This is the accepted version of the journal article: Chetboun, Mikael; Raverdy, Violeta; Labreuche, Julien; [et al.]. «BMI and pneu- monia outcomes in critically ill COVID-19 patients : an international multicenter study». Obesity (Silver Spring, Md.), may 2021. DOI 10.1002/oby.23223 This version is available at https://ddd.uab.cat/record/245160 under the terms of the license Accepted Article DR. MIKAEL CHETBOUN (Orcid ID : 0000-0001-9870-4310) DR. VIOLETA RAVERDY (Orcid ID : 0000-0001-5754-2028) DR. ARTHUR SIMONNET (Orcid ID : 0000-0003-1529-3259) DR. CYRIELLE CAUSSY (Orcid ID : 0000-0001-8089-2907) DR. ELEFTHERIOS MYLONAKIS (Orcid ID : 0000-0002-4624-0777) DR. LEONIDAS PALAIODIMOS (Orcid ID : 0000-0003-4682-5991) DR. LUCA BUSETTO (Orcid ID : 0000-0003-4883-8980) Title: BMI and pneumonia outcomes in critically ill COVID-19 patients: an international multicenter study. Mikael CHETBOUN, M.D., 1,2 $; Violeta RAVERDY, M.D., 1,2 $; Julien LABREUCHE, BST 3; Arthur SIMONNET, M.D.-Ph.D., 4; Florent WALLET, M.D., 5,6; Cyrielle CAUSSY, M.D.-Ph.D., 7,8; Massimo ANTONELLI, M.D., 9; Antonio ARTIGAS, M.D.-Ph.D., 10,11; Gemma GOMA, Research Nurse, 12; Ferhat MEZIANI, M.D.-Ph.D., 13,14; Julie HELMS, M.D.-Ph.D., 13,15; Eleftherios MYLONAKIS, M.D.; 16,17 Mitchell M. LEVY, M.D., 17,18; Markos KALLIGEROS, M.D., 16,17; Nicola LATRONICO, M.D., 19,20; Simone PIVA, M.D., 19,20; Charles CERF, M.D., 21; Mathilde NEUVILLE, M.D., 21; Kada KLOUCHE, M.D.- Ph.D., 22,23; Romaric LARCHER, M.D., 22,23; Fabienne TAMION, M.D.-Ph.D., 24,25; Emilie OCCHIALI, M.D., 26; Morgane SNACKEN, M.D., 27; Jean-Charles PREISER, M.D.-Ph.D. 27; Loay KONTAR, M.D., 28; Antoine RIVIERE, M.D., 29; Stein SILVA, M.D.-Ph.D., 30,31; Benjamine SARTON, M.D.-Ms.C., 30,31; Raphael KROUCHI, M.D., 32; Victoria DUBAR, 32; Leonidas PALAIODIMOS, M.D., 33; Dimitrios KARAMANIS, Ph.D. 34; Juliette PERCHE, M.D., 35; Erwan L'HER, M.D.-Ph.D., 36,37; Luca BUSETTO, M.D., 38; Dror DICKER, M.D., 39,40; Shaul LEV, M.D., 40,41; Alain DUHAMEL, M.D., 3; Mercè JOURDAIN, M.D.-Ph.D., 1,4 *; François PATTOU, M.D., 1,2 * $ : co-first authors; * : co-last authors; This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1002/oby.23223 This article is protected by copyright. All rights reserved On behalf of the "LICORNE Lille Intensive Care COVID-19 and Obesity Study Group", Accepted Article Lille, France; the "COVID-19 Toulouse ICU Network", Toulouse, France; the "Hasharon Hospital-Rabin Medical Center Covid-19 and Obesity Study Group", Petach-Tikva, Israel; the "ID and ICU Covid-19 Study Group", Providence, RI, USA; the "Rouen Covid-19 and Obesity Study", Rouen, France; the "COVID-O-HCL Consortium", Lyon, France; the "MICU Lapeyronie", Montpellier, France; the "Bronx COVID-19 and Obesity Study Group", Bronx, NY, USA; the "MIR Amiens Covid19", Amiens, France; the "Strasbourg NHC", Strasbourg, France; the "Foch COVID-19 Study Group", Suresnes, France. See complete list of co-authors in Online data supplement. Corresponding authors: - Prof. François PATTOU Address: Centre Hospitalier Universitaire de Lille, Chirurgie Génerale et Endocrinienne, Centre Integré de l'Obésité, Hôpital Huriez, 1 rue Polonovski, 59000 Lille, France Tel: +333 20 44 42 73 - Fax: +333 20 44 47 58 Email: francois.pattou@univ-lille.fr - Prof. Mercè JOURDAIN Address: Centre Hospitalier Universitaire de Lille, Pôle Réanimation, Hôpital Roger Salengro, avenue du Professeur Emile Laine, 59000 Lille, France Tel: +33 3 20 44 40 84 E-mail: mercedes.jourdain@univ-lille.fr running title: BMI and COVID-19 pneumonia type: Original article / subject category list: Clinical Trials and Investigations TRIAL REGISTRATION: ClinicalTrials.gov Identifier: NCT04391738 Key words: COVID-19, Obesity, Critical Care, Risk Factors, Prevention This article has online data supplement Affiliations 1 Univ Lille, Inserm, Lille Pasteur Institute, European Genomic Institute for Diabetes, U1190, F-59000, Lille, France This article is protected by copyright. All rights reserved 2 CHU Lille, Centre Intégré de l'Obésité, F-59000, Lille, France Accepted Article 3 Univ. Lille, CHU Lille, ULR 2694 - METRICS: Évaluation des technologies de santé et des pratiques médicales, F-59000 