Sylvie Chevret

Paris Diderot University, Lutetia Parisorum, Île-de-France, France

Are you Sylvie Chevret?

Claim your profile

Publications (482)3219.71 Total impact

  • Revue d Épidémiologie et de Santé Publique 05/2015; 63. DOI:10.1016/j.respe.2015.03.014 · 0.66 Impact Factor
  • S. Chevret
    Revue d Épidémiologie et de Santé Publique 05/2015; 63. DOI:10.1016/j.respe.2015.03.003 · 0.66 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Systemic rheumatic diseases (SRD) patients may require ICU management for SRD exacerbation, treatment-related infectious or toxicities. Observational study in 10 university-affiliated ICUs in France. Consecutive patients with SRD were included. Determinants of ICU mortality were identified through multivariable logistic analysis. 363 patients (65.3 % women, median age 59y (IQR, 42-70)) accounted for 381 admissions. Connective tissue disease (primarily Systemic Lupus Erythematosus) accounted for 66.1% of SRD and systemic vasculitides for 26.2 % (chiefly ANCA-associated vasculitides). SRD was newly diagnosed in 43 (11.3%) cases. Direct admission to ICU occurred in 143 (37.9%) cases. Reasons for ICU admissions were infection (39.9%), SRD exacerbation (34.4%), toxicity (5.8 %) or miscellaneous (19.9%). Respiratory involvement was the leading cause of admission (56.8 %), followed by shock (41.5 %) and acute kidney injury (42.2%). Median SOFA on day-1 was 5 [3-8]. Mechanical ventilation was required in 57% cases, vasopressors in 33.9% and renal replacement therapy in 28.1%. ICU mortality rate was 21.0% (80 deaths). Factors associated with ICU mortality were shock (OR: 3.77 [95%CI, 1.93 -7.36]), SOFA score at day 1 (OR: 1.19 [95%CI, 1.10-1.30]) and direct admission (OR: 0.52, [CI 0.28-0.97]. Neither comorbidities nor SRD characteristics were associated with survival. In SRD patients, critical care management is mostly needed in patients with previously known SRD; still, diagnosis can be made in the ICU in 12% of the patients. Infection and SRD exacerbation account for more than two-third of the situations, both targeting chiefly the lungs. Direct admission to the ICU might improve outcomes.
    Chest 05/2015; DOI:10.1378/chest.14-3098 · 7.13 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: The propensity score (PS) is a balancing score. Following PS matching, balance checking usually relies on estimating separately the standardized absolute mean difference for each baseline characteristic. The average standardized absolute mean difference and the Mahalanobis distances have been proposed to summarize the information across the covariates. However, they might be minimized when nondesirable variables such as instrumental variables (IV) are included in the PS model. We propose a new weighted summary balance measure that takes into account, for each covariate, its strength of association with the outcome. This new measure was evaluated using a simulation study to assess whether minimization of the measure coincided with minimally biased estimates. All measures were then applied to a real data set from an observational cohort study. Contrarily to the other measures, our proposal was minimized when including the confounders, which coincided with minimal bias and mean squared error, but increased when including an IV in the PS model. Similar findings were observed in the real data set. A balance measure taking into account the strength of association between the covariates and the outcome may be helpful to identify the most parsimonious PS model. Copyright © 2015 Elsevier Inc. All rights reserved.
