Search

Mechanisms and consequences of late-onset hyperlactatemia after cardiac surgery remain unclear. The aim of this study was to identify risk factors and outcomes of late hyperlactatemia, defined as a lactate value ≥3 mmol/L developing in the intensive care unit (ICU) after not being elevated on admission after cardiac surgery with cardiopulmonary bypass.A retrospective analysis of prospectively collected data.A single-center University Hospital.Patients who underwent elective cardiac surgery with cardiopulmonary bypass in 2019 and who had normal lactatemia at ICU admission.None.Patients were divided in 2 groups according to their lactatemia on postoperative day 1 (14.0 ± 3.0 hours after ICU admission): normal lactatemia (control group) and late hyperlactatemia (HL group). Risk factors for late hyperlactatemia were identified using a multivariate analysis, and postoperative outcomes were compared using a composite criterion of severe outcomes. Of the 432 included patients, 37 (8.5%) presented with late hyperlactatemia. Risk factors independently associated with hyperlactatemia were afternoon surgery (odds ratio [OR] 4.24, 95% CI 2.00-9.35), a bleeding300 mL within the 6 hours after surgery (H6) (OR 3.77, 95% CI 1.71-8.30), and H6 fluid loading250 mL (OR 2.64, 95% CI 1.22-5.55). Patients with hyperlactatemia presented more frequently with major postoperative complications, including acute kidney injury, and received more frequent red-cell transfusion.The strongest risk factors associated with late-onset hyperlactatemia in the authors' population were afternoon surgery and H6 bleeding300 mL. Poor postoperative outcomes were more frequent in patients with late hyperlactatemia, even in the absence of early hyperlactatemia or severe obvious condition.

Le baclofène, initialement utilisé comme traitement antispastique, est depuis plusieurs années prescrit chez les patients alcoolodépendants pour favoriser le sevrage alcoolique. La France présente la spécificité d’avoir autorisé la mise sur le marché du baclofène dans cette indication depuis 2018. Les patients alcoolodépendants présentent un risque accru de tentative de suicide, notamment par intoxication volontaire. L’incidence des intoxications volontaires au baclofène s’est majoré sur la dernière décennie. Comparativement aux autres intoxications, celle-ci présente un recours fréquent à la ventilation mécanique (50% des cas) et une léthalité significative (plus de 2% des cas). Le toxidrome de l’intoxication aiguë au baclofène se résume principalement à une dépression du système nerveux central, pouvant aller de la sédation jusqu’au coma, et nécessitant une surveillance continue pour des doses suspectées ingérées supérieures à 200 mg. Le dosage plasmatique du baclofène permet de confirmer le diagnostic et de suivre l’évolution de l’élimination de la molécule. La principale voie d’élimination est rénale, motivant une surveillance accrue de la fonction rénale dans ce contexte. L’épuration extrarénale pourra se discuter dans les intoxications les plus sévères associées à une insuffisance rénale aiguë. Le reste de la prise en charge reste symptomatique, sans antidote existant. Le syndrome de sevrage au baclofène se rencontre lors d’un arrêt brutal du traitement, et peut être composé des symptômes suivants : spasticité, hyperthermie, agitation, rhabdomyolyse. Il peut compliquer l’hospitalisation de patients traités au long cours admis pour une intoxication aiguë. Son traitement nécessite une réintroduction précautionneuse du baclofène.

Case Presentation A 78-year-old woman was admitted to the ED with a 10-day history of diarrhea and recent onset of dry cough, fever, and asthenia. She had a medical history of obesity (BMI 32) and arterial hypertension treated with irbesartan. In the context of a large-scale lockdown in France during the COVID-19 pandemic, she only had physical contact with her husband, who did not report any symptoms. She required mechanical ventilation because of severe hypoxemia within 1 hour after admission to the ED.

