sud Essonne, Dourdan, France; 14Réanimation chirurgicale, Hôpital Antoine Béclère, Clamart, France; 15Réanimation médicale, C.H.U. Grenoble, Greno‑ ble, France; 16Réanimation médicale et infectieuse, Hôpital Bichat‑Claude Bernard, Paris, France Correspondence: Nicolas Terzi ‑ Annals of Intensive Care 2017, 7(Suppl 1):S112 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 PaCO2 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. S113 Nutritional support in patients receiving temporary extracorporeal life support: a retrospective cohort study Arthur Bailly1, Laurent Brisard1, Philippe Bizouarn1, Thierry Lepoivre1, Johanna Nicolet1, Jean Christophe Rigal1, Jean Christian Roussel2, Ber‑ trand Rozec1 1Réanimation ctcv transplantation thoracique, CHU de Nantes ‑ Hôpital Nord Laennec, Saint‑Herblain, France; 2Chirurgie ctcv transplantation tho‑ racique, CHU de Nantes ‑ Hôpital Nord Laennec, Saint‑Herblain, France Correspondence: Laurent Brisard ‑ laurent.brisard@chu‑nantes.fr Annals of Intensive Care 2017, 7(Suppl 1):S113 Fig. 27 Daily proportion of target energy delivered on first 8 days Introduction The optimal nutritional intake in patients receiving temporary extracorporeal life support (ECLS), including extracorporeal membrane oxygenation (ECMO) venovenous (VV) or venoarterial (VA), remains controversial. Enteral nutrition (EN) is suspect to increase risk of gastrointestinal (GI) intolerance and intestinal ischemia. So, total parenteral nutrition (TPN) is often preferred. The purpose of this study is to describe the nutrition practices for critically ill patients receiving ECLS and identify opportunities for improving nutrition therapy in this population. Patients and methods Retrospective analysis of patients requiring ECMO-VA or ECMO-VV between 2010 and 2014 in the cardiac surgery intensive care unit of the University Hospital of Nantes. Nutritional support was daily monitored with parenteral intake (glucose, lipid and propofol, protein and albumin, parenteral nutrition) and enteral nutrition until ECLS weaning. Two groups were compared during ECLS period: no enteral nutrition delivered (none or TPN) (ANEC, n  =  73) and at least once enteral nutrition delivered (NEC, n  =  50) including EN alone and supplemental parenteral nutrition (SPN). Primary outcome was incidence of GI intolerance and risk factors. Secondary outcomes were nutritional adequacy (calculated as overall of calories and protein delivered divided by the theoretical amount requirements: 20 kcal/kg/d and 1.2 g/kg/d) and clinical outcome. Data are reported as median (25th and 75th percentiles) or number (%), and analyzed with student’s t test for continuous variables and χ2 test for categorical variables. P < 0.05 was considered as significant. Results 123 patients were enrolled [age 49  years (38–59), IGS-II 51 (35–67), IMC 24 (22–29)], represented 1009 nutrition days [ANEC = 428 (42), NEC = 581 (58)] with duration of 7 days (5–10) [ANEC = 6 (5–8), NEC  =  9 (6–13); P  <  0.001] including 96 ECMO-VA and 27 ECMOVV. None nutrition support and TPN were respectively received on 42% (n  =  428) and 32% (n  =  321) of patient days (NEC  =  105, SPN = 155). Patients received median intake of 1188 kcal (751–1784) [NEC = 1458 kcal (952–2016), ANEC = 990 kcal (605–1389); P = 0.001]. The median ratio of prescribed/required calories per day was 83% (50–127) [NEC  =  105% (61–144), ANEC  =  70% (42–101); P  =  0.001]. Evolution of daily energy balance is reported in Fig.  27. Digestive intolerance was reported in 27 patients (54%) including gastric residual volume considered high for physician in 15 cases (30%). Motility agents were used in 18 patients(36%). No serious GI complications or clinical signs of mesenteric ischemia were reported. Risk factors for GI intolerance were low weight [66 kg (56–75) vs. 80 kg (68.5–86); P = 0.034] and long duration of circulatory support [8 days (5–9.5) vs. 11  days (6.5–14.5); P  =  0.018]. Clinical outcome was similar between the groups (NEC vs. ANEC; ECMO-VA vs. ECMO-VV). Discussion TPN was preferably used in this specific cohort of critically ill patients. Although, the number of patients receiving no nutrition 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. S114 Refeeding hypophosphoremia in a medical critical care unit: 3‑month observational study Gioia Gastaldi1, Cherifa Cheurfa2, Julien Abily1, Steven Grange1, Dorothée Carpentier1, Christophe Girault1, Gaetan Beduneau1, Thomas Lescot3, Fabienne Tamion1 1Réanimation médicale, Centre Hospitalier Universitaire Rouen, Rouen, France; 2Anesthésie réanimation, Centre Hospitalier Universitaire Rouen, Rouen, France; 3Réanimation chirurgicale digestive, Hôpital Saint‑Antoine, Paris, France Correspondence: Gioia Gastaldi ‑ gioia.gastaldi@chu‑rouen.fr Annals of Intensive Care 2017, 7(Suppl 1):S114 Introduction Refeeding syndrome (RS) is a potentially lethal condition that remains underdiagnosed. It is characterized by severe electrolyte and fluid shifts associated with metabolic abnormalities in malnourished patients undergoing refeeding orally, enterally, or parenterally. Clinical criteria