Lille, France 4 CHU Lille, Pôle de Réanimation, Hôpital Roger Salengro, F-59000, Lille, France 5 Hospices Civils de Lyon, Medical Intensive Care Unit, Hôpital Lyon Sud, 69495 Pierre-Bénite, France 6 Laboratoire des Pathogènes Emergents-Fondation Mérieux, Centre International de Recherche en Infectiologie, Inserm U1111, Lyon, France 7 Hospices Civils de Lyon, Département Endocrinologie, Diabète et Nutrition, Hôpital Lyon Sud, 69495 Pierre-Bénite, France 8 Univ Lyon, CarMen Laboratory, INSERM, INRA, INSA Lyon, Université Claude Bernard Lyon 1,69495 Pierre-Bénite, France 9 Department of Anesthesia and Intensive Care Medicine, Fondazione Policlinico Universitario A. Gemelli IRCCS, Rome Italy I 00168 10 Corporació Sanitaria Universitaria Parc Tauli , Critical Care Department, CIBER Enfermedades Respiratorias , Sabadell , 08208 , Spain 11 Autonomous University of Barcelona , Department of Medecine , Bellaterra , Spain 12 Corporació Sanitaria Universitaria Parc Tauli , Critical Care Department, Sabadell, 08208 , Spain 13 Hôpitaux Universitaires de Strasbourg, Service de Médecine Intensive Réanimation, Nouvel Hôpital Civil, Strasbourg, France 14 Inserm, UMR 1260, Regenerative Nanomedicine (RNM), FMTS, Strasbourg, France 15 Inserm UMR_S1109, LabEx TRANSPLANTEX, Centre de Recherche d'Immunologie et d'Hématologie, Faculté de Médecine, FHU OMICARE, FMTS, Université de Strasbourg (UNISTRA), Strasbourg, France 16 Infectious Diseases Division, Warren Alpert Medical School of Brown University, Providence, RI, 02903, USA 17 Rhode Island Hospital and The Miriam Hospital, Providence, RI, 02903, USA 18 Division of Critical Care, Pulmonary, and Sleep Medicine,, Warren Alpert Medical School of Brown University, Providence, RI, 02903, USA 19 Department of Medical and Surgical Specialties, Radiological Sciences and Public Health, University of Brescia 20 Department of Anesthesia, Critical Care and Emergency, Spedali Civili University Hospital, Brescia, Italy This article is protected by copyright. All rights reserved 21 Hôpital Foch, Medico-surgical ICU, Suresnes, 92150, France Accepted Article 22 Department of Intensive Care Medicine, Lapeyronie University Hospital, Montpellier, France 23 PhyMedExp, University of Montpellier, INSERM, CNRS, Montpellier, France 24 Medical Intensive Care Unit, University Hospital, 76000 Rouen, France 25 Normandie Univ, UNIROUEN, Inserm U1096, FHU- REMOD-VHF, 76000 Rouen, France 26 Department of Anesthesia and Critical Care, Rouen University Hospital, Rouen, 76031 cedex, France 27 Erasme Hospital, Université Libre de Bruxelles, Department of Intensive Care, B- 1070 Brussels, Belgium 28 Medical ICU Amiens University Hospital, Amiens, 80054, France 29 Abbeville General Hospital, Intensive Care Unit, Abbeville, 80100, France 30 Critical Care Unit, University Hospital of Purpan, Toulouse, 31300 France 31 Paul Sabatier University, ToNIC, Toulouse NeuroImaging Center, Inserm1214, Toulouse, 27091, France 32 Intensive Care Unit, Centre hospitalier de Dunkerque, 59240 Dunkerque 33 Division of Hospital Medicine, Montefiore Medical Center, Albert Einstein College of Medicine, Bronx 10467, New York, USA 34 Department of Economics, University of Piraeus, Piraeus 18534, Attica, Greece 35 Centre Hospitalier de Roubaix, Réanimation Polyvalente, Roubaix, 59100, France 36 Médecine Intensive et Réanimation, CHRU de Brest La Cavale Blanche, Brest, France 37 LATIM Inserm UMR 1101, Université de Bretagne Occidentale 38 Department of Medicine, University of Padova, Padova. 35128, Italy 39 Internal Medicine D, Hasharon Hospital-Rabin Medical Center, Petach-Tikva, Israel 40 Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel 41 General Intensive Care Unit, Hasharon Hospital-Rabin Medical Center, Petach-Tikva, Israel Author Contributions: FP and MJ had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. This article is protected by copyright. All rights reserved Concept and design: MJ, FP Accepted Article Acquisition, analysis, or interpretation of data: MC, VR, JL, AD, MJ, FP Drafting of the manuscript: MC, VR, MJ, FP Critical revision of the manuscript for important intellectual content: All authors. Statistical analysis: JL, AD