Aims We aimed at assessing the effectiveness of renal replacement therapy in patients severely self-poisoned with baclofen and with normal kidney function. Methods A population pharmacokinetic model was built using analytical data extracted from 26 baclofen poisoning cases reported to a French Poison Centre: 8 patients underwent renal replacement therapy (RRT), 18 did not. In the RRT group, 2 patients suffered from kidney failure. Mechanical ventilation was required for 20 patients with normal kidney function; 15 were not treated by RRT and 5 were. Pharmacokinetic profiles of baclofen were measured in 28 patients and further modelled by a non-parametric approach (PMetrics®). The total data set was divided into a building data set (26 patients, 57 observations) and a validation set (2 external patients, 6 observations). Then, the estimated elimination half-life of baclofen and the duration of intubation were compared in patients with or without RRT using Wilcoxon–Mann–Whitney test. Results A model using three parameters plus a lag time and bioavailability was necessary to determine the pharmacokinetics of baclofen. Estimated elimination half-life in the ‘RRT’ group and the ‘no RRT’ group were respectively 3.1 [2.2−4.8] h (n = 6 patients) and 3.4 [1.4−5.5] h (n = 19 patients, p = 0.53). The median duration of intubation was not significantly different between groups (72 [48−72] h and 72 [24−96] h, respectively; p = 0.38). Conclusion Renal replacement therapy did not appear to significantly increase baclofen clearance in patients without kidney failure.

Purpose To describe acute respiratory distress syndrome (ARDS) severity, ventilation management, and the outcomes of ICU patients with laboratory-confirmed COVID-19 and to determine risk factors of 90-day mortality post-ICU admission. Methods COVID-ICU is a multi-center, prospective cohort study conducted in 138 hospitals in France, Belgium, and Switzerland. Demographic, clinical, respiratory support, adjunctive interventions, ICU length-of-stay, and survival data were collected. Results From February 25 to May 4, 2020, 4643 patients (median [IQR] age 63 [54–71] years and SAPS II 37 [28–50]) were admitted in ICU, with day-90 post-ICU admission status available for 4244. On ICU admission, standard oxygen therapy, high-flow oxygen, and non-invasive ventilation were applied to 29%, 19%, and 6% patients, respectively. 2635 (63%) patients were intubated during the first 24 h whereas overall 3376 (80%) received invasive mechanical ventilation (MV) at one point during their ICU stay. Median (IQR) positive end-expiratory and plateau pressures were 12 (10–14) cmH2O, and 24 (21–27) cmH2O, respectively. The mechanical power transmitted by the MV to the lung was 26.5 (18.6–34.9) J/min. Paralyzing agents and prone position were applied to 88% and 70% of patients intubated at Day-1, respectively. Pulmonary embolism and ventilator-associated pneumonia were diagnosed in 207 (9%) and 1209 (58%) of these patients. On day 90, 1298/4244 (31%) patients had died. Among patients who received invasive or non-invasive ventilation on the day of ICU admission, day-90 mortality increased with the severity of ARDS at ICU admission (30%, 34%, and 50% for mild, moderate, and severe ARDS, respectively) and decreased from 42 to 25% over the study period. Early independent predictors of 90-day mortality were older age, immunosuppression, severe obesity, diabetes, higher renal and cardiovascular SOFA score components, lower PaO2/FiO2 ratio and a shorter time between first symptoms and ICU admission. Conclusion Among more than 4000 critically ill patients with COVID-19 admitted to our ICUs, 90-day mortality was 31% and decreased from 42 to 25% over the study period. Mortality was higher in older, diabetic, obese and severe ARDS patients. Electronic supplementary material The online version of this article (10.1007/s00134-020-06294-x) contains supplementary material, which is available to authorized users.