have been proposed for determination of its risk and reported in the National Institute for Clinical Excellence (NICE) Clinical Guidelines. Hypophosphoremia (hP) is a prominent feature of the RS and seems to be the earliest abnormality. Phosphorus is a vital component of nucleic acids, enzyme systems, and various metabolic pathways. Objective To determine the incidence of refeeding hypophosphoremia (RH)  <  0.9  mmol/L, and severe RH  <  0.6  mmol/L in a medical critical care unit. Patients and methods Monocentric, retrospective and observational study with patients from FRench-speAking icu Nutritional Survey study FRANS. Critically ill adults (more than 18 yo) were enrolled if they were hospitalized for more than 3 days during a 3-month period and had an artificial nutritional support. Refeeding hypophosphoremia is defined by the occurrence of hypophosphoremia after refeeding. We studied the incidence of HR, risk factors, and prognosis. Results 34 patients were enrolled between 03/01/2015 and 05/31/2015. RH appears in 73.5% and severe RH < 0.6 mmol/L in 35.3% (Fig. 28). There is no correlation between RS risk factors and RH in our study. Logistic regression did not permit to identify neither risk factor nor prognostic modification. There is a lack in phosphoremia measuring (27.6%), and overfeeding during the first 3 days occurs in 29.4%. Discussion We define that an hypophosphoremia appearing after refeeding is a refeeding hypophosphoremia, and we do not consider others etiologies, such as mechanical ventilation, alkalosis, sepsis, alcoholism, malabsorptive states, poor intake, some medication. Our cohort is too small to find some possible correlations with risk factors or prognosis. Conclusion Refeeding hypophosphoremia is common in our population. Hypophosphoremia is not an independent predictor of ICU or in-hospital mortality in critically ill patients. The knowledge of the SRI requires the follow-up of the phosphoremia during nutrition after critical illness in particular in the undernourished patients. Fig. 28 Incidence of hypophosphoremia at admission, the first day, and refeeding hypophosphoremia Introduction To determine the possible relationship between 10 days cumulated proteins (10  days CPD) and energy deficits (10  days CED) observed in ventilated patients and ICU length of stay, duration of ventilator support, incidence of infections and 28 days mortality. Patients and methods Mixed medical or surgical ventilated for at least 2 days adult patients from 5 ICUs from CHU Liège Belgium were enrolled into the study. They were fed by enteral route with a target of 25 kcal and 1.25 g of proteins by corrected kg of bodyweight and by day. If 50% of the target was not reached on day seven, parenteral nutrition was added with the same target. CED and CPD were calculated for 10  days, taking into account all the sources of nutrition, and was defined as the difference between the amount of energy or protein intake and the target. Results From 10/12/2014 till 30/5/2015, 99 patients were followed. Data from 62 patients could be cumulated on the first 10 days. There were 45 males, mean BMI was 26.9 ± 6.3; SAPS II score on day 1 was 51.3 ± 15.8, SOFA score at day 1 was 9.1 ± 3.5. They were ventilated for a median of 18  days (IQR 13–26), median ICU length of stay was 27  days (IQR 19–35). Mean SOFA max calculated for the first 10  days was 12.1  ±  4.2 and the 28  day mortality was 32.3%. On day 5, only 42% reached the target of 25 kcal/kg and 27% the target of 1.25 g of protein/kg. Mean 10 days CED was −5555.2 ± 4639.9 kcal and mean 10  days CPD was −350.8  ±  252.3  g. There was a significant negative relationship between both deficits and the SOFA max (p  =  0.0064 for CED and p  =  0.0053 for CPD). However, there were no correlations between any of the deficits and ICU length of stay, duration of mechanical ventilation, occurrence of infections and 28 days mortality. Discussion Saps II level, SOFA max level, ICU length of stay, all these parameters emphasize the high severity of this cohort of patients. It could indeed been thought that it is in this group of critically ill patients that the impact of nutrition could be easily demonstrated. Clear relationships between SOFAmax on day 10 and the 10 days CED and CPD could be seen. However, both the deficit and the level of organ dysfunctions could be cause or consequence. Unlike previous studies, usually performed in less severely ill patients, we did not find any relationship between CED or CPD and patient’s outcome. Conclusion Contrary to some recent studies, we found no relationship between CED and CPE and outcome of patients. Future studies are needed. S116 Cardiopulmonary bypass induces lymphopenia and decreases lymphocyte proliferation ability: IL‑10 and PD‑L1 as potential therapeutic targets to reduce postoperative infection Fabrice Uhel1, Mathieu Lesouhaitier1, Murielle Grégoire2, Baptiste Gaudriot3, Arnaud Gacouin1, Yves Le Tulzo1, Erwan Flecher4, Karin Tarte5, JeanM‑arc Tadié1 1Réanimation médicale, Centre Hospitalier Universitaire de Rennes, Rennes, France; 2Biosit and inserm