Search

Abstract Backgroud Modelling discrete-time cause-specific hazards in the presence of competing events and non-proportional hazards is a challenging task in many domains. Survival analysis in longitudinal cohorts often requires such models; notably when the data is gathered at discrete points in time and the predicted events display complex dynamics. Current models often rely on strong assumptions of proportional hazards, that is rarely verified in practice; or do not handle sequential data in a meaningful way. This study proposes a Transformer architecture for the prediction of cause-specific hazards in discrete-time competing risks. Contrary to Multilayer perceptrons that were already used for this task (DeepHit), the Transformer architecture is especially suited for handling complex relationships in sequential data, having displayed state-of-the-art performance in numerous tasks with few underlying assumptions on the task at hand. Results Using synthetic datasets of 2000–50,000 patients, we showed that our Transformer model surpassed the CoxPH, PyDTS, and DeepHit models for the prediction of cause-specific hazard, especially when the proportional assumption did not hold. The error along simulated time outlined the ability of our model to anticipate the evolution of cause-specific hazards at later time steps where few events are observed. It was also superior to current models for prediction of dementia and other psychiatric conditions in the English longitudinal study of ageing cohort using the integrated brier score and the time-dependent concordance index. We also displayed the explainability of our model’s prediction using the integrated gradients method. Conclusions Our model provided state-of-the-art prediction of cause-specific hazards, without adopting prior parametric assumptions on the hazard rates. It outperformed other models in non-proportional hazards settings for both the synthetic dataset and the longitudinal cohort study. We also observed that basic models such as CoxPH were more suited to extremely simple settings than deep learning models. Our model is therefore especially suited for survival analysis on longitudinal cohorts with complex dynamics of the covariate-to-outcome relationship, which are common in clinical practice. The integrated gradients provided the importance scores of input variables, which indicated variables guiding the model in its prediction. This model is ready to be utilized for time-to-event prediction in longitudinal cohorts.

IntroductionUreteropelvic junction obstruction (UPJO) syndrome is one of the most common causes of neonatal hydronephrosis. Management varies from simple monitoring to surgical intervention, with indications differing between institutions. A consensus of 8 societies recently described a new Urinary Tract Dilation (UTD) classification which aims to standardize ultrasound description of hydronephrosis, but which is also supposed to have predictive value in children with hydronephrosis. Our aim was to compare, in a monocentric prospective cohort of children with UPJO, the ability of UTD to predict the occurrence of a clinically significant event within the first year of life, as compared to anteroposterior diameter of the renal pelvis (APD).Study designWe used a preexisting cohort of children followed in a prospective study on UPJO. A pediatric radiologist, blinded to the children's outcome, classified the last antenatal ultrasound and postnatal ultrasound according to the UTD-A and UTD-P classification. He also confirmed the APD-A and APD-P measures. We defined a clinically significant event as being: increased pelvic dilation (>5 mm) and/or the presence of a febrile urinary tract infection (fUTI) and/or impaired renal function on initial nuclear scan (5 mm in dilation, 6 presented an episode of fUTI and 9 had impaired function of the affected kidney. APD-A was the most effective individual criterion for predicting the occurrence of a significant clinical event (AUC = 0.867).ConclusionIn our series, for children with UPJO, the most significant marker was prenatal APD >15 mm to predict an increase in dilation >5 mm.

Introduction Reunion Island, a French overseas department, is located in the southern hemisphere, close to the Capricorn tropic. This island has a multicultural and multiethnic population of 860 000 inhabitants, a quarter of whom are at high risk of developing skin cancer due to intense ultraviolet radiation. Melanoma is responsible for the majority of skin cancer deaths. The early prevention of melanoma is based on sun protection in childhood, but French regulations are not adapted to the environmental conditions of this tropical region.The main objective of our study is to evaluate the effectiveness of three sun protection programs conducted in Reunionese primary schools through a pupil knowledge questionnaire.Methods and analysis PRESOLRE is an interventional, open-label, cluster-randomised controlled trial, in four parallel arms, that is being conducted throughout 2022–2023 on Reunion Island. The trial design assumes an escalation interventional effect using: first, a control arm without proposed intervention (arm 1); second, an arm whose classes are encouraged to use the validated educational programme ‘Living With the Sun’ (LWS) (arm 2); third, an arm whose classes are encouraged to use both ‘LWS’ combined with ‘Mission Soleil Réunion’s sun protection awareness programme (arm 3); fourth, an arm benefiting from an intervention similar to arm 3, combined with the distribution of hats, sunglasses and sun creams (arm 4). In all, 1780 pupils from 18 classes of 20 pupils, on average, will be included. Randomisation applies to the classes of pupils (so defined as clusters). The primary outcome is based on the proportion of correct answers to the knowledge questions after the awareness programme, compared between the four arms using a linear mixed model with random intercept.Ethics and dissemination The study obtained ethics approval in 2022 (ID: 2022-A00350-43). Results will be published in peer-reviewed journals.Trial registration number NCT05367180.