    Journal of Clinical Epidemiology 05/2015; DOI:10.1016/j.jclinepi.2015.04.009 · 5.48 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: In this study, we randomly compared high doses of the tyrosine kinase inhibitor (TKI) imatinib combined with reduced-intensity chemotherapy (arm A) to standard imatinib/HyperCVAD treatment (arm B) in 268 adults (median age, 47 years) with Ph-positive acute lymphoblastic leukemia (ALL). The primary objective was the major molecular response (MMolR) rate after cycle 2, patients being then eligible for allogeneic stem cell transplantation (SCT) if they had a donor or autologous SCT if in MMolR and no donor. With less induction deaths, the CR rate was higher in arm A (98% versus 91%, P= 0.006), while MMolR rate was similar in both arms (66% versus 64%). With a median follow-up of 4.8 years, 5-year EFS and OS were estimated at 37.1% and 45.6% respectively, without difference between both arms. Allogeneic transplantation was associated with a significant benefit in RFS (HR, 0.69; P= 0.036) and OS (HR, 0.64; P= 0.02), initial WBC being the only factor significantly interacting with this SCT effect. In patients achieving MMolR, outcome was similar after autologous and allogeneic transplantation. This study validates an induction regimen combining reduced-intensity chemotherapy and imatinib in Ph-positive ALL adult patients and suggests that SCT in first CR is still a good option in Ph-positive ALL adult patients. The study is registered to as NCT00327678. Copyright © 2015 American Society of Hematology.
    Blood 04/2015; 125(24). DOI:10.1182/blood-2015-02-627935 · 10.43 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Human adenovirus (HAdV) infections are a major cause of morbidity in pediatric hematopoietic stem cell transplant (HSCT) patients. New antiviral treatments offer promising perspectives. However it remains challenging to identify patients at risk for disseminated infection and which should receive early antiviral intervention. We conducted a longitudinal study of allogeneic HSCT recipients, including a weekly HAdV monitoring, to determine the risks factors associated with HAdV infection and dissemination, and to assess whether HAdV loads in stool may be used as surrogate markers for HAdV dissemination. Between September 2010 and December 2011, out of 72 patients, the cumulative incidences at day 100 of HAdV digestive infection, systemic infection and related disease were 35.9%, 24.0% and 18.3%, respectively. In multivariate analysis, the risk factors for HAdV digestive and systemic infection were cord blood and in vitro T-cell depletion. Graft versus host disease (GvHD) grade >2 was also associated with systemic infection. In patients with HAdV digestive shedding, GvHD grade >2 and HAdV load in stool were the only risk factors of systemic infection. Median peak levels of HAdV in stool were 7.9 and 4.0 log10 copies/ml respectively in patients with HAdV systemic infection and those without. HAdV monitoring in stool of pediatric HSCT recipients receiving cord blood or in vitro T-cell depleted transplants helps to predict patients at risk for HAdV systemic infection. Our results provide a rationale to conduct randomized controlled trial to evaluate the benefit of anti-HAdV preemptive treatments based on HAdV DNA levels in stool. Copyright © 2015. Published by Elsevier Ltd.
    Clinical Microbiology and Infection 04/2015; 21(7). DOI:10.1016/j.cmi.2015.03.011 · 5.20 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Reports of patients with secondary acute promyelocytic leukemia (APL) have increased in recent years, particularly for those who received treatment with mitoxantrone, and retrospective studies have suggested that their characteristics and outcomes were similar to those of patients with de novo APL. The authors investigated patients with de novo and secondary APL who were included in the ongoing APL-2006 trial. Patients with secondary APL who were included in that trial also were compared with a previous retrospective cohort of patients with secondary APL. In the APL-2006 trial, 42 of 280 patients (15%) had secondary APL. Compared with the retrospective cohort, patients with secondary APL in the APL-2006 trial had a lower incidence of prior breast carcinoma (35.7% vs 57%; P = .03) and a higher incidence of prior prostate carcinoma (26.2% vs 4.7%; P < .001). Treatment of the primary tumor in the APL-2006 trial less frequently included combined radiochemotherapy (28.6% vs 47.2%; P = .044) and no mitoxantrone (0% vs 46.7%; P = .016) but more frequently included anthracyclines (53.3% vs 38.3%; P = .015). In the APL-2006 trial, patients who had secondary APL, compared with those who had de novo APL, were older (mean, 60.2 years vs 48.7 years, respectively; P < .0001) but had a similar complete response rate (97.6% vs 90.3%, respectively), cumulative incidence of relapse (0% vs 1.8%, respectively), and overall survival (92.3% vs 90.9%, respectively) at 18 months. Although the incidence of secondary APL appears to be stable over time, evolving strategies for the treatment of primary cancers have reduced its occurrence among breast cancer patients but have increased its incidence among patients with prostate cancer. The current results confirm prospectively that patients with secondary APL have characteristics and outcomes similar to those of patients with de novo APL. Cancer 2015. © 2015 American Cancer Society. © 2015 American Cancer Society.