Background Since the 2000s, opioid-free anesthesia (OFA) protocols have been spreading worldwide in anesthesia daily practice. These protocols avoid using opioid drugs during anesthesia to prevent short- and long-term opioid side effects while ensuring adequate analgesic control and optimizing postoperative recovery. Proofs of the effect of OFA protocol on optimizing postoperative recovery are still scarce. The study aims to compare the effects of an OFA protocol versus standard anesthesia protocol on the early quality of postoperative recovery (QoR) from major surgeries. Methods The SOFA trial is a prospective, randomized, parallel, single-blind, monocentric study. Patients (n = 140) scheduled for major plastic, visceral, urologic, gynecologic, or ear, nose, and throat (ENT) surgeries will be allocated to one of the two groups. The study group (OFA group) will receive a combination of clonidine, magnesium sulfate, ketamine, and lidocaine. The control group will receive a standard anesthesia protocol based on opioid use. Both groups will receive others standard practices for general anesthesia and perioperative care. The primary outcome measure is the QoR-15 value assessed at 24 h after surgery. Postoperative data such as pain intensity, the incidence of postoperative complication, and opioid consumption will be recorded. We will also collect adverse events that may be related to the anesthetic protocol. Three months after surgery, the incidence of chronic pain and the quality of life will be evaluated by phone interview. Discussion This will be the first study powered to evaluate the effect of OFA versus a standard anesthesia protocol using opioids on global postoperative recovery after a wide range of major surgeries. The SOFA trial will also provide findings concerning the OFA impact on chronic pain incidence and long-term patient quality of life. Trial registration ClinicalTrials.gov NCT04797312. Registered on 15 March 2021

When anesthesia checklists and preparations are performed urgently, omissions may occur and be deleterious to the patient. The aim of this study was to evaluate in simulation the interest of a cognitive aid to effectively prepare an anesthetic room for an emergency. In a prospective single-center simulation-based study, 32 anesthesia residents had to prepare an anesthetic room in an emergency scenario, without cognitive aid in the first phase. Three months later (phase 2), they were randomly assigned to receive a cognitive aid (aid group) or no additional aid (control) and were involved in the same scenario. The primary outcome was the validation rate of each essential item in the first 5 min in phase 2. Eight items were significantly more frequently completed in the first 5 min in the aid group in phase 2 (vs. phase 1), compared with two only in the control group. However, there were no significant differences in the overall number of completed items between the two groups, as both groups completed significantly more items in phase 2, either in the first 5 min (19 (14–23) vs. 13 (9–15) in phase 1 for all residents, p < 0.001) or without time limit. Preparation times were reduced in phase 2 in both groups. In conclusion, the use of a cognitive aid allowed anesthesia residents to complete some safety items of a simulated urgent anesthesia preparation more frequently. In addition, despite daily clinical experience, a single simulation session improved anesthesia preparation and reduced the preparation time with or without cognitive aid.

In causal studies, the near-violation of the positivity may occur by chance, because of sample-to-sample fluctuation despite the theoretical veracity of the positivity assumption in the population. It may mostly happen when the exposure prevalence is low or when the sample size is small. We aimed to compare the robustness of g-computation (GC), inverse probability weighting (IPW), truncated IPW, targeted maximum likelihood estimation (TMLE), and truncated TMLE in this situation, using simulations and one real application. We also tested different extrapolation situations for the sub-group with a positivity violation. The results illustrated that the near-violation of the positivity impacted all methods. We demonstrated the robustness of GC and TMLE-based methods. Truncation helped in limiting the bias in near-violation situations, but at the cost of bias in normal conditions. The application illustrated the variability of the results between the methods and the importance of choosing the most appropriate one. In conclusion, compared to propensity score-based methods, methods based on outcome regression should be preferred when suspecting near-violation of the positivity assumption.

International audience; Background: Ventilator-associated pneumonia (VAP) is the most frequent health care-associated infection in severely ill patients, and aspiration of contaminated oropharyngeal content around the cuff of the tracheal tube is the main route of contamination.Research question: Is continuous regulation of tracheal cuff pressure using a pneumatic device superior to manual assessment three times daily using a portable manometer (routine care) in preventing VAP in patients with severe trauma?Study design and methods: In this open-label, randomized controlled superiority trial conducted in 13 French ICUs, adults (age ≥ 18 years) with severe trauma (Injury Severity Score > 15) and requiring invasive mechanical ventilation for ≥ 48 h were enrolled. Patients were randomly assigned (1:1) via a secure Web-based random number generator in permuted blocks of variable sizes to one of two groups according to the method of tracheal cuff pressure control. The primary outcome was the proportion of patients developing VAP within 28 days following the tracheal intubation, as determined by two assessors masked to group assignment, in the modified intention-to-treat population. This study is closed to new participants.Results: A total of 434 patients were recruited between July 31, 2015, and February 15, 2018, of whom 216 were assigned to the intervention group and 218 to the control group. Seventy-three patients (33.8%) developed at least one episode of VAP within 28 days following the tracheal intubation in the intervention group compared with 64 patients (29.4%) in the control group (adjusted subdistribution hazard ratio, 0.96; 95% CI, 0.76-1.20; P = .71). No serious adverse events related to the use of the pneumatic device were noted.Interpretation: Continuous regulation of cuff pressure of the tracheal tube using a pneumatic device was not superior to routine care in preventing VAP in patients with severe trauma.Clinical trial registration: ClinicalTrials.gov; No.: NCT02534974; URL: www.clinicaltrials.gov.