Abstract Background Chest radiographs are routinely performed in intensive care unit (ICU) to confirm the correct position of an endotracheal tube (ETT) relative to the carina. However, their interpretation is often challenging and requires substantial time and expertise. The aim of this study was to propose an externally validated deep learning model with uncertainty quantification and image segmentation for the automated assessment of ETT placement on ICU chest radiographs. Methods The CarinaNet model was constructed by applying transfer learning to the RetinaNet model using an internal dataset of ICU chest radiographs. The accuracy of the model in predicting the position of the ETT tip and carina was externally validated using a dataset of 200 images extracted from the MIMIC-CXR database. Uncertainty quantification was performed using the level of confidence in the ETT–carina distance prediction. Segmentation of the ETT was carried out using edge detection and pixel clustering. Results The interrater agreement was 0.18 cm for the ETT tip position, 0.58 cm for the carina position, and 0.60 cm for the ETT–carina distance. The mean absolute error of the model on the external test set was 0.51 cm for the ETT tip position prediction, 0.61 cm for the carina position prediction, and 0.89 cm for the ETT–carina distance prediction. The assessment of ETT placement was improved by complementing the human interpretation of chest radiographs with the CarinaNet model. Conclusions The CarinaNet model is an efficient and generalizable deep learning algorithm for the automated assessment of ETT placement on ICU chest radiographs. Uncertainty quantification can bring the attention of intensivists to chest radiographs that require an experienced human interpretation. Image segmentation provides intensivists with chest radiographs that are quickly interpretable and allows them to immediately assess the validity of model predictions. The CarinaNet model is ready to be evaluated in clinical studies. Graphical Abstract

Abstract Background In the older general population, neurodegenerative diseases (NDs) are associated with increased disability, decreased physical and cognitive function. Detecting risk factors can help implement prevention measures. Using deep neural networks (DNNs), a machine-learning algorithm could be an alternative to Cox regression in tabular datasets with many predictive features. We aimed to compare the performance of different types of DNNs with regularized Cox proportional hazards models to predict NDs in the older general population. Methods We performed a longitudinal analysis with participants of the English Longitudinal Study of Ageing. We included men and women with no NDs at baseline, aged 60 years and older, assessed every 2 years from 2004 to 2005 (wave2) to 2016–2017 (wave 8). The features were a set of 91 epidemiological and clinical baseline variables. The outcome was new events of Parkinson’s, Alzheimer or dementia. After applying multiple imputations, we trained three DNN algorithms: Feedforward, TabTransformer, and Dense Convolutional (Densenet). In addition, we trained two algorithms based on Cox models: Elastic Net regularization (CoxEn) and selected features (CoxSf). Results 5433 participants were included in wave 2. During follow-up, 12.7% participants developed NDs. Although the five models predicted NDs events, the discriminative ability was superior using TabTransformer (Uno’s C-statistic (coefficient (95% confidence intervals)) 0.757 (0.702, 0.805). TabTransformer showed superior time-dependent balanced accuracy (0.834 (0.779, 0.889)) and specificity (0.855 (0.0.773, 0.909)) than the other models. With the CoxSf (hazard ratio (95% confidence intervals)), age (10.0 (6.9, 14.7)), poor hearing (1.3 (1.1, 1.5)) and weight loss 1.3 (1.1, 1.6)) were associated with a higher DNN risk. In contrast, executive function (0.3 (0.2, 0.6)), memory (0, 0, 0.1)), increased gait speed (0.2, (0.1, 0.4)), vigorous physical activity (0.7, 0.6, 0.9)) and higher BMI (0.4 (0.2, 0.8)) were associated with a lower DNN risk. Conclusion TabTransformer is promising for prediction of NDs with heterogeneous tabular datasets with numerous features. Moreover, it can handle censored data. However, Cox models perform well and are easier to interpret than DNNs. Therefore, they are still a good choice for NDs.