    Cancer 04/2015; DOI:10.1002/cncr.29389 · 4.90 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Rationale Systemic steroids are the standard treatment for bronchiolitis obliterans syndrome (BOS) after allogeneic hematopoietic stem cell transplantation (HSCT), despite their poor efficacy and disabling side effects. Objectives To evaluate the effectiveness and tolerance of budesonide/formoterol as an alternative treatment for BOS following HSCT. Methods In this randomized, double-blind, placebo-controlled study, we randomly assigned 32 HSCT recipients with mild/severe BOS to receive budesonide/formoterol or placebo for six months. The primary outcome was the change in the FEV1 after one month of treatment (M1) compared with the baseline value. Patients were unblinded at M1 if there was no improvement in the FEV1. Those who had initially received the placebo were switched to budesonide/formoterol. Intention-to-treat analysis was performed to assess the primary outcome. Additional analyses took scheduled treatment contamination into account. Measurements and main results At M1, the median FEV1 increased by 260 mL in the budesonide/formoterol arm compared with 5 mL in the placebo arm (p=0.012). The median increases in the FEV1 at M1 relative to the baseline value for the treated and placebo groups were 13% and 0%, respectively (p=0.019). Twenty-five patients received budesonide/formoterol during the study. The median difference in the FEV1 between the baseline and after one month of treatment for these patients was +240 mL (p=0.0001). The effect of budesonide/formoterol on the FEV1 was maintained in the 13 patients who completed six months of treatment. Conclusions Budesonide/formoterol administration led to a significant improvement in the FEV1 in patients with mild/severe BOS after allogeneic HSCT. Clinical trial registration available at, ID NCT00624754.
    American Journal of Respiratory and Critical Care Medicine 04/2015; 191(11). DOI:10.1164/rccm.201410-1818OC · 11.99 Impact Factor
  • Journal of Hepatology 04/2015; 62:S213-S214. DOI:10.1016/S0168-8278(15)30057-X · 10.40 Impact Factor
  • Journal of Hepatology 04/2015; 62:S196. DOI:10.1016/S0168-8278(15)30020-9 · 10.40 Impact Factor
  • Leukemia Research 04/2015; 39:S63. DOI:10.1016/S0145-2126(15)30123-5 · 2.69 Impact Factor
  • Leukemia Research 04/2015; 39:S44. DOI:10.1016/S0145-2126(15)30088-6 · 2.69 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Intensive care unit (ICU) admission is associated with high mortality in allogeneic hematopoietic stem cell transplant (HSCT) recipients. Whether mortality has decreased recently is unknown. The 497 adult allogeneic HSCT recipients admitted to three ICUs between 1997 and 2011 were evaluated retrospectively. Two hundred and nine patients admitted between 1997 and 2003 were compared with the 288 patients admitted from 2004 to 2011. Factors associated with 90-day mortality were identified. The recent cohort was characterized by older age, lower conditioning intensity, and greater use of peripheral blood or unrelated-donor graft. In the recent cohort, ICU was used more often for patients in hematological remission (67% vs 44%; P<0.0001) and without GVHD (73% vs 48%; P<0.0001) or invasive fungal infection (85% vs 73%; P=0.0003) despite a stable admission rate (21.7%). These changes were associated with significantly better 90-day survival (49% vs 31%). Independent predictors of hospital mortality were GVHD, mechanical ventilation (MV) and renal replacement therapy (RRT). Among patients who required MV or RRT, survival was 29% and 18%, respectively, but dropped to 18% and 6% in those with GVHD. The use of ICU admission has changed and translated into improved survival, but advanced life support in patients with GVHD usually provides no benefits.Bone Marrow Transplantation advance online publication, 23 March 2015; doi:10.1038/bmt.2015.55.