We aim to assess physicians’ level of resilience and define factors that improve or decrease the resilience level during the COVID-19 pandemic. Physicians from hospitals located in areas with different COVID-19 caseload levels, were invited to participate in a national e-survey between April and May 2020. Study participants were mainly emergency physicians, and anaesthesiologists, infectious disease consultants, and intensive care. The survey assessed participant’s characteristics, factors potentially associated with resilience, and resilience using the Connor-Davidson Resilience Scale (RISC-25), with higher scores indicative of greater resilience. Factors associated with the resilience score were assessed using a multivariable linear regression. Of 451 responding physicians involved in the care of COVID-19 patients, 442 were included (98%). Age was 36.1 ± 10.3 years and 51.8% were male; 63% worked in the emergency department (n = 282), 10.4% in anesthesiology (n = 46), 9.9% in infectious disease department (n = 44), 4.8% in intensive care unit (n = 21) or other specialties (n = 49). The median RISC-25 score was at 69 (IQR 62–75). Factors associated with higher RISC scores were anesthesia as a specialty, parenthood, no previous history of anxiety or depression and nor increased anxiety. To conclude, this study is the first to characterize levels of resilience among physicians involved in COVID-19 unit. Our data points to certain protective characteristics and some detrimental factors, such as anxiety or depression, that could be amenable to remediating or preventing strategies to promote resilience and support caregivers in a pandemic.

Sağır, Özlem (Balikesir Author), Introduction More than 300 million surgical procedures are performed each year. Acute kidney injury (AKI) is a common complication after major surgery and is associated with adverse short-term and long-term outcomes. However, there is a large variation in the incidence of reported AKI rates. The establishment of an accurate epidemiology of surgery-associated AKI is important for healthcare policy, quality initiatives, clinical trials, as well as for improving guidelines. The objective of the Epidemiology of Surgery-associated Acute Kidney Injury (EPIS-AKI) trial is to prospectively evaluate the epidemiology of AKI after major surgery using the latest Kidney Disease: Improving Global Outcomes (KDIGO) consensus definition of AKI. Methods and analysis EPIS-AKI is an international prospective, observational, multicentre cohort study including 10 000 patients undergoing major surgery who are subsequently admitted to the ICU or a similar high dependency unit. The primary endpoint is the incidence of AKI within 72 hours after surgery according to the KDIGO criteria. Secondary endpoints include use of renal replacement therapy (RRT), mortality during ICU and hospital stay, length of ICU and hospital stay and major adverse kidney events (combined endpoint consisting of persistent renal dysfunction, RRT and mortality) at day 90. Further, we will evaluate preoperative and intraoperative risk factors affecting the incidence of postoperative AKI. In an add-on analysis, we will assess urinary biomarkers for early detection of AKI. Ethics and dissemination EPIS-AKI has been approved by the leading Ethics Committee of the Medical Council North Rhine-Westphalia, of the Westphalian Wilhelms-University Münster and the corresponding Ethics Committee at each participating site. Results will be disseminated widely and published in peer-reviewed journals, presented at conferences and used to design further AKI-related trials. Trial registration number NCT04165369. ©