u917, faculte de medecine, université rennes 1, Centre Hospitalier Universitaire de Rennes, Rennes, France; 3Anesthésier‑éan‑ imation, Centre Hospitalier Universitaire de Rennes, Rennes, France; 4Service de chirurgie thoracique, cardiaque et vasculaire, Centre Hospitalier Universi‑ taire, Rennes, France; 5Biosit and inserm u917, faculte de medecine, universite rennes 1, Centre Hospitalier Universitaire de Rennes, Rennes, France Correspondence: Jean‑Marc Tadié ‑ jeanmarc.tadie@chu‑rennes.fr Annals of Intensive Care 2017, 7(Suppl 1):S116 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. S117 Influence of neutropenia on mortality of critically ill cancer patients: results of a systematic review on individual data Quentin Georges1, Elie Azoulay2, Djamel Mokart3, M Soares4, Kyeongman Jeon5, Sandra Oeyen6, Chin Kook Rhee7, Pascale Gruber8, Marlies Oster‑ mann9, Quentin Hill10, Peter Depuydt11, Christelle Ferra12, Alice Muller13, Virginie Lemiale2, Bourmaud Aurelie14, Michaël Darmon1 1Réanimation médicale, CHU Saint‑Etienne ‑ Hôpital Nord, Saint‑Étienne, France; 2Réanimation médicale, Hôpital Saint‑Louis, Paris, France; 3Réanimation, Institut Paoli‑Calmettes, Marseille, France; 4Post‑graduation program, D’Or Institute for Research and Education, Rio de Janeiro, Brazil; 5Department of critical care medicine and division of pulmonary and critical care medicine, Sungkyunkwan University, Seoul, Republic of Korea; 6Department of intensive care, Ghent University hospital, Gent, Belgium; 7Division of pulmonary, allergy and critical care medicine, Seoul St. Mary’s Hospital, Seoul, Republic of Korea; 8Intensive care, anaesthesia, and surgery, The Royal Marsden Hospital, London, United Kingdom; 9Nephrology and intensive care, St Thomas’ Hospital, London, United Kingdom; 10Hematology, Leeds Teaching Hospital, Leeds, United King‑ dom; 11Department of intensive care, Ghent University Hospital, Ghent, Belgium; 12Hematology, Catalan Institute of Oncology, Barcelona, Spain; 13Department of critical care medicine and pulmonary medicine, hOSPI‑ TAL DE CLINICAS DE PORTO ALEGRE, Porto Alegre, Brazil; 14Département de santé publique, Institut de Cancérologie de la Loire Lucien Neuwirth, Saint‑Priest‑enJ‑arez, France Correspondence: Michaël Darmon ‑ michael.darmon@chu‑st‑etienne.fr Annals of Intensive Care 2017, 7(Suppl 1):S117 Introduction The prognostic impact of neutropenia in criticallyill cancer patients remains controversial. Hence, several studies in critically ill cancer patients failed to demonstrate the impact of neutropenia on outcome [1]. This lack of statistical association might however, reflect a lack of statistical power. A previous meta-analysis of aggregated data suggested 11% (95% CI 9–14%) raw increase in mortality in neutropenic patients. The available data were, however insufficient to allow adjustment with confounders [2]. The aim of this study was to assess the influence of neutropenia on mortality of critically ill cancer patients using individual data obtained from studies identified by our systematic review. Secondary objectives were to assess the influence of neutropenia on mortality of critically ill patients while taking into account underlying malignancy, use of G-CSF or changes related to period of admission. Patients and methods This systematic review and meta-analysis was performed according to the PRISMA statements. Public-domain databases including PubMed and the Cochrane database were searched by using predefined keywords. The research was restricted to articles published in English and studies focusing on critically ill adult patients from May 2005 to May 2015. The methods and objectives of this systematic review were reported in the PROSPERO database (CRD42015026347). Selected manuscripts’ authors were then contacted to obtained part of their dataset. Mortality was defined as either hospital or day-28 mortality. This preliminary analysis reports results from the whole dataset before and after adjustment using logistic regression. Period of admission and use of G-CSF were then assessed and were a pre-planned analysis. Results Our initial search yielded 1528 citations and 131 studies were retained for further analysis. Overall, 9 studies were excluded for redundancy with other included studies, 5 as containing only neutropenic patients, and two as containing only palliative patients. Finally 30 datasets (26%) containing sufficient data to allow comparison were obtained from authors. Overall, 7356 patients were included in this study, including 1666 patients with neutropenia at ICU admission. Median age was of 60  years (IQR 49–69). Median SAPSII score at ICU admission was 42 (IQR 28–57). Respectively 4101 and 3255 patients had underlying haematological malignancy and solid tumours, and 438 patients underwent allogeneic stem cell transplantation. Mechanical ventilation, vasopressors, and renal replacement therapy were required in 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. S118 Cytokinic profiles kinetic in response to Candida bloodstream infections Christopher Niles1, Fabien Herbert2, Sylviane Pied2, Séverine Loridant,3, Nadine