IntroductionHealth quality of life assessment is particularly important to measure the impact of chronic diseases. The aims of this study were to provide a cross-culturally adapted Creole-translation of the Medical Outcome Study Short-Form 36 (SF-36) and to assess psychometric performance of the Creole and French versions of the SF-36 among patients with type II diabetes in Reunion Island.Materials and methodsThe Creole translation and cross-cultural adaptation processes were based on the International Quality Of Life Assessment (IQOLA) methods. Internal consistency, test–retest reliability, convergent and discriminant validity using Multi-Trait-Multi-Method analysis and structural validity using exploratory factor analysis of the SF-36 for both versions were performed.ResultsIn the Creole version of the SF-36, Cronbach’s alpha exceeded 0.70 for all subscales except general health. In the French SF-36, Cronbach’s alpha exceeded 0.70 on all subscales except general health and bodily pain. In the Creole SF-36, intraclass correlation coefficient (ICC) for reproducibility was suboptimal. Multi-trait multi-method analysis showed that item-scale correlation exceeded 0.4 for all items except two general health items of the Creole SF-36 and one of the French SF-36. Factor analysis of 2 versions showed that the physical functioning, vitality, and mental health were each divided into two subscales.DiscussionOverall, our findings provided evidence that the SF-36 is adapted to Reunion Island in both Creole and French versions. However, further research could be conducted to investigate French–Creole differences in perceived health status and a cultural adaptation of the French version will be considered.

IntroductionChildhood chronic diseases affect family functioning and well-being. The aim of this study was to measure the impact of caring for a child with PUV, and the factors that most impact the burden of care.Patients and methodWe gave a questionnaire on the familial impact of having a child with posterior urethral valves to all parents of a child included in the CIRCUP trial from 2015 onwards. The questionnaire included questions about the parents' demographics, health, professional, financial and marital status and how these evolved since the child's birth as well as the “impact on family scale” (IOFS), which gives a total score ranging from 15 (no impact) to 60 (maximum impact). We then analyzed both the results of the specific demographic questions as well as the factors which influenced the IOFS score.ResultsWe retrieved answers for 38/51 families (74.5% response rate). The average IOFS score was 23.7 (15–51). We observed that the child's creatinine level had an effect on the IOFS score (p = 0.02), as did the parent's gender (p = 0.008), health status (p = 0.015), being limited in activity since the birth of the child (p = 0.020), being penalized in one's job (p = 0.009), being supported in one's job (p = 0.002), and decreased income (p = 0.004). Out of 38 mother/father binomials, 8/33 (24.2%) declared that they were no longer in the same relationship afterwards.ConclusionIn conclusion, having a boy with PUV significantly impacts families. The risk of parental separation and decrease in revenue is significant. Strategies aiming to decrease these factors should be put in place as soon as possible.

Abstract To describe the relationship between the use of laboratory tests and changes in laboratory parameters in ICU patients is necessary to help optimize routine laboratory testing. A retrospective, descriptive study was conducted on the large eICU-Collaborative Research Database. The relationship between the use of routine laboratory tests (chemistry and blood counts) and changes in ten common laboratory parameters was studied. Factors associated with laboratory tests were identified in a multivariate regression analysis using generalized estimating equation Poisson models. The study included 138,734 patient stays, with an ICU mortality of 8.97%. For all parameters, the proportion of patients with at least one test decreased from day 0 to day 1 and then gradually increased until the end of the ICU stay. Paradoxically, the results of almost all tests moved toward normal values, and the daily variation in the results of almost all tests decreased over time. The presence of an arterial catheter or teaching hospitals were independently associated with an increase in the number of laboratory tests performed. The paradox of routine laboratory testing should be further explored by assessing the factors that drive the decision to perform routine laboratory testing in ICU and the impact of such testing on patient.