The QoR-15 scale is a validated tool to assess the quality of postoperative recovery (QoR). Our objective was to assess the association between the early QoR-15 values and the occurrence of one-month postoperative complications.We used the data from a prospective single-centre cohort study conducted in the Angers University Hospital from July 2019 to February 2020 that validated the French version of the QoR-15 (FQoR-15).Preoperative room, ward, and home.363 French-speaking adult patients, undergoing elective surgery, were enrolled (217 (59,8%) men, median age 60 (range 44 to 71) years old), including 139 (38.3%) ambulatory surgeries.Patients completed the QoR-15 scale the day before, at 24 and 48 h after surgery.Postoperative complications were recorded according to the PostOperative Morbidity Survey (POMS) classification till 30 days after surgery. The QoR was classified as excellent (QoR-15 135), good (122 ≤ QoR-15 ≤ 135), moderate (90 ≤ QoR-15 ≤ 121) or poor (QoR-15 90). Days alive and out of hospital up to 30 days after surgery was also recorded.According to the POMS classification, 176 (48.5%) patients had at least one complication up to 30 days after surgery. Among the 69 (19.0%) patients with a poor recovery at H24, 58 (84.1%) had at least one complication up to 30 days after surgery compared to 10 (23.8%) among the 42 (11.6%) in the excellent recovery group (p 0.0001). The QoR-15 score at H24 allowed suitable discrimination of the occurrence of at least one complication up to 30 days after surgery (AUC 0.732 (95% CI 0.680 to 0.784)).The early QoR-15 scale after surgery is moderately associated with the occurrence of postoperative complications up to 30 days after elective surgeries (i.e. it has predictive validity).

International audience; Barbiturates are proposed as a second/third line treatment for intracranial hypertension in traumatic brain injury (TBI) patients, but the literature remains uncertain regarding their benefit/risk balance. We aimed to evaluate the impact of barbiturates therapy in TBI patients with early intracranial hypertension on the intensive care unit (ICU) survival, the occurrence of ventilator-associated pneumonia (VAP), and the patient’s functional status at three months. We used the French AtlanREA prospective cohort of trauma patients. Using a propensity score-based methodology (inverse probability of treatment weighting), we compared patients having received barbiturates within the first 24 hours of admission (barbiturates group) and those who did not (control group). We used cause-specific Cox models for ICU survival and risk of VAP, and logistic regression for the 3-month Glasgow Outcome Scale (GOS) evaluation. Among the 1396 patients with severe trauma, 383 had intracranial hypertension on admission and were analyzed. Among them, 96 (25.1%) received barbiturates. The early use of barbiturates was significantly associated with increased ICU mortality (HR = 1.85, 95%CI 1.03–3.33). However, barbiturates treatment was not significantly associated with VAP (HR = 1.02, 95%CI 0.75–1.41) or 3-month GOS (OR = 1.67, 95%CI 0.84–3.33). Regarding the absence of relevant clinical trials, our results suggest that each early prescription of barbiturates requires a careful assessment of the benefit/risk ratio.

Controlling for confounding bias is crucial in causal inference. Distinct methods are currently employed to mitigate the effects of confounding bias. Each requires the introduction of a set of covariates, which remains difficult to choose, especially regarding the different methods. We conduct a simulation study to compare the relative performance results obtained by using four different sets of covariates (those causing the outcome, those causing the treatment allocation, those causing both the outcome and the treatment allocation, and all the covariates) and four methods: g-computation, inverse probability of treatment weighting, full matching and targeted maximum likelihood estimator. Our simulations are in the context of a binary treatment, a binary outcome and baseline confounders. The simulations suggest that considering all the covariates causing the outcome led to the lowest bias and variance, particularly for g-computation. The consideration of all the covariates did not decrease the bias but significantly reduced the power. We apply these methods to two real-world examples that have clinical relevance, thereby illustrating the real-world importance of using these methods. We propose an R package RISCA to encourage the use of g-computation in causal inference.