Obtained funding: no specific funding Administrative, technical, or material support: NA Supervision: FP and MJ Conflict of Interest Disclosures The authors declared no conflict of interest Funding/Support: no specific funding Manuscript word count: 3721 4 tables and 2 figures in the main text Online-only supplement: 3 supplemental eTables; 3 supplemental eFigures; Study Group of Collaborators Study Importance What is already known? - Early evidence has linked the overall severity of COVID-19 with older age, male sex, and preexisting chronic conditions such as cancer, and chronic renal, liver, respiratory, and cardiovascular disease. - Some studies also found an overall association of disease severity with obesity, and other metabolic risk factors, including diabetes, hypertension, and smoking. - A wider geographic representation, across centers with variable prevalence of obesity, is needed. This article is protected by copyright. All rights reserved - Accepted Article What does this study add? - We showed a significant relation between BMI and invasive mechanical ventilation (IMV): adjusted OR 1.27 (95% CI, 1.12-1.45) in the whole cohort, and 1.65 (95% CI, 0.97-2.79) per 5 kg/m2 in females under 50 years. - Adjusted Cox regression model showed a significant association between BMI and 28-day all-cause mortality, which was only increased in obesity class III (≥40 kg/m2) (adjusted HR 1.68 (95% CI 1.06-2.64). How might these results change the focus of clinical practice? - We observed a linear association between BMI and the need for IMV in ICU COVID-19 patients, across multiple countries, independent of other metabolic risk factors, and a non-linear association between BMI and mortality risk. - These findings will help to determine the risk of severe COVID-19 pneumonia, in all BMI categories, in order to provide clear guidance for specific prevention, including vaccination in patients at highest risk. Abstract Background: Previous studies unveiled a relation between the severity of COVID-19 pneumonia and obesity. The aims of this multicenter retrospective cohort study were to disentangle the association of BMI and associated metabolic risk factors (diabetes, This article is protected by copyright. All rights reserved hypertension, hyperlipidemia, current smoking) in critically ill patients with COVID-19. Accepted Article Methods: This multicenter retrospective cohort study enrolled patients admitted in intensive care for COVID-19, in 21 centers (Europe, Israel, USA) between 02/19/2020 and 05/19/2020. Primary and secondary outcomes were the need for invasive mechanical ventilation (IMV), and 28-day mortality. Results: A total of 1,461 patients were enrolled, median(IQR) age was 64 years (40.9-72.0); 73.2% males; BMI 28.1 kg/m2 (25.4-32.3); 1,080 patients (73.9%) required IMV; the 28-day mortality estimate was 36.1% (95%CI, 33.0-39.5). Adjusted mixed logistic regression model showed a significant linear relation between BMI and IMV: OR 1.27 (95%CI, 1.12-1.45) per 5 kg/m2. Adjusted Cox proportional hazards regression model showed a significant association between BMI and mortality, which was only increased in obesity class III (≥40 kg/m2) (HR 1.68 (95%CI 1.06-2.64). Conclusion: In critically ill COVID-19 patients, we observed a linear association between BMI and the need for IMV, independent of other metabolic risk factors, and a non-linear association between BMI and mortality risk. (NCT04391738). Introduction Eighteen months after the outbreak of severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2), the coronavirus disease 2019 (COVID-19) pandemic has expanded globally, having already affected more than 150 million individuals worldwide and claimed more than 3 millions lives (1). Identifying risk factors of worse outcomes is essential for reducing the future overall burden of COVID-19 on health system and enforcing prevention efforts in populations at higher risk. Early evidence has linked the overall severity of COVID-19 with older age, male sex, and preexisting chronic conditions such as cancer, and chronic renal, liver, respiratory, and cardiovascular disease (2–4). Numerous studies also found an overall This article