IntroductionDue to the high prevalence of diabetes and its complications, evaluating the patient’s quality of life is critical. EQ-5D-5L is a valid tool for assessing health-related quality of life (HRQOL) in chronic diseases, including diabetes. However, no psychometric measures have been validated in a Creole-speaking population. Therefore, this study aimed for the first time to validate and cross-culturally adapt Creole and French versions of EQ-5D-5L on Type II diabetes patients in Reunion Island.Materials and methodsThe Creole translation and cross-cultural adaptation process were based on the EUROQOL methods. Internal consistency and construct validity were determined using confirmatory factor analysis (CFA) of EQ-5D-5L for both versions. CFA model for HRQOL and global fit measures were calculated based on the EQ-5D-5L items using the maximum likelihood (ML) method.ResultsFrom November 2016 to October 2017, 148 patients were included in the Creole group and 152 in the French group. EQ-5D-5L measures were unidimensional for both versions. Cronbach’s coefficient alpha was 0.76 for the Creole version and 0.81 for the French version in CFA models. The root mean square error of approximation (RMSEA) was 0.06 for the Creole version and 0.02 for the French version. The Comparative Fit Index (CFI) was closer to 1 for both versions. CFA models for both the Creole and French versions fit the data adequately.DiscussionOverall, our findings provided evidence that both the Creole and French versions of EQ-5D-5L are suitable for assessing HRQOL in diabetes patients in Reunion Island. However, further research could be done to investigate French–Creole differences concerning the perception of health status, and a cultural adaptation of the French version will be considered.

Abstract The aim of this study was to compare the prognosis of patients with acute respiratory failure (ARF) due to the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variant 501Y.V2 to that of patients with ARF due to the original strain. This retrospective matched cohort study included all consecutive patients who were hospitalized for ARF due to SARS-CoV-2 in Reunion Island University Hospital between March 2020 and March 2021. Twenty-eight in hospital mortality was evaluated before and after matching. A total of 218 patients with ARF due to SARS-CoV-2 were enrolled in the study. Of these, 83 (38.1%) were infected with the 501Y.V2 variant. During intensive care unit stay, 104 (47.7%) patients received invasive mechanical ventilation and 20 (9.2%) patients were supported by venovenous extracorporeal membrane oxygenation. Patients infected with the 501Y.V2 variant were younger (58 [51–68] vs. 67 [56–74] years old, P = 0.003), had less hypertension (54.2% vs 68.1%, P = 0.04), and had less chronic kidney disease (13.3% vs. 31.9%, P = 0.002) than patients infected with the original strain. After controlling for confounding variables (62 matched patients in each group), 28-day mortality was higher in the group of patients infected with the 501Y.V2 variant (30.6%) than in the group of patients infected with the original strain (19.4%, P = 0.04). In Reunion Island, where SARS-CoV-2 incidence remained low until February 2021 and the health care system was never saturated, mortality was higher in patients with ARF infected with the 501Y.V2 variant than in patients infected with the original strain.

Abstract Background Ventilator-associated pneumonia (VAP) caused by Stenotrophomonas maltophilia is poorly described in the literature. However, it has been shown to be associated with increased morbidity and mortality. Probabilistic antibiotic therapy against S. maltophilia is often ineffective as this pathogen is resistant to many antibiotics. There is no consensus at present on the best therapeutic strategy to adopt (class of antibiotics, antibiotic combination, dosage, treatment duration). The aim of this study was to evaluate the effect of antibiotic therapy strategy on the prognosis of patients with VAP caused by S. maltophilia. Results This retrospective study evaluated all consecutive patients who developed VAP caused by S. maltophilia between 2010 and 2018 while hospitalized in the intensive care unit (ICU) of a French university hospital in Reunion Island, in the Indian Ocean region. A total of 130 patients with a median Simplified Acute Physiology Score II of 58 [43–73] had VAP caused by S. maltophilia after a median duration of mechanical ventilation of 12 [5–18] days. Ventilator-associated pneumonia was polymicrobial in 44.6% of cases, and ICU mortality was 50.0%. After multivariate Cox regression analysis, the factors associated with increased ICU mortality were older age (hazard ratio (HR): 1.03; 95% CI 1.01–1.04, p = 0.001) and high Sequential Organ Failure Assessment score on the day of VAP onset (HR: 1.08; 95% CI 1.03–1.14, p = 0.002). Appropriate antibiotic therapy, and in particular trimethoprim–sulfamethoxazole, was associated with decreased ICU mortality (HR: 0.42; 95% CI 0.24–0.74, p = 0.003) and decreased hospital mortality (HR: 0.47; 95% CI 0.28–0.79, p = 0.04). Time to start of appropriate antibiotic therapy, combination therapy, and duration of appropriate antibiotic therapy had no effect on ICU mortality (p > 0.5). Conclusion In our study, appropriate antibiotic therapy, and in particular trimethoprim–sulfamethoxazole, was associated with decreased ICU and hospital mortality in patients with VAP caused by S. maltophilia.