Various regional anesthesia (RA) techniques were shown to reduce pain after lung surgery, but controversies remain regarding the best technique to use to improve recovery. In this observational prospective study, the aim was to assess the efficacy of an RA strategy depending on the surgical approach. Patients who underwent lung surgery were included if an RA was planned following our unit procedure (erector spinae plane block (ESP) for video-assisted thoracic surgery (VATS) and thoracic epidural analgesia (TEA) or intrathecal analgesia (IA) for thoracotomy). Patients were compared according to the RA used. In total, 116 patients were included, 70 (60%), 32 (28%), 14 (12%) in the ESP, TEA and IA groups, respectively. Between Day 1 and Day 3, median NRS values were ≤4 at rest, and

Introduction Tandis que les approches d’apprentissage automatique (c.-a-d., « machine learning ») sont de plus en plus utilisees en prediction, leurs applications en inference causale sont plus recentes. Nous proposons une approche combinant la G-computation et l’apprentissage automatique pour estimer l’effet causal d’une exposition binaire sur un evenement binaire. Methodes Nous avons evalue et compare, via une etude de simulation, les performances de regressions logistiques penalisees, d’un reseau de neurones, d’une machine a vecteurs de support, d’une classification renforcee, d’arbres de regression et d’un « super learner ». Nous avons etudie six scenarios differents faisant varier les tailles d’echantillon et les relations entre les covariables, l’exposition binaire et l’evenement binaire. Resultats L’utilisation du super learner pour specifier le modele de nuisance de la G-computation surpassait les autres approches en termes de biais ou de variance lors de l’estimation de l’effet causal, particulierement pour les tailles d’echantillon moderees. L’utilisation de la machine a support de vecteurs permettait egalement d’obtenir des estimations causales performantes, bien que le biais moyen fut legerement plus eleve qu’avec le super learner. Conclusion L’utilisation d’approches d’apprentissage automatique peut etre pertinente en inference causale. Contrairement a une idee recue, ceci est egalement vrai pour des tailles d’echantillon constituees d’une centaine d’individus comme dans la majorite des etudes medicales. La G-computation avec le super learner est disponible dans le package R RISCA.

Aims Baclofen has gained interest in treating alcohol use disorder patients. These patients are at risk of voluntary drug intoxications. Understanding the context of baclofen's widespread use for alcohol use disorder, a recent evaluation of the epidemiology, morbidity and mortality of baclofen self-poisoning episodes is necessary. Methods All baclofen self-poisoning reported to the Western France Poison Control Centre between January 2008 and December 2015 were included for analysis in this retrospective study. Baclofen poisoning cases were compared to the 31,859 non-baclofen voluntary poisoning cases declared over the same period. Results Number of baclofen poisoning grew from 8 cases in 2008 to 91 cases in 2015, attributable to patients purposely treated for alcohol use disorder. Overall, 190 cases were considered: 111 patients (59%) had Glasgow coma score (GCS) ≤12 at hospital admission and 77 had GCS > 12 (41%). Eighty patients required mechanical ventilation (42.6%). Neurological severity (GCS ≤ 12) was associated with seizures (n = 24, 22% vs. n = 2, 3%, P < 0.001). Two patients were deceased at first medical evaluation and three died in the hospital (mortality rate 2.6%). Non-baclofen cases had lower rate of endotracheal intubation (n = 1833, 6%, P < 0.001) and mortality rate (n = 299, 0.1%, P = 0.02). Conclusions Baclofen self-poisoning cases has risen in recent years due to baclofen intake for alcohol use disorder therapies. The risk-benefit ratio may be evaluated before baclofen is safely prescribed.

International audience

Reintubation after weaning from mechanical ventilation is relatively common and is associated with poor outcomes. Different methods to decrease the reintubation rate post extubation, including noninvasive ventilation, and more recently high-flow oxygen (HFO) therapy, have been proposed. In this study, we aimed to assess the safety of introducing HFO in the post-extubation care of intensive care unit (ICU) patients. We conducted a single-center cohort study of extubated adult patients hospitalized in a surgical ICU and previously mechanically ventilated for > 1 day. Our study consisted of two phases: Phase 1 (before the introduction of HFO from April 2015 to April 2016) and Phase P2 (after the introduction of HFO from April 2017 to April 2018). The primary endpoint was the reintubation rate within 48 hours of extubation. In total, 290 patients (median age 65 years [50–74]; 190 men [65.5%]) were included in the analysis (181 and 109 in Phases 1 and 2, respectively). The results of the post-extubation use of noninvasive methods (noninvasive ventilation and/or HFO) were not significantly different between the two phases (41 [22.7%] versus 29 [26.6%] patients; p = 0.480), however these methods were implemented earlier in Phase 2 (0 versus 4 hours; p = 0.009) and HFO was used significantly more often than noninvasive ventilation (24 [22.0%] versus 25 [13.8%] patients; p = 0.039). The need for reintubation within 48 hours post extubation was significantly lower in Phase 2 (4 [3.7%] versus 20 [11.0%] patients; p = 0.028) but was not significantly different at 7 days post extubation (10 [9.2%] versus 30 [16.6%] patients; p = 0.082). The earlier implementation of noninvasive methods and the increased use of HFO beginning in Phase 2 were safe and effective based on the reintubation rates within the first 48 hours post extubation and after 7 days.