is protected by copyright. All rights reserved association of disease severity with obesity (5–9), and other metabolic risk factors, including Accepted Article diabetes (10, 2, 5, 11), hypertension (3), and smoking (12). One distinct feature of COVID-19 is pneumonia and the frequent need for invasive mechanical ventilation (IMV), resulting in great strain on intensive care resources worldwide (13). Simonnet et al. (14) reported that the need for IMV gradually increased with body mass index (BMI) in COVID-19 patients admitted in intensive care units (ICU). The association between obesity and the need for IMV was also found statistically significantly in four independent studies (15–18), in contrast with other studies (19–21). Some reports suggested that the relation between BMI and COVID-19 severity might be restricted to younger patients (22–25). Based on existing data, public health agencies in Europe and the USA have issued guidelines that include obesity as an increased risk factor of severe COVID-19 (26–28). Likewise, the COVID-19 pandemic has already impacted people living with obesity in multiple ways, ranging from food shortages and insecurity, reduced physical activity during lockdown, anxiety from cancellation of care, and mental health issues compounded by isolation (29). More global data is therefore urgently needed to determine the risk of severe COVID-19 pneumonia, across all BMI categories, in order to provide clear guidance and inform patient care (30). A wider geographic representation, across centers and countries with variable prevalence of obesity, is also needed to ensure the validity and generalizability of the findings. This multicenter international retrospective cohort study was designed to examine the relationship between BMI and COVID-19 pneumonia severity, as defined by the need for IMV (primary outcome) and the 28-day all-cause mortality rate (secondary outcome) among patients admitted in ICU. The study secondary objectives were to disentangle the influence of BMI from other metabolic risk factors, such as diabetes, hypertension, dyslipidemia, and current smoking, as well as to investigate the modifying effects of age and sex on this relationship. This article is protected by copyright. All rights reserved Methods Accepted Article The BMI-SARS-CoV-2 study was a multicenter, international, retrospective cohort study designed to investigate the relationship between BMI and the need of IMV, among adult patients admitted in ICU for COVID-19 (ClinicalTrials.gov Identifier: NCT04391738, sponsored by the Lille University Hospital, France). The study complied with standard operating procedures in place, in accordance with the European Data Protection Directive (95/46/EC) and, upon its entry into force, Regulation (EU) 2016/679 (also referred to as the General Data Protection Regulation (GDPR)), and French CNIL frameworks n° MR004 regarding the processing of personal data in clinical studies. The institutional review board from other centers in USA and Israel approved the retrospective case series as minimal-risk research using data collected for routine clinical practice and waived the requirement for informed consent. The study report followed the Strengthening Reporting of Observational Studies in Epidemiology (STROBE) guidelines. Data Sources Centers that routinely registered BMI in patients admitted in ICU for COVID-19 during the study period were identified among a preexisting network of ICUs in France (French Network of Simulation in Intensive Care) or through the published literature. The Medline database (https://pubmed.ncbi.nlm.nih.gov) and the Clarivate Web of Science database (https://webofknowledge.com) were searched using the following keywords: intensive care, COVID-19, SARS-CoV-2, obesity, body mass index to identify cohort studies that reported BMI in patients admitted in ICU for COVID-19. Overall, 27 centers from seven countries (France, China, Belgium, USA, Israel, Italy, and Spain), were identified and contacted, and 21 agreed to participate as follows: France (13), Italy (3), USA (2), Israel (1), Belgium (1), and Spain (1). In order to minimize data input errors or possible bias, each