ObjectiveBoys with posterior urethral valves (PUV) present an increased risk of febrile urinary tract infection (fUTI). Identifying specific risk factors could allow for tailoring UTI prevention. The aim of this study was to use the data from the CIRCUP randomized controlled trial data to identify patient characteristics associated with a higher risk of fUTI.Patients and methodsWe performed a secondary analysis of the data from the CIRCUP randomized trial which included boys with PUV, randomized to circumcision and antibiotic prophylaxis vs. antibiotic prophylaxis alone and followed for 2 years. There was only 1 episode of fUTI in the circumcision group vs. 17 in the uncircumcised group. We therefore only studied the antibiotic prophylaxis alone group and compared age at prenatal diagnosis, size and weight at birth, presence of dilating VUR at diagnosis, abnormal DMSA scan at 2 months, and nadir creatinine between children who presented a fUTI and those who did not, as well as age at first episode of fUTI.ResultsThe study group consisted of 42 patients of which 17 presented at least on fUTI. Presence of dilating VUR was significantly associated with risk of fUTI (p = 0.03), OR: 6 [CI 95% = (1.13–27.52)]. None of the other parameters were associated with increased risk of fUTI. We observed three distinct time periods for presenting a fUTI with a decrease in infection rate after the first 40 days of life, then at 240 days of life.ConclusionIn boys with PUV, presence of high-grade VUR is associated with a higher risk of presenting a fUTI. The rate of febrile UTIs seems to decrease after 9 months.

Abstract Introduction No hierarchy of first-line treatments for anogenital warts (AGWs) is provided in international guidelines. This study aimed to determine the efficacy of topical treatments and ablative procedures for the management of AGWs. Methods Twelve electronic databases were systematically searched from inception to August 2018. All randomized controlled trials (RCTs) comparing immunocompetent adults with AGWs who received at least 1 provider-administered or patient-administered treatment in at least 1 parallel group were included. Risk of bias assessment followed the Cochrane Handbook. The study endpoint was complete lesion response after clearance and recurrence assessment. A network meta-analysis was performed. Results A network geometry was constructed based on 49 of the 70 RCTs included in our systematic review. All but 4 RCTs had a high risk of bias. The most efficacious treatments compared to placebo were surgery (RR 10.54; CI 95% 4.53–24.52), ablative therapy + imiquimod (RR 7.52; CI 95% 4.53–24.52), and electrosurgery (RR 7.10; CI 95% 3.47–14.53). SUCRA values confirmed the superiority of surgery (90.9%), ablative therapy + imiquimod (79.8%), and electrosurgery (77.1%). The most efficacious patient-administered treatments were podophyllotoxin 0.5% solution (63.5%) and podophyllotoxin 0.5% cream (62.2%). Conclusions With low-level evidence of most included RCTs, surgery and electrosurgery were superior to other treatments after clearance and recurrence assessment. Podophyllotoxin 0.5% was the most efficacious patient-administered treatment. Combined therapies should be evaluated in future RCTs in view of their identified effectiveness. The results of future RCTs should systematically include clinical type, number and location of AGWs, and sex of the patient, to refine therapeutic indications. Protocol Registration PROSPERO-CRD42015025827