    Bone marrow transplantation 03/2015; 50(6). DOI:10.1038/bmt.2015.55 · 3.47 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The natural history of pulmonary Langerhans cell histiocytosis (PLCH) has been unclear due to the absence of prospective studies. The rate of patients who experience an early progression of their disease is unknown. Additionally, conflicting effects of smoking cessation on the outcome of PLCH have been reported. In this prospective, multicentre study, 58 consecutive patients with newly diagnosed PLCH were comprehensively evaluated over a two-year period. Our objectives were to estimate the incidence of early progression of the disease and to evaluate the impact of smoking status on lung function outcomes. Lung function deterioration was defined as a decrease of at least 15% in FEV1 and/or FVC and/or DLCO, compared with baseline values. At each visit, smoking status was recorded based on the patients' self-reports and urinary cotinine measurements that were blinded for the patients. The cumulative incidence of lung function outcomes over time was estimated using the non-parametric Kaplan-Meier method. Multivariate Cox models with time-dependent covariates were used to calculate the hazards ratios of the lung function deterioration associated with smoking status with adjustment for potential confounders. The cumulative incidence of lung function deterioration at 24 months was 38% (22% for FEV1 and DLCO, and 9% for FVC). In the multivariate analysis, smoking status and PaO2 at inclusion were the only factors associated with the risk of lung function deterioration. The patients' smoking statuses markedly changed over time. Only 20% of the patients quit using tobacco for the entire study period. Nevertheless, being a non-smoker was associated with a decreased risk of subsequent lung function deterioration, even after adjustment for baseline predictive factors. By serial lung computed tomography, the extent of cystic lesions increased in only 11% of patients. Serial lung function evaluation on a three- to six-month basis is essential for the follow-up of patients with recently diagnosed PLCH to identify those who experience an early progression of their disease. These patients are highly addicted to tobacco, and robust efforts should be undertaken to include them in smoking cessation programs. No: NCT01225601 .
    Orphanet Journal of Rare Diseases 03/2015; 10(1):30. DOI:10.1186/s13023-015-0249-2 · 3.96 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Various critical events, liver-related or not, occur in patients with compensated cirrhosis but their respective burden remain to be prospectively assessed. The aim of this prospective cohort study involving 35 French centres was to capture the whole spectrum of complications occurring in compensated viral cirrhosis using competing risks analyses. Inclusion criteria were: histologically proven cirrhosis due to HCV or HBV; Child-Pugh A; no previous hepatic complications. The cohort was considered as a multi-state disease model, cumulative incidences (CumI) of events were estimated in a competing risks framework. 1,654 patients were enrolled from 2006 to 2012 (HCV 1308, HBV 315, HCV-HBV 31). During a median follow-up of 34 months, at least one liver nodule was detected in 271 patients, confirmed as hepatocellular carcinoma (HCC) in 128 (4-yr cumI: 10.5%) and cholangiocarcinoma in 3. HCC incidence was higher in HCV (4-yr cumI: 11.4% vs. 7.4%, P=0.05). HCC fulfilled Milan criteria in 79.3%, leading to curative treatment in 70.4%. Liver decompensation occurred more frequently in HCV patients (4-yr cumI: 10.8% vs. 3.6%, P=0.0004). Virological eradication/control was achieved in 34.1% HCV and 88.6% HBV patients and was associated to a marked decrease in HCC, decompensation and bacterial infection incidences. Survival was shorter in HCV patients (4-yr cumI 91.6% vs. 97.2%, P=0.0002). Death (n=102, missing data 6) was attributed to liver disease in 48 (47%; liver cancer: n=18, miscellaneous: n=30) and to extra-hepatic causes in 48 (47%; bacterial infection n=13, extra-hepatic cancers n=10, cardiovascular events n=5, miscellaneous: n=20). Conclusion: After 3 years of follow-up, extra-hepatic events still explained half of deaths in patients with compensated viral cirrhosis. A strong decrease in complications was linked to virological eradication/control. This article is protected by copyright. All rights reserved. © 2015 by the American Association for the Study of Liver Diseases.