François3, Anne Bignon4, Boualem Sendid5, Julien Poissy1 1Pôle de réanimation, hôpital salengro, C.H.R.U. ‑ Lille, Avenue Oscar Lambret, Lille, France, Lille, France; 2Centre d’infection et d’immunité de lille equipe 4 ‑ basic and clinical immunity of parasitic di, Institut Pasteur de Lille, Rue du Professeur Calmette, Lille, France, Lille, France; 3Centre de biologie et pathologie génétique, laboratoire de mycologie et parasitolo‑ gie, Centre Hospitalier Régional Universitaire de Lille, Lille, France; 4Service de réanimation chirurgicale, hôpital huriez, Centre Hospitalier Régional Universitaire de Lille, Lille, France; 5Inserm u995‑2, Universite Lille 2 ‑ Droit et Santé, Lille, France Correspondence: Julien Poissy ‑ Annals of Intensive Care 2017, 7(Suppl 1):S118 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. S120 Neutrophil‑to‑lymphocyte ratio as an independent predictor of mortality in critically ill cirrhotic patients Mikhael Giabicani1, Caroline Lemaitre1, Emmanuel Weiss2, Steven Grange1, Dorothée Carpentier1, Gaetan Beduneau1, Christophe Girault1, Catherine Paugam‑Burtz 2, Fabienne Tamion1 1Intensive care, Hospital Center University Rouen, Rouen, France; 2Anes‑ thesiology and critical care, Hospital Center university Beaujon (AP‑HP), Clichy, France Correspondence: Mikhael Giabicani ‑ Annals of Intensive Care 2017, 7(Suppl 1):S120 Introduction Prognosis of cirrhotic patients hospitalized in intensive care unit (ICU) remains poor. Many studies suggested a negative impact of systemic inflammation on organ failure and outcome in cirrhosis(1). In ICUs, cirrhotic patients are widely admitted and revalued after receiving optimal treatments for 3 days. However, little is known about how manage these patients after day 3 according to their prognosis. The blood neutrophil-to-lymphocyte ratio (NLR) as a novel inflammation index biomarker has been reported to be a predictor of clinical outcomes in various malignancies and in unselected critically ill patients(2). NLR has also been identified as a predictor of mortality in patients with stable liver cirrhosis. To our knowledge, the ability of NLR to predict outcome in critically ill cirrhotic patients has never been studied. The aim of this study was to evaluate the usefulness of inflammatory marker such as NLR for diagnosis of infection and predicting the outcome in hospitalized critically ill cirrhotic patients. Patients and methods We performed a retrospective monocentric study including consecutively cirrhotic patients hospitalized in a medical ICU from 2010 to 2014. For each patient, clinical and biological data at admission and day 3 were collected. NLR at admission (“NLRD0”), at day 3 (“NLRD3”) and the variation of NLR between admission and D3 Fig. 29 Multivariate analysis of factors associated with mortality (“delta NLR”) were calculated. Statistical analysis used appropriate non parametric tests and Cox regression for survival analysis. The ability of the variables to discriminate survivors from non-survivors was determined using ROC curves. Results are expressed as median (IQR). Results During the study period, 149 cirrhotic patients were admitted in ICU. The etiologies of liver cirrhosis were alcoholic in 87% of cases with severe score: median Child-Pugh score  =  9 [7–11], median MELD score  =  25 [18.3–30.8]. Main reasons for ICU admission were sepsis (40.9%), gastrointestinal bleeding (23.5%) and renal failure (19.5%). NLRD0 was higher for patients hospitalized for septic shock (p  <  0.001). Patients were followed up for 23  days [5.5–289.5]. 90 (59.6%) patients died including 57 (37.7%) deaths in ICU and 18 (12.1%) deaths after ICU discharge during the same hospitalization. NLR decreased for survivors between D0 and D3 (−4.5 [−9.8 to 0.0]) whereas it increased for non-survivors (+1.2 [−3.1 to +10.1]). In univariate analysis, for predicting survival, higher values of NLRD3, delta NLR, MELD score at admission, SOFA score at admission and at day 3 and delta SOFAD0-D3 were significant factors. Predictors of death in multivariate analysis are shown in Fig.  29. Area under delta NLR ROC curve was 0.72 (CI = 0.61–0.83). Conclusion The blood NLR is a novel inflammation index that has been shown to independently predict poor clinical outcomes. We have demonstrated that delta NLR is an independent predictor of mortality in critically ill cirrhotic patients. Delta NLR could help to identify cirrhotic patients at risk of unfavorable outcome. References 1. Salciccioli JD, et al. The association between the neutrophil‑to ‑lympho ‑ cyte ratio and mortality in critical illness: an observational cohort study. Crit Care. 2015;19:13. 2. Gandoura S, et al. Gene‑ and exon‑ expression profiling reveals an exten‑ sive LPS‑induced response in immune cells in patients with cirrhosis. J Hepatol. 