Abstract Background The recommendations of learned societies mention risk factors for the presence of multidrug resistant bacteria in hospital-acquired infections, but they do not propose a scoring system to guide empiric antibiotic therapy. Our study was aimed at developing a simple score for predicting “the presence of bacteria requiring carbapenem treatment” in ICU-acquired bloodstream infection and pneumonia. Methods Between December 2011 and January 2015, we conducted a retrospective study using a prospectively collected French database of nosocomial infections in the polyvalent intensive care unit of a French university hospital. All patients with ICU-acquired bloodstream infection or pneumonia were included in the study. Bivariate and multivariate analyses were performed to develop the CarbaSCORE, and this score was internally validated. Results In total, 338 patients were analyzed, including 27 patients requiring carbapenem treatment. The CarbaSCORE was composed of four criteria: “presence of bloodstream infection” (as opposed to pneumonia) scored 2 points, “chronic hemodialysis” scored 4 points, “travel abroad in the last 6 months” scored 5 points, and “MDR-colonization or prior use of a β-lactam of class ≥ 3” scored 6 points. Internal validation by bootstrapping showed an area under the receiver operating characteristic curve of 0.81 [0.73–0.89]. Sensitivity was 96% at the 6-point threshold and specificity was 91% at the 9-point threshold. Conclusions The CarbaSCORE is a simple and efficient score for predicting the presence of bacteria requiring carbapenem treatment. Further studies are needed to test this score before it can be used in practice.

Abstract Background Few data are available on the impact of levosimendan in refractory cardiogenic shock patients undergoing peripheral venoarterial extracorporeal membrane oxygenation (VA-ECMO). The aim of this study was to evaluate the impact of levosimendan on VA-ECMO weaning in patients hospitalized in intensive care unit (ICU). Methods This retrospective cohort study was conducted in a French university hospital from 2010 to 2017. All patients hospitalized in ICU undergoing VA-ECMO were consecutively evaluated. Results A total of 150 patients undergoing VA-ECMO were eligible for the study. Thirty-eight propensity-matched patients were evaluated in the levosimendan group and 65 in the non-levosimendan group. In patients treated with levosimendan, left ventricular ejection fraction had increased from 21.5 ± 9.1% to 30.7 ± 13.5% (P

BACKGROUND:Intensive care unit (ICU) patients with the most severe forms of acute coronary syndrome (ACS) require invasive therapies such as extracorporeal life support. The risk of bleeding in ICU patients with ACS treated with a dual antiplatelet therapy of aspirin and ticagrelor is unknown. The primary objective of this study was to compare the bleeding risk of ticagrelor and clopidogrel in ICU patients with ACS. METHODS AND FINDINGS:We conducted a retrospective study based on a propensity score and a proportional hazards model. All patients with ACS hospitalized in the ICU of a French university hospital between January 2013 and January 2017 were included in the study. Bleeding during ICU stay was defined as all Thrombolysis in myocardial infarction (TIMI) major or minor events. A total of 155 patients were included in the study. According to propensity score matching, 57 patients treated with aspirin and ticagrelor were matched with 57 patients treated with aspirin and clopidogrel. Median (first-third quartile) Simplified Acute Physiology Score II was 61.5 (41.0-85.0). Bleeding during ICU stay occurred in 12 patients (21.1%) treated with clopidogrel and in 35 patients (61.4%) treated with ticagrelor (p

PurposePatients with acute severe asthma (ASA) may in rare cases require invasive mechanical ventilation (IMV). However, recent data on this issue are lacking.Materials and methodsIn this retrospective and bicentric study conducted on a 10 year period, we investigate the in-hospital mortality in patients with ASA requiring IMV. We compare this mortality to that of patients with other types of respiratory distress using a standardized mortality ratio (SMR) model.ResultsEighty-one episodes of ASA requiring IMV were evaluated. Factors significantly associated with in-hospital mortality were cardiac arrest on day of admission, cardiac arrest as the reason for intubation, absence of decompensation risk factors, need for renal replacement therapy on day of admission, and intubation in pre-hospital setting. Non-survivors had higher SAPS II, SOFA, creatinine and lactate levels as well as lower blood pressure, pH, and HCO3 on day of admission. In-hospital mortality was 15% (n = 12). Compared to a reference population of 2,670 patients, the SMR relative to the SAPS II was very low at 0.48 (95% CI, 0.25-0.84). The only factor independently associated with in-hospital mortality was cardiac arrest on day of admission. In-hospital mortality was 69% in patients with cardiac arrest on day of admission and 4% in others (p < 0.01). Salvage therapies were given to 7 patients, sometimes in combination with each other: ECMO (n = 6), halogenated gas (n = 1) and anti-IL5 antibody (n = 1). Death occurred in only 2 of these 7 patients, both of whom had cardiac arrest on day of admission.ConclusionNowadays, the mortality of patients with ASA requiring IMV is low. Death is due to multi-organ failure, with cardiac arrest on day of admission being the most important risk factor. In patients who did not have cardiac arrest on day of admission the mortality is even lower (4%) which allows an aggressive management.