  • J. Lambert, R. Porcher, S. Chevret
    01/2015; 4(1):103-113. DOI:10.6000/1929-6029.2015.04.01.12
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: An increased proportion of deaths occur in the intensive care unit (ICU). We performed this prospective study in 41 ICUs to determine the prevalence and determinants of complicated grief after death of a loved one in the ICU. Relatives of 475 adult patients were followed up. Complicated grief was assessed at 6 and 12 months using the Inventory of Complicated Grief (cut-off score >25). Relatives also completed the Hospital Anxiety and Depression Scale at 3 months, and the Revised Impact of Event Scale for post-traumatic stress disorder symptoms at 3, 6 and 12 months. We used a mixed multivariate logistic regression model to identify determinants of complicated grief after 6 months. Among the 475 patients, 282 (59.4%) had a relative evaluated at 6 months. Complicated grief symptoms were identified in 147 (52%) relatives. Independent determinants of complicated grief symptoms were either not amenable to changes (relative of female sex, relative living alone and intensivist board certification before 2009) or potential targets for improvements (refusal of treatment by the patient, patient died while intubated, relatives present at the time of death, relatives did not say goodbye to the patient, and poor communication between physicians and relatives). End-of-life practices, communication and loneliness in bereaved relatives may be amenable to improvements. Copyright ©ERS 2015.
    European Respiratory Journal 01/2015; 45(5). DOI:10.1183/09031936.00160014 · 7.13 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: The prognosis of critically ill cancer patients has improved recently. Controversies remain as regard to the specific prognosis impact of neutropenia in critically ill cancer patients. The primary objective of this study was to assess hospital outcome of critically ill neutropenic cancer patients admitted into the ICU. The secondary objective was to assess risk factors for unfavorable outcome in this population of patients and specific impact of neutropenia. We performed a post hoc analysis of a prospectively collected database. The study was carried out in 17 university or university-affiliated centers in France and Belgium. Neutropenia was defined as a neutrophil count lower than 500/mm(3). Among the 1,011 patients admitted into the ICU during the study period 289 were neutropenic at the time of admission. Overall, 131 patients died during their hospital stay (hospital mortality 45.3 %). Four variables were associated with a poor outcome, namely allogeneic transplantation (OR 3.83; 95 % CI 1.75-8.35), need for mechanical ventilation (MV) (OR 6.57; 95 % CI 3.51-12.32), microbiological documentation (OR 2.33; CI 1.27-4.26), and need for renal replacement therapy (OR 2.77; 95 % CI 1.34-5.74). Two variables were associated with hospital survival, namely age younger than 70 (OR 0.22; 95 % CI 0.1-0.52) and neutropenic enterocolitis (OR 0.37; 95 % CI 0.15-0.9). A case-control analysis was also performed with patients of the initial database; after adjustment, neutropenia was not associated with hospital mortality (OR 1.27; 95 % CI 0.86-1.89). Hospital survival was closely associated with younger age and neutropenic enterocolitis. Conversely, need for conventional MV, for renal replacement therapy, and allogeneic hematopoietic stem cell transplantation (HSCT) were associated with poor outcome.