2013;58(5):936–48. S121 Serum procalcitonin is a good diagnostic marker of bacterial infection in patients with hematologic malignancies in the Intensive Care Unit Celine Dupre1, Saad Nseir2, Anne‑Sophie Moreau 1 1Centre de réanimation, C.H. Régional Universitaire de Lille (CHRU de Lille), Lille, France; 2Centre de Réanimation, Centre Hospitalier Régional Universitaire de Lille, Lille, France Correspondence: Celine Dupre ‑ Annals of Intensive Care 2017, 7(Suppl 1):S121 Introduction Diagnosis of infection in immunocompromised patients can be difficult. However, diagnosing infection is very important, particularly in critically ill. This study aims to evaluate the benefit of procalcitonin (PCT) blood level as a diagnostic marker for bacterial infection in patients with hematological malignancies admitted to the Intensive Care Unit (ICU). Patients and methods This retrospective single-center study included all consecutive patients with acute myeloid leukemia or high grade lymphoid malignancy admitted to the ICU. Patients were sorted into three subgroups, according to clinical and microbiological data: «Infectious disease», «no infectious disease» and «Unknown». Initial serum PCT and when available at day 3 and day 5 were recorded. Receiver Operating Characteristic (ROC) curve, sensitivity and specificity were calculated. Serum PCT was considered as decreasing when the decrease was ≥25% at day 3 and/or ≥50% at day 5. Mortality rates in the ICU and at day-90 were also studied. Results Fifty-four patients were included in the study. At diagnosis, PCT levels were significantly different between the “Infection disease” group and the “No infection disease” group (p  =  0.002). There was no difference between the “Infection disease” group and the “Unknown” group (p  =  0.052). For the diagnosis of bacterial infection, best initial serum PCT threshold was 0.65  ng per milliliter. For that threshold, sensitivity was 94.6% and specificity was 78.6%. PCT area under the ROC curve was 0.93 [CI 95%  =  0.855–1]. Youden’s  J statistic was 0.73. PCT levels weren’t different between groups according to the presence of neutropenia or in case of inaugural disease. There was a significant difference in PCT values between groups according to the SOFA score (p  =  0.027), but not the SAPS2 score. Mortality rate in the ICU and at day-90 were significantly lower for the patients with decreasing PCT (p < 0.001 and p < 0.001, respectively). When comparing serum PCT and CRP predictive values, PCT was significantly a better marker of bacterial infection (Fig. 30). Discussion We found that serum PCT, with a threshold of 0.66  ng/ mL, is a reliable marker of bacterial infection disease in patients with aggressive hematological malignancy admitted to the ICU. Our study confirms the results of a previous study in unselected immunocompromised patients admitted to the ICU, showing a 100% sensitivity, a 63% specificity and an area under ROC curve of 0.851 [0.78–0.92] for a threshold of 0.5 ng/mL (1). The main limitations of our study are its retrospective design and the small number of included patients. Conclusion PCT is a reliable marker of bacterial infection in patients with hematological malignancies admitted to the ICU. PCT kinetic seems to be an interesting prognostic marker in this population. Aymen Zayene1, Lucie Portier2, Nathalie De Freitas Caires2, Philippe Las‑ Introduction Endocan is a proteogycan secreted by pulmonary and renal endothelial cells (Lassalle et  al.). Its synthesis is increased in sepsis and is regulated by Lipopolysaccharide (LPS) and cytokines. Toll Like Receptors (TLRs) are receptors of innate immunity and recognize diverse exogenous and endogenous patterns. In this study, we have compared the effect of 2 TLRs agonists on regulation of secretion and expression of Endocan by Human Umbilical Veinous Endothelial Cells (HUVECs). Materials and methods The stimulation of HUVECs was performed by 2 TLRs agonists: LPS and PolyInossinic Polycytidylic acid (Poly (I:C)). The secretion of Endocan was evaluated by Enzyme Linked Immuno Sorbant Assay (ELISA) in the supernatant of stimulated HUVECs. Real Time Polymerase Chain Reaction (qPCR) study compared the expression of Endocan in the two groups. Results The secretion of Endocan by HUVECs is upregulated by LPS and poly (I:C). Kinetics of secretion of Endocan was faster with LPS than Poly (I:C). q PCR study confirmed that Endocan is overexpressed in HUVECs stimulated with LPS than Poly (I:C). Discussion TLR 4 is involved in the recognition of LPS (Kawai et  al.), a component of the outer membrane of Gram Negative Bacteria, respiratory syncytial virus fusion proteins, Steptococcus pneumoniae pneumolysin. TLR 3 was originally identified as recognizing a synthetic analog of double stranded Ribonucleic Acid (dsRNA), poly(I:c) which mimics viral infections. In this study, we have found that kinetics of secretion and expression of Endocan is faster with HUVECs stimlated by TLR 4 agonist than TLR 3 agonist. This results could suggest that Endocan may be not only a marker of septic shock but could be also a specific marker to recognize the nature of pathogenic microorganisms in septic shock. Furthermore, other studies with more TLR agonists could be useful to confirm these results. Conclusion Studying the effects of diverse TLRs agonists could make the plasmatic dosage of Endocan more specific and helpful to