International audience; In this paper, we present work conducted in order to explain the results of a commercial software used for real-time decision support for the flow management of a combined and sanitary wastewater system. This tool is deployed in many major cities and is used on a daily basis. We apply decision trees to build rules for classifying and interpreting the solutions of the optimization model. Our main goal is to build a classifier that would help a user to understand why a proposed solution is good and why other solutions are worse. We demonstrate the feasibility of the approach to our industrial application by generating a large dataset of feasible solutions and classifying them as satisfactory or not based on whether the objective function is a certain percentage higher than the optimal objective. We evaluate the performance of the learned classifier on unseen examples. Our results show that our approach is very promising according to reactions from analysts and potential users.

Objectif Les intoxications au baclofene provoquent essentiellement des troubles neurologiques, coma, convulsions, depression respiratoire et complications cardiovasculaires. Actuellement, le traitement de ces intoxications est symptomatique avec une prise en charge en reanimation et une ventilation mecanique dans les tableaux les plus severes. L’epuration extrarenale est indiquee chez les patients insuffisants renaux dans les cas d’intoxications graves. Son efficacite reste debattue chez les patients avec une fonction renale preservee. Cette etude a pour objectif de determiner l’efficacite de l’epuration extrarenale dans la prise en charge des intoxications au baclofene chez les patients sans dysfonction renale. Methodes Nous avons etudie les cas signales au Centre Antipoison d’Angers entre 2008 et 2015. Nous avons identifie tous les cas par baclofene chez des patients sans dysfonction renale traites par hemodialyse ou non. Tous les cas avec une dose supposee ingeree connue et au moins un dosage plasmatique de baclofene ont ete inclus. Les dosages ont ete faits par des methodes LC-MS/MS ou HPLC-DAD selon les laboratoires. Tous les profils de pharmacocinetique ont ete analyses a l’aide de l’approche non parametrique avec le logiciel R-package (Pmetrics version 1.5.0) [1] . L’analyse a ete basee sur un modele mono-compartimental. Les demi-vies de baclofene ont ensuite ete comparees entre les patients qui ont beneficie ou non d’une epuration extra renale. La duree de l’intubation a egalement ete rapportee. Resultats Les profils pharmacocinetiques ont ete analyses pour 24 patients. 7 ont beneficie d’une hemodialyse pendant leur sejour en reanimation et 18 n’ont pas ete hemodialyses. Les baclofenemies etaient en moyenne de 6,5 ± 3,6 mg/L contre 2,4 ± 1,9 mg/L, respectivement. Les demi-vies d’elimination chez ces patients etaient respectivement de 3,93 ± 2,45 h et de 4,09 ± 4,02 h. La difference n’etait pas significative ( p = 0,520). Une ventilation mecanique a ete necessaire pour 20 patients. Parmi eux, 15 n’ont pas ete hemodialyses et 5 l’ont ete. La duree moyenne d’intubation n’a pas ete significativement differente pour chaque groupe (93,9 ± 56,5 et 80 ± 44,6 h respectivement, p = 0,530). Conclusion L’epuration extra renale ne semble pas indiquee dans la prise en charge des intoxications au baclofene chez les patients sans dysfonction renale en raison de l’absence de benefice biologique et clinique, et du fait des complications qui sont associees a cette methode d’epuration.