    Intensive Care Medicine 01/2015; 41(2). DOI:10.1007/s00134-014-3615-y · 5.54 Impact Factor
  • S Chevret, S Seaman, M Resche-Rigon
    [Show abstract] [Hide abstract]
    ABSTRACT: Missing values in clinical studies are almost unavoidable. When analyzing such data, the standard response is to exclude the patients with missing data. This is known as ‘complete case analysis’ (CCA) and has been shown to be the leading strategy in the epidemiology [1] and intensive care unit (ICU) literature [2]. However, if the excluded patients are not a representative subsample from the whole sample, their exclusion can lead to bias and loss of precision in estimation, both of which can, for example, adversely affect the performance of predictive risk models in the ICU (Supplementary 1). To deal with this issue, numerous imputation methods have been developed. The simplest method is “simple imputation.” This involves replacing each missing value with a single value, such as the mean of the observed data [3]. Thereafter, all patients present in the sample can be included in the analysis. The simplicity and ease of implementation of this method make it attractive. However, it tends ...
    Intensive Care Medicine 01/2015; 41(2). DOI:10.1007/s00134-014-3624-x · 5.54 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Improved mortality prediction for patients in intensive care units is a big challenge. Many severity scores have been proposed, but findings of validation studies have shown that they are not adequately calibrated. The Super ICU Learner Algorithm (SICULA), an ensemble machine learning technique that uses multiple learning algorithms to obtain better prediction performance, does at least as well as the best member of its library. We aimed to assess whether the Super Learner could provide a new mortality prediction algorithm for patients in intensive care units, and to assess its performance compared with other scoring systems. From January, 2001, to December, 2008, we used the Multiparameter Intelligent Monitoring in Intensive Care II (MIMIC-II) database (version 26) including all patients admitted to an intensive care unit at the Beth Israel Deaconess Medical Centre, Boston, MA, USA. We assessed the calibration, discrimination, and risk classification of predicted hospital mortality based on Super Learner compared with SAPS-II, APACHE-II, and SOFA. We calculated performance measures with cross-validation to avoid making biased assessments. Our proposed score was then externally validated on a dataset of 200 randomly selected patients admitted at the intensive care unit of Hôpital Européen Georges-Pompidou, Paris, France, between Sept 1, 2013, and June, 30, 2014. The primary outcome was hospital mortality. The explanatory variables were the same as those included in the SAPS II score. 24 508 patients were included, with median SAPS-II of 38 (IQR 27-51) and median SOFA of 5 (IQR 2-8). 3002 of 24 508 (12%) patients died in the Beth Israel Deaconess Medical Centre. We produced two sets of predictions based on the Super Learner; the first based on the 17 variables as they appear in the SAPS-II score (SL1), and the second, on the original, untransformed variables (SL2). The two versions yielded average predicted probabilities of death of 0·12 (IQR 0·02-0·16) and 0·13 (0·01-0·19), whereas the corresponding value for SOFA was 0·12 (0·05-0·15) and for SAPS-II 0·30 (0·08-0·48). The cross-validated area under the receiver operating characteristic curve (AUROC) for SAPS-II was 0·78 (95% CI 0·77-0·78) and 0·71 (0·70-0·72) for SOFA. Super Learner had an AUROC of 0·85 (0·84-0·85) when the explanatory variables were categorised as in SAPS-II, and of 0·88 (0·87-0·89) when the same explanatory variables were included without any transformation. Additionally, Super Learner showed better calibration properties than previous score systems. On the external validation dataset, the AUROC was 0·94 (0·90-0·98) and calibration properties were good. Compared with conventional severity scores, Super Learner offers improved performance for predicting hospital mortality in patients in intensive care units. A user-friendly implementation is available online and should be useful for clinicians seeking to validate our score. Fulbright Foundation, Assistance Publique-Hôpitaux de Paris, Doris Duke Clinical Scientist Development Award, and the NIH. Copyright © 2014 Elsevier Ltd. All rights reserved.