participating center received a standardized template, based on the registered study design and accompanied by a glossary of required variables and a data entry support guide, in order to enforce the homogeneity and validity of the data and to lower the number of missing values. At each center, investigators extracted primary data from medical records from all consecutive patients admitted in ICU for COVID-19, between February19th 2020 and May 11th, 2020. Data cleaning involved repeated cycles of screening, diagnosing, treatment, and documentation of this process. Finally, each center removed all potential identifiers from the dataset, which was assembled in random order, protected with a password, prior to being sent This article is protected by copyright. All rights reserved to the study sponsor, in accordance to the data transfer agreement. Individual data were Accepted Article eventually aggregated by the sponsor center, in random order, with an inclusion number corresponding to each center. Study Patients and Covariates Participants were patients admitted in ICU for confirmed COVID-19 related pneumonia with acute respiratory distress syndrome, as diagnosed on the basis of WHO guidance (31). SARS- CoV-2 infection was defined as a positive result on real-time reverse transcriptase– polymerase chain reaction assay of nasal or pharyngeal swab specimens, as previously described (32). All primary data were reviewed and collected by trained physicians. The variables collected at the time of admission included sex, age, height, and body weight, measured or estimated by a physician, as well as pre-specified metabolic risk factors such as current smoking status and history of diabetes, hypertension, and dyslipidemia. Other associated comorbidities included cardiovascular disease (including chronic heart disease, cerebrovascular disease, and peripheral arterial disease), chronic obstructive pulmonary disease, and immunodeficiency (including steroid use, pre-existing immunological condition, or current chemotherapy in individuals with cancer). Current or previous history of cancer and chronic kidney disease were also collected, as well as the use of angiotensin converting enzyme inhibitors, angiotensin receptor blockers, and antidiabetic treatment, with or without insulin. Exposure of Interest: The exposure of interest was BMI, defined as the weight in kilograms divided by the square of height in meters, and measured at the time of admission in ICU. BMI was analyzed as a continuous variable, or classified in BMI categories as defined by the World Health Organization (33): underweight (under 18.5 kg/m2); normal weight (18.5 to 24.9 kg/m2); overweight (25 to 29.9 kg/m2); obesity class I (30 to 34.9 kg/m2); obesity class II (35 to 39.9 kg/m2); and obesity class III (40 kg/m2 and above). In some subgroup analyses, underweight and normal weight categories, on one hand, and obesity class I, II and III categories, on the other hand, were combined as lean (under 25 kg/m2), and obesity (30 kg/m2 and above), respectively, in order to gather sufficient numbers of patients. Outcomes and Follow-up This article is protected by copyright. All rights reserved The primary outcome was the need for IMV prior to, or after, ICU admission. The secondary Accepted Article outcome was all-cause mortality rate within 28 days following ICU admission. Data were collected from February 19th to May 19th, 2020. The following events were collected: the date of ICU admission, the number of days between admission and intubation (0 day means that IMV occurred the day of admission), and the number of days spent under IMV (1 day means that the patient was extubated after a day spent under IMV). When the patient was tracheotomized, the return to spontaneous ventilation with ambient air was considered as an extubation event. Patient status at last news (deceased, discharged from, or still hospitalized in ICU) and the number of days that elapsed since ICU admission was also collected. Statistical Analysis Having examined histograms, all quantitative variables were summarized by median and quartiles, and groups compared using the Mann-Whitney U-test. Categorical