recognize the nature of pathogenic microorganisms in septic shock. PHYSIOTHERAPISTS ABSTRACTS O79 Lung ultrasound: help to the diagnostic and the monitoring of response to physiotherapy. A case report of pneumonia Aymeric Le Neindre1, Pascal Selot1, Daniel Ferreiro,1, Maria Bonarek,1, Stépahen Henriot,1, Julie Rodriguez,1 1Physiotherapy, Hôpital Forcilles ‑ Fondation CognacqJ‑ay, Férolles‑Attilly, France Correspondence: Aymeric Le Neindre ‑ Annals of Intensive Care 2017, 7(Suppl 1):O79 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 (PaO2  =  59  mmHg and SaO2  =  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 SpO2 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 Chronic Critical Illness (CCI) syndrome is a new condition affecting an increasing number of patients, who survived an acute critical illness but have persistent severe organ dysfunction, requiring prolonged specialized care. CCI is a iatrogenic process, reflecting the efficacy of modern life support technologies(1), and encompasses multiple organ failure, need for prolonged mechanical ventilation (MV), organ support, and palsy due to polineuromyopathy. The transition from acute to CCI is gradual: definitions are based on duration of MV, with cut-offs of 7, 14 or 21 consecutive days of MV for ≥6 h/day. CCI patients may come from either medical or surgical ICU; their health status fluctuates between improvements and deteriorations implying recurrent transitions between different levels of care (1).The risk of death is reported to be as high as 50%. Despite a relatively young age (65  years on average), functional status of CCI patients discharged is seriously impaired, thus CCI patients require long-term rehabilitation. AIM: To estimate the frequency of CCI Syndrome in Careggi, a large academic, tertiary care hospital; to describe the clinical course of CCI patients through discharge, and their functional status at discharge. Patients and methods Administrative data on admission, transfer, death and discharge of all CCI patients, consecutively admitted in one of the 56 ICU beds at Careggi Hospital from January 1 to December 31, 2014, were collected. CCI was defined with the cut off of ≥21 days of ICU stay, representing the index event (IE) without contribution of previous or subsequent hospitalization in other hospitals. Reasons for admission were grouped into the 4 broad categories of medical causes, surgery, major trauma and cardio-respiratory arrest. Patients discharged were evaluated in daily living, cognitive status, and mobility using Barthel Index. Results We identified 123 subjects who developed CCI (71 males; age 61.7  ±  1.5  years, mean  ±  SEM); 36 of them came from an external ICU, 87 began their CCI course within Careggi hospital (60 from the Emergency Room, 27 from a regular ward). Average duration of the IE was 36.1  ±  2.1  days. These sample developed accumulative length of ICU stay of 4440  days, corresponding to a 22% ICU bed occupation over the theoretical total of 20,440. When days of subintensive care and regular ward were separately added, 5500  days of highly specialized care and 6266 days of total acute hospital stay were reached. Surgical patients had longer hospitalizations (p = 0.009).CCI patients confirmed to be highly erratic: a total of 302 transitions across different services were recorded in the 123 patients, with a maximum of 9 in 6 of them. Mean age was comparable between the 27 patients who died (22%) and the remaining 95 who were discharged alive (66.9  ±  2.5 vs. 60.3  ±  1.7  years; p  =  0.058).Fourteen subjects continued their ICU stay out of hospital. Only 6, whose age was lower (37.7  ±  3.7  years), were discharged home; half of the participants (n  =  68, 55.2%) were admitted to a residential rehabilitation facility. Younger subjects scored better in the domains of self care (p = 0.018) and cognitive status (p  =  0.008) but not in the domain of mobility, including walking ability: 45 patients required maximal assistance in performing activities of daily living and transfers, other 12 required medium/maximal assistance, with no statistical difference between DG group. Conclusion CCI is a relevant clinical condition that need to be assessed and possibly prevented, as it causes severe morbidity, long-term functional impairment and exceeding healthcare costs. 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 (µm2) 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. O82 Where should we place the stethoscope’s chestpiece to hear the noise of the primary bronchi? Frédéric Duprez1, Bastien Dupuis2, Grégory Cuvelier2, Thierry Bonus1, Sandra Ollieuz1, Sharam Machayeckhi1, Gregory Reychler3 1Icu, C.H. Epicura, Hornu, Belgium; 2Laboratoire de l’effort et du mouve‑ ment, Condorcet, Tournai, Belgium; 3Service de pneumologie, Cliniques Universitaires Saint Luc, Bruxelles, Belgium Correspondence: Frédéric Duprez ‑ Annals of Intensive Care 2017, 7(Suppl 1):O82 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: a) Lowered perpendicular (LP) of: • Bi‑Axillary Line (BAL) and the sternal carina (SC) • BAL’s and the position of