    The Lancet Respiratory Medicine 11/2014; 3(1). DOI:10.1016/S2213-2600(14)70239-5

Publication Stats

18k Citations
3,219.71 Total Impact Points


  • 2009–2015
    • Paris Diderot University
      Lutetia Parisorum, Île-de-France, France
  • 2004–2014
    • Université de Versailles Saint-Quentin
      Versailles, Île-de-France, France
    • Centre Hospitalier Régional Universitaire de Lille
      Lille, Nord-Pas-de-Calais, France
    • Centre Hospitalier de Versailles
      • Department of Onco-Hematology
      Versailles, Île-de-France, France
  • 2002–2014
    • Unité Inserm U1077
      Caen, Lower Normandy, France
    • University of Helsinki
      Helsinki, Uusimaa, Finland
    • Centre Hospitalier Universitaire Rouen
      Rouen, Haute-Normandie, France
  • 2013
    • University of Leeds
      Leeds, England, United Kingdom
  • 2006–2013
    • University of Paris-Est
      La Haye-Descartes, Centre, France
    • University of Glasgow
      Glasgow, Scotland, United Kingdom
  • 2011
    • French Institute of Health and Medical Research
      Lutetia Parisorum, Île-de-France, France
  • 1992–2011
    • Université René Descartes - Paris 5
      Lutetia Parisorum, Île-de-France, France
    • Institut National de la Transfusion Sanguine, Paris
      Lutetia Parisorum, Île-de-France, France
  • 2010
    • Sunnybrook Health Sciences Centre
      • Department of Critical Care Medicine
      Toronto, Ontario, Canada
    • University of Limoges
      Limages, Limousin, France
  • 2003–2010
    • Assistance Publique – Hôpitaux de Paris
      Lutetia Parisorum, Île-de-France, France
    • Sapienza University of Rome
      Roma, Latium, Italy
    • Clinique Ambroise Paré
      Tolosa de Llenguadoc, Midi-Pyrénées, France
  • 2005–2009
    • Université Paris 13 Nord
      Île-de-France, France
    • Hôpital Henri Mondor (Hôpitaux Universitaires Henri Mondor)
      Créteil, Île-de-France, France
  • 2008
    • Institut de Cancérologie Gustave Roussy
      Villejuif, Île-de-France, France
  • 2007
    • University of Washington Seattle
      Seattle, Washington, United States
    • Hôpitaux Universitaires La Pitié salpêtrière - Charles Foix
      Lutetia Parisorum, Île-de-France, France
    • Centre Hospitalier Universitaire de Nancy
      Nancy, Lorraine, France
  • 2003–2007
    • Mercy Hospital St. Louis
      San Luis, Missouri, United States
  • 2001–2006
    • Université Paris-Est Créteil Val de Marne - Université Paris 12
      Créteil, Île-de-France, France
    • Centre Hospitalier Universitaire de Caen
      Caen, Lower Normandy, France
    • University of Hamburg
      Hamburg, Hamburg, Germany
    • American Hospital of Paris
      Lutetia Parisorum, Île-de-France, France
  • 1999–2003
    • Hôpital Bichat - Claude-Bernard (Hôpitaux Universitaires Paris Nord Val de Seine)
      • Service de Réanimation Médicale et des Maladies Infectieuses
      Lutetia Parisorum, Île-de-France, France
  • 2000–2002
    • Hôpital Antoine-Béclère – Hôpitaux universitaires Paris-Sud
      Clamart, Île-de-France, France
  • 1999–2002
    • Hôpital La Pitié Salpêtrière (Groupe Hospitalier "La Pitié Salpêtrière - Charles Foix")
      • Service d’Hépato - Gastro - Entérologie
      Lutetia Parisorum, Île-de-France, France
  • 1992–2002
    • Hôpital Paris Saint Joseph
      Lutetia Parisorum, Île-de-France, France
  • 1994–2000
    • Hôpital Raymond-Poincaré – Hôpitaux universitaires Paris Ile-de-France Ouest
      Île-de-France, France
  • 1998
    • Hôpital Cochin (Hôpitaux Universitaires Paris Centre)
      Lutetia Parisorum, Île-de-France, France
  • 1996
    • Hôtel-Dieu de Paris – Hôpitaux universitaires Paris Centre
      Lutetia Parisorum, Île-de-France, France