variables were expressed as numbers (percentage) and compared by Chi-square test with the use of Yates' continuity correction. The association of BMI with the need of IMV in patients with COVID-19 admitted in intensive care was assessed by using a mixed logistic regression model, including center as random effect. BMI was analyzed as a categorical variable using modified WHO classification (see exposure description) and as a continuous variable. Odds ratio (OR) was calculated using the lean category (BMI < 25 kg/m2) as reference, or per 5 kg/m2 increase in BMI. The log-linearity assumption was examined using restricted cubic spline functions. A multivariable analysis using a mixed logistic regression model was performed to adjust the association between BMI and IMV on predefined confounding variables, including age, sex, and the presence of pre-specified metabolic risk factors (diabetes, hypertension, dyslipidemia, and current smoking). Finally, in a secondary and exploratory analysis, we investigated the heterogeneity in the association of BMI and IMV into the mixed logistic regression models according to sex (male vs. female) and age subgroups (< 50 years, 50 to 74.9 years and ≥ 75 years). The association between BMI and 28-day all-cause mortality was assessed by using a Cox proportional hazard regression models with a random center effect (frailty model) with and without predefined confounding variables. Similarly, to previous analyses, BMI was analyzed as a categorical variable using modified WHO classification (see exposure description) and This article is protected by copyright. All rights reserved continuous variable. Since the log-linearity assumption was not satisfied, only results for BMI Accepted Article treated as categorical variable were reported. Hazard Ratio (HR) were calculated using the lean category (BMI < 25 kg/m2) as reference. Primary analyses for both outcomes were done after handling missing values by multiple imputations. Imputation procedure was performed using a regression switching approach (chained equations with m=10 imputations) under missing at random assumption using all patient characteristics with a predictive mean matching method for quantitative variables and a logistic regression model (binary, ordinal, or multinomial) for categorical variables. Regression estimates obtained in the different imputed datasets were combined using Rubin’s rules. An available-case sensitivity analysis was also performed. Statistical testing was done at the two-tailed α level of 0.05. Data were analyzed using the SAS software package, release 9.4 (SAS Institute, Cary, NC). This article is protected by copyright. All rights reserved Results Accepted Article Characteristics of Study Population A total of 1,461 individuals admitted in ICU with confirmed COVID-19 - in 21 institutions from 6 countries: France (13), Italy (3), USA (2), Israel (1), Belgium (1), and Spain (1), were included in this study (See study flow chart, Figure S1). Table 1 details the characteristics of the participants who required IMV, and those who did not, as well as the prevalence of pre- existing chronic conditions and treatments at the time of admission. Overall, study participants were predominantly males (73.2%), with an age varying from 19 to 93 years, and a median (IQR) of 64 (56-73) years. The median (IQR) BMI was 28.1 (25.4-32.3) kg/m2, with an overall prevalence of obesity (BMI above 30 kg/m2) of 37.5%. The study participants had markedly higher BMI than those observed, after adjustment for sex and age, in the general population of the corresponding country (Figure S2). Half of participants had hypertension (51.5%), while the prevalence of diabetes and dyslipidemia were 29.2% and 29.0%, respectively. Overall, 1,080 (73.9%) participants required IMV, after a median (IQR) time of 0 days (0-1) following ICU admission. As expected, patients who required IMV had a higher initial severity score than those who did not require IMV. The patients who required IMV were also predominantly males (76.3%), had a higher BMI, and were more frequently current smokers (Table 1). Patient characteristics and data collected in each of the 21 centers