right and left PB • Middle of the body‑sternum (BS) and the perpendicular middle of right and left PB. b) Hyoid bone and the SC The exclusion criteria were: BMI < 18.5 kg/m2 and BMI > 30 kg/m2, 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 cm2 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. O83 Study of efficacy on ICU acquired weakness of early standing with the assistance of a tilt table in critically ill patients Celine Sarfati1, Alex Moore1, Paula Mendialdua1, Emilie Rodet1, Catherine Pilorge2, Francois Stephan2, Saida Rezaiguia‑Delclaux 2 1Physiotherapy department, Surgical Center Marie Lannelongue, Le Plessis‑Robinson, France; 2Réanimation adulte, Surgical Center Marie Lan‑ nelongue, Le Plessis‑Robinson, France Correspondence: Celine Sarfati ‑ Annals of Intensive Care 2017, 7(Suppl 1):O83 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/ Fig. 31 50 X‑Rays (Men = 26, women = 24) have been analyzed. Normality test passed. LP lower perpendicular, BAL bi‑axillary line Fig. 32 See text for description norepinephrine >2 mg/h; continuous renal replacement; ECMO). Interventions for patients assigned to the standard physiotherapy group (Std) included sitting in armchair for at least 2 h per day. In addition to the standard physiotherapy mobilization protocol, patients assigned to the Tilt table physiotherapy group (Tilt) were positioned on a tilt table for 1 h per day. The primary outcome was the muscles strength evaluated using the Medical Research Council (MRC) score (range 0 = no muscle contraction, to 5 = normal strength), scoring 3 muscle groups in each limb, at ICU and hospital discharge, compared to MRC score evaluated in ICU before the implementation of the physiotherapy program. Secondary outcome variables included ICU and hospital length of stay. Results The flow chart of the study, conducted between October 2013 and October 2014, is presented in the Fig. 32. Both groups (Std vs Tilt) are comparable for age (63 ± 16 vs 60 ± 15, p = 0.37), gender (21 Females vs 26 F and 52 Males vs 46 M, p = 0.34) and the SAPSII (43  ±  13 vs 42  ±  12, p  =  0.65). 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. O84 Aerosol delivery using two nebulizers through high flow nasal cannula: a randomized cross‑over SPECT‑CT study Jonathan Dugernier1, Michel Hesse2, Thibaud Jumetz1, Emilie Bialais1, Jean Roeseler1, Virginie Depoortere2, Jean Bernard Michotte3, Xavier Wit‑ tebole1, Pierre‑François Laterre 1, François Jamar4, Gregory Reychler5 1Sevice des soins intensifs, Cliniques Universitaires Saint‑Luc, Bruxelles, Belgium; 2 Médecine nucléaire, Cliniques Universitaires Saint‑Luc, Bruxelles, Belgium; 3Respiratory, Haute École de Santé Vaud, Lausanne, Switzerland; 4Service de médecine nucléaire, Cliniques Universitaires Saint‑Luc, Woluwe‑Saint‑Lambert, Belgium; 5Service de pneumologie, Cliniques Universitaires Saint Luc, Bruxelles, Belgium Correspondence: Jonathan Dugernier ‑ Annals of Intensive Care 2017, 7(Suppl 1):O84 Introduction Patients with high flow nasal cannula may benefit from combined aerosol therapy. Clinical efficacy depends on pulmonary deposition which is related to the type of nebulizer. All new nebulizers or delivery methods require rigorous evaluation. The aim of this study was to compare lung deposition between two nebulizers (jet nebulizer vs vibrating-mesh nebulizer) through high flow nasal cannula in healthy subjects. Patients and methods Aerosol delivery of diethylenetriaminepentaacetic acid labelled with technetium-99m (99mTc-DTPA, 4  mCi/4  mL) to the lungs using a vibrating-mesh nebulizer (Aerogen Solo®, Aerogen Ltd., Galway, Ireland) and a constant-output jet nebulizer (Opti-Mist Plus Nebulizer®, ConvaTec, Bridgewater, NJ) through high flow nasal cannula (Optiflow®, Fisher & Paykel, New Zealand) was compared in 6 healthy subjects. Flow rate was set at 30 L/min through the heated humidified circuit. Pulmonary and extrapulmonary deposition were measured by single photon emission computed tomography combined with a low dose CT-scan (SPECT-CT) and by planar scintigraphy. Results Lung deposition was only 3.3  ±  1.3 and 1.2  ±  0.8% of the nominal dose with the vibrating-mesh nebulizer and the jet nebulizer, respectively (p  <  0.05). Dose lost in the high flow circuit, humidification chamber and nasal cannula was higher with the vibrating-mesh nebulizer as compared to the jet nebulizer (75.2 ± 8.2 vs 45.0 ± 6.2% of the nominal dose, p = 0.001). Expressed as percentage of emitted dose, lung deposition was similar with both nebulizers. Conclusion This study demonstrated that aerosol delivery through HFNC is poor in the specific conditions of the study despite the higher efficiency of the vibrating-mesh nebulizer as compared to the jet nebulizer. Placing the nebulizer on the HFNC circuit at 30 L/min induces high aerosol loss on the circuit and the oropharynx. Competing interests J. Dugernier: unrestricted grant by Aerogen Ltd. (Galway, Ireland). 1. Am J Respir Crit Care Med . 2007 ; 175 : 480 - 9 .


This is a preview of a remote PDF: https://link.springer.com/content/pdf/10.1186%2Fs13613-016-0224-7.pdf