BACKGROUND:Treatment by venoarterial extracorporeal membrane oxygenation (VA-ECMO) is widely used today, even though it is associated with high risks of complications and death. While studies have focused on the relationship between some of these complications and the risk of death, the relationship between different complications has never been specifically examined, despite the fact that the occurrence of one complication is known to favor the occurrence of others. Our objective was to describe the relationship between complications in patients undergoing VA-ECMO in intensive care unit (ICU) and to identify, if possible, patterns of patients according to complications. METHODS AND FINDINGS:As part of a retrospective cohort study, we conducted a multiple correspondence analysis followed by a hierarchical ascendant classification in order to identify patterns of patients according to main complications (sepsis, thromboembolic event, major transfusion, major bleeding, renal replacement therapy) and in-ICU death. Our cohort of 145 patients presented an in-ICU mortality rate of 50.3%. Morbidity was high, with 36.5% of patients presenting three or more of the five complications studied. Multiple correspondence analysis revealed a cumulative inertia of 76.9% for the first three dimensions. Complications were clustered together and clustered close to death, prompting the identification of four patterns of patients according to complications, including one with no complications. CONCLUSIONS:Our study, based on a large cohort of patients undergoing VA-ECMO in ICU and presenting a mortality rate comparable to that reported in the literature, identified numerous and often interrelated complications. Multiple correspondence analysis and hierarchical ascendant classification yielded clusters of patients and highlighted specific links between some of the complications studied. Further research should be conducted in this area.

Our objective was to evaluate the effectiveness of peer-led self-management education in improving glycaemic control in patients with type 2 diabetes in a low-income country (Mali).We conducted an open-label randomised controlled trial. A total of 151 adults (76% women, mean age 52.5) with type 2 diabetes (HbA1c≥8%), treated in the diabetes consultation units of two secondary health centres in Bamako, were allocated to peer-led structured patient education (n = 76) or conventional care alone (n = 75). The intervention group received 1 year of culturally tailored structured patient education (3 courses of 4 sessions) delivered in the community by five trained peer educators. Both groups underwent conventional diabetes monitoring and follow-up. Primary outcome was the mean absolute change in HbA1c from baseline to 12 months.177 education sessions were delivered to the intervention group. Patient attrition was 8%. From baseline to 12 months, the decrease in HbA1c levels was 1.05% (SD = 2.0; CI95%: 1.54;-0.56) in the intervention group compared with 0.15% (SD = 1.7; CI95%: -0.56; 0.26) in the control group, p = 0.006. Mean BMI change was -1.65 kg/m2 (SD = 2.5; CI95%: -2.25; -1.06) in the intervention group and +0.05 kg/m2 (SD = 3.2; CI95%: -0.71; 0.81) in the control group, p = 0.0005. Mean waist circumference decreased by 3.34 cm (SD = 9.3; CI95%: -5.56;-1.13) in the intervention group and increased by 2.65 cm (SD = 10.3; CI95%: 0.20; 5.09) in the control group, p = 0.0003.Peer-led structured patient education delivered over 1 year to patients with poorly controlled type 2 diabetes in Mali yielded substantial improvements in glycaemic control and anthropometric parameters. This is of importance for the scaling up of efficient interventions in low-resource settings in the future.ClinicalTrials.gov NCT01485913.