are detailed in Figure S3. At the time of analysis (May 19th, 2020), 903 (61.8%) patients had been discharged alive, 385 (26.4%) had died following a median (IQR) 11 days (6-18) in the ICU, and 173 (11.8%) were still hospitalized in ICU. The median (IQR) follow-up period of observation following ICU admission was 13 days (7-26) in the overall population, and 15 days (7-29) among survivors. Among 1,080 patients who required IMV, 633 (58.6%) patients had been extubated alive at the time of analysis, after a median (IQR) duration of IMV of 14 days (8-22). Association between BMI and the Need for IMV In a multivariable mixed logistic regression model adjusted on center and/or age, sex, and pre- specified metabolic risk factors (diabetes, hypertension, dyslipidemia, and current smoking), BMI was statistically significantly associated with the need for IMV (Table 2). Moreover, the relation between BMI and the risk of the need for IMV was linear, as illustrated by the This article is protected by copyright. All rights reserved gradual increase of OR with each BMI category, reaching 3.06 (1.53 to 6.10) in patients with Accepted Article class III obesity (BMI ≥ 40 kg/m2) (Figure 1a). The relation between BMI and the need for IMV was further confirmed in a sensitivity analysis limited to available cases (Table S1). In addition, we observed that older age and male sex were independent predictors of the need for IMV, in contrast to diabetes, hypertension, dyslipidemia, and current smoking (Table 3 and Table S2). The heterogeneity of the association between BMI and IMV with age was studied in key subgroups exploratory analysis, on the overall study population and on the complete cases population (Figure 2). As illustrated, the association of BMI with IMV was more pronounced in females under 50 years, with an adjusted OR of 1.65 (95%CI, 0.97 to 2.79) in primary analysis and 1.86 (95%CI, 1.67 to 3.22) in sensitivity analysis. Association of BMI with 28-day Mortality. The overall estimate of 28-day mortality was 36.1% (95%CI, 33.0 to 39.5%). As shown in table 2, we found a non-linear relationship between BMI categories and 28-day mortality, with an increased mortality risk only for patients with severe or class III obesity (≥ 40 kg/m2) (Figure 1b). The non-linearity of the relation between BMI and 28-day mortality was further confirmed in a sensitivity analysis (Table S1). In data driven model, considering BMI as binary variable (≥ 40 kg/m2 vs. < 40 kg/m2), obesity class III remained an independent predictor of mortality (HR=1.84; 95% CI 1.23 to 2.75) in addition to age (HR per 10-year increase, 1.74; 95% CI 1.54 to 1.95) (Table 4). These results were further confirmed in a sensitivity analysis (Table S3). This article is protected by copyright. All rights reserved Discussion Accepted Article To our knowledge, this multicenter cohort study represents the first international collaborative effort to explore the association of BMI with the outcomes of pneumonia among Covid-19 patients admitted in ICU. Our main finding was a linear correlation between BMI and the need for IMV, after adjustment on center, age, sex, and other pre-specified metabolic risk factors. Of note, we observed in age-sex subgroups analyses that the relation between BMI and the need for IMV, was more pronounced in females than in males under 50 years (Figure 2). An influence of age on the relation between obesity and COVID-19 severity has been previously suggested, (22–24), but one that existed regardless of sex. A milder severity of COVID-19 has been previously reported in pre-menopausal women (34). In line with our findings, obesity was also the variable most associated with COVID-19 disease severity in pregnant women in a recent report (35). Of note estrogen are classically playing a protective role in women (36), and Estradiol levels decreased with overweight and/or obesity in pre- menopausal women (36, 37). However, the observational nature of our study cannot address the complexity of this question. As expected, we also observed an overall association of the need for IMV with older age and male sex (11-19). However, we did not unveil any independent relation between