Chtara Kamilia, Kais Regaieg, Najeh Baccouch, Hedi Chelly, Mabrouk Bahloul, Mounir Bouaziz, Ali Jendoubi, Ahmed Abbes, Houda Belhaouane, Oussama Nasri, Layla Jenzri, Salma Ghedira, Mohamed Houissa, Kamal Belkadi, Youness Harti, Afak Nsiri, Khalid Khaleq, Driss Hamoudi, Rachid Harrar, Camille Thieffry, Frédéric Wallet, Erika Parmentier-Decrucq, Raphaël Favory, Daniel Mathieu, Julien Poissy, Thomas Lafon, Philippe Vignon, Emmanuelle Begot, Alexandra Appert, Mathilde Hadj, Paul Claverie, Morgan Matt, Olivier Barraud, Bruno François, Amira Jamoussi, Amira Ben Jazia, Takoua Marhbène, Dhouha Lakhdhar, Jalila Ben Khelil, Mohamed Besbes, Julien Goutay, Caroline Blazejewski, Isabelle Joly-Durand, Isabelle Pirlet, Marie Pierre Weillaert, Sebastien Beague, Soufi Aziz, Reda Hafiane, Khalid Hattabi, Mohamed Aziz Bouhouri, Driss Hammoudi, Abdelaziz Fadil, Rachid Al Harrar, Khalid Zerouali, Fatma Kaaniche Medhioub, Rania Allela, Najla Ben Algia, Samar Cherif, Mohamed Taoufik Slaoui, Souhail Boubia, Y. Hafiani, A. Khaoudi, R. Cherkab, W. Elallam, C. Elkettani, L. Barrou, M. Ridaii, Rihi El Mehdi, Caroline Schimpf, Assaf Mizrahi, Benoît Pilmis, Alban Le Monnier, Kelly Tiercelet, Mélanie Cherin, Cédric Bruel, Francois Philippart, Sébastien Bailly, Jc Lucet, Alain Lepape, François L’hériteau, Martine Aupée, Caroline Bervas, Sandrine Boussat, Anne Berger-Carbonne, Anaïs Machut, Anne Savey, Jean-François Timsit, Keyvan Razazi, Jérémy Rosman, Nicolas de Prost, Guillaume Carteaux, Chloe Jansen, Jean Winoc Decousser, Christian Brun-Buisson, Armand Mekontso Dessap, Aymen M’rad, Zouhour Ouali, Manel Barghouth, Achille Kouatchet, Rafael Mahieu, Emmanuel Weiss, David Schnell, Jean-Ralph Zahar, Margaux Artiguenave, Paktoris-Papine Sophie, Florence Espinasse, Faten El Sayed, Aurélien Dinh, Cyril Charron, Guillaume Geri, Antoine Vieillard-Baron, Xavier Repessé, Hatem Kallel, Claire Mayence, Stéphanie Houcke, Pascal Guegueniat, Didier Hommel, Kaouther Dhifaoui, Zied Hajjej, Amira Fatnassi, Walid Sellami, Iheb Labbene, Mustapha Ferjani, Fahmi Dachraoui, Sabrine Nakkaa, Abdelwaheb M’ghirbi, Ali Adhieb, Dhouha Ben Braiek, Kmar Hraiech, Ali Ousji, Islem Ouanes, Hammouda Zaineb, Saousen Ben Abdallah, Lamia Ouanes-Besbes, Fekri Abroug, Simon Klein, Mattéo Miquet, Jean-Marc Thouret, Vincent Peigne, Jean-Louis Daban, Mathieu Boutonnet, Bernard Lenoir, Takoua Merhbene, Celine Derreumaux, Thierry Seguin, Jean-Marie Conil, Charlotte Kelway, Valery Blasco, Cyril Nafati, Karim Harti, Laurent Reydellet, Jacques Albanese, Narjess Ben Aicha, Khaoula Meddeb, Ahmed Khedher, Jihene Ayachi, Nesrine Fraj, Nesrine Sma, Imed Chouchene, Mohamed Boussarsar, Soumaya Ben Yedder, Walid Samoud, Bousselmi Radhouene, Bousselmi Mariem, Asma Ammar, Asma Ben Cheikh, Hend Ben Lakhal, Messaouda Khelfa, Yamina Hamdaoui, Nabiha Bouafia, Timothée Trampont, Thomas Daix, Vincent Legarçon, Henri Hani Karam, Nicolas Pichon, Fatma Essafi, Nasreddine Foudhaili, Hafedh Thabet, Youssef Blel, Nozha Brahmi, Hanane Ezzouine, Mahmoud Kerrous, Saad El Haoui, Soufiane Ahdil, Abdellatif Benslama, Khalid Abidi, Tarek Dendane, Ssouni Oussama, Jihane Belayachi, Naoufal Madani, Redouane Abouqal, Amine Ali Zeggwagh, Hatem Ghadhoune, Anis Chaari, Guissouma Jihene, Hend Allouche, Insaf Trabelsi, Habib Brahmi, Mohamed Samet, Hatem El Ghord, Ben Sik Ali Habiba, Nouira Hajer, Najla Tilouch, Sondes Yaakoubi, Oussama Jaoued, Rim Gharbi, Mohamed Fekih Hassen, Souheil Elatrous, Julien Arcizet, Bertrand Leroy, Caroline Abdulmalack, Catherine Renzullo, Maël Hamet, Jean-Marc Doise, Jérôme Coutet, Chaigar Mohammed Cheikh, Zakaria Quechar, Magalie Joris, Dimitri Titeca Beauport, Loay Kontar, Delphine Lebon, Bérengère Gruson, Michel Slama, Jean-Pierre Marolleau, Julien Maizel, Julie Gorham, Lieveke Ameye, Thierry Berghmans, Marianne Paesmans, Jean-Paul Sculier, Anne-Pascale Meert, Max Guillot, Marie-Pierre Ledoux, Thierry Braun, Quentin Maestraggi, Baptiste Michard, Vincent Castelain, Raoul Herbrecht, Francis Schneider, Severine Couffin, David Lobo, Nicolas Mongardon, Gilles Dhonneur, Roman Mounier, Pierrick Le Borgne, Sophie Couraud, Jean-Etienne Herbrecht, Alexandra Boivin, François Lefebvre, Pascal Bilbault, Setti-Aouicha Zelmat, Djamila-Djahida Batouche, Fatima Mazour, Belkacem Chaffi, Nadia Benatta, Ali Habiba Sik, I. Talik, Maxime Perrier, Eliane Gouteix, Claude Koubi, Annabelle Escavy, Victoria Guilbaut, Jean-Philippe Fosse, Rahma Ben Jazia, Ahmed Abdelghani, Pierre-Julien Cungi, Julien Bordes, Cédric Nguyen, Candice Pierrou, Maximilien Cruc, Alain Benois, Frédéric Duprez, Thierry Bonus, Grégory Cuvelier, Sandra Ollieuz, Sharam Machayekhi, Frédéric Paciorkowski, Gregory Reychler, Remi Coudroy, Arnaud W. Thille, Xavier Drouot, Véronique Diaz, Jean-Claude Meurice, René Robert, Olfa Turki, Hmida Chokri Ben, Mona Assefi, Romain Deransy, Hélène Brisson, Antoine Monsel, Filomena Conti, Olivier Scatton, Olivier Langeron, Hassen Ben Ghezala, Salah Snouda, Chiekh Imen Ben, Moez Kaddour, Anwar Armel. Proceedings of Réanimation 2017, the French Intensive Care Society International Congress, Annals of Intensive Care, 2017, 7, DOI: 10.1186/s13613-016-0224-7