Objectives: While the prevalence of Group B streptococcus (GBS) colonization is important, little is known about invasive GBS (iGBS) disease in tropical areas. Our objective was to assess the burden of iGBS disease among non-pregnant adults. Methods: A prospective hospital-based study of all non-pregnant adult patients with iGBS disease was conducted between January and December 2011 in Saint Pierre, Réunion Island, to assess its cumulative incidence rate (CIR). Capsular serotyping and multilocus sequence typing were performed to characterize GBS isolates. Case-control study was done to identify risk factors. Results: The overall CIR of iGBS disease was 10.1 per 100,000. The CIR in elderly patients (≥ 65 yrs) was estimated at 40.6 per 100.000, and that of adults (15-64 years) at 6.7 per 100.000. Aboriginal origin in the Indian Ocean and overweight were both associated with iGBS disease. The most prominent clinical forms were osteo-articular and skin/soft tissue infections, as a consequence of diabetic foot. The serotypes were classic, type-Ia being the most prevalent. The hyper virulent ST-17 (CC17) was associated with type-III. Conclusions: The incidence of iGBS disease found in Réunion island is twofold that usually reported. This burden is linked to overweight in aboriginal people from the Indian Ocean.

The benefits of cardiac surgery are sometimes difficult to predict and the decision to operate on a given individual is complex. Machine Learning and Decision Curve Analysis (DCA) are recent methods developed to create and evaluate prediction models.We conducted a retrospective cohort study using a prospective collected database from December 2005 to December 2012, from a cardiac surgical center at University Hospital. The different models of prediction of mortality in-hospital after elective cardiac surgery, including EuroSCORE II, a logistic regression model and a machine learning model, were compared by ROC and DCA. Of the 6,520 patients having elective cardiac surgery with cardiopulmonary bypass, 6.3% died. Mean age was 63.4 years old (standard deviation 14.4), and mean EuroSCORE II was 3.7 (4.8) %. The area under ROC curve (IC95%) for the machine learning model (0.795 (0.755-0.834)) was significantly higher than EuroSCORE II or the logistic regression model (respectively, 0.737 (0.691-0.783) and 0.742 (0.698-0.785), p < 0.0001). Decision Curve Analysis showed that the machine learning model, in this monocentric study, has a greater benefit whatever the probability threshold.According to ROC and DCA, machine learning model is more accurate in predicting mortality after elective cardiac surgery than EuroSCORE II. These results confirm the use of machine learning methods in the field of medical prediction.

End-of-life decision-making in Intensive care Units (ICUs) is difficult. The main problems encountered are the lack of a reliable prediction score for death and the fact that the opinion of patients is rarely taken into consideration. The Decision Curve Analysis (DCA) is a recent method developed to evaluate the prediction models and which takes into account the wishes of patients (or surrogates) to expose themselves to the risk of obtaining a false result. Our objective was to evaluate the clinical usefulness, with DCA, of the Simplified Acute Physiology Score II (SAPS II) to predict ICU mortality.We conducted a retrospective cohort study from January 2011 to September 2015, in a medical-surgical 23-bed ICU at University Hospital. Performances of the SAPS II, a modified SAPS II (without AGE), and age to predict ICU mortality, were measured by a Receiver Operating Characteristic (ROC) analysis and DCA.Among the 4.370 patients admitted, 23.3% died in the ICU. Mean (standard deviation) age was 56.8 (16.7) years, and median (first-third quartile) SAPS II was 48 (34-65). Areas under ROC curves were 0.828 (0.813-0.843) for SAPS II, 0.814 (0.798-0.829) for modified SAPS II and of 0.627 (0.608-0.646) for age. DCA showed a net benefit whatever the probability threshold, especially under 0.5.DCA shows the benefits of the SAPS II to predict ICU mortality, especially when the probability threshold is low. Complementary studies are needed to define the exact role that the SAPS II can play in end-of-life decision-making in ICUs.

To assess the Burgundy perinatal network (18 obstetrical units; 18 500 births per year), discharge abstracts and additional data were collected for all mothers and newborns. In accordance with French law, data were rendered anonymous before statistical analysis, and were linked to patients using a specific procedure. This procedure allowed data concerning each mother to be linked to those for her newborn(s). This study showed that all mothers and newborns were included in the regional database; the data for all mothers were linked to those for their infant(s) in all cases. Additional data (gestational age) were obtained for 99.9% of newborns.