the need of IMV and current smoking, nor with any other pre- specified metabolic risk factors. A distinct association of diabetes with the overall severity of COVID-19 had been suggested in previous reports (4, 16, 38). However, one study did not include BMI among the variables analyzed (38), while, in the two others, obesity was the main driver of the association between diabetes and the need for IMV (4, 16). The second original finding of our study was the non-linear relation observed between BMI and the 28-day all-cause mortality rate in patients admitted in ICU. When fully adjusted on center, age, sex and pre-specified metabolic risk factors, obesity class III (BMI ≥ 40 kg/m2) was associated with a 68% increase in mortality, as compared with lean patients (BMI <25 kg/m2). On the other hand, mortality risk was not increased in patients with overweight or obesity class I and II (BMI between 25 kg/m2 and 39.9 kg/m2). This seemingly paradoxical relationship between mortality and BMI categories has not yet been reported in COVID-19 patients (5, 6), but echoes the “obesity survival paradox,” generally observed in critically ill patients (39), with overweight and moderate obesity being protective as compared with lean BMI or normal BMI or more severe obesity (40) ici . Further research should focus on This article is protected by copyright. All rights reserved identifying the underlying pathophysiologic mechanisms of the “obesity survival paradox” in Accepted Article COVID-19. As expected, the all-cause mortality was also closely linked with age, with a 74% increase per ten-years. Of note, male sex and diabetes were only marginally associated with mortality in the present study, which enrolled only patients admitted in ICU, in contrast to other multicenter studies conducted in hospitalized patients (6, 38, 41) or in the general population (5). Strengths The main strength of the present study was the large number of participants, across all BMI categories, and both sexes, which allowed us to perform, all planned analyses, with BMI as exposure, and association with pre-specified metabolic risk factors. Second was the choice of IMV as the study primary outcome. As opposed to other less specific endpoints like hospital or ICU admission, all-cause mortality, or composite severity end points (6, 10, 20), IMV allowed us to specifically explore the association between BMI and severity of pneumonia in COVID-19. Finally, the wide geographic representation of participating centers provides insight related to the generalizability of effect across regions and countries with variable prevalence of obesity, thus improving the validity of our findings and enhancing their general relevance. Limitations The main limitation of our study lies in its retrospective nature. Standard clinical care may have varied between centers, and participants received various treatments which were not considered in our analyses. Second, the data from patients who remained hospitalized at the final study date (11.8%) were censored, and they may have led to an underestimation of the outcomes. This risk appears limited, however, since 98.7% of patients requiring IMV did so within 7 days from admission, and since, at the time of analysis, 95.8% of patients had been already discharged alive or had been hospitalized for more than 28 days. Third, the multicenter design of this international study creates a complex confounding structure. Results were not adjusted for other potential confounders, including race, ethnicity, socio- economic status indicators, and comorbidities such as chronic kidney, cardiovascular, and respiratory diseases. Finally, we did not enrolled participants from Africa, South America, or Asia, in whom different BMI cut points may have been more relevant. This may limit the generalizability of our findings which remain to be confirmed more globally. This article is protected by copyright. All rights reserved Conclusion Accepted Article Taken together, the data of this international multicenter cohort study provide direct evidence of the independent association between obesity and the severity of pneumonia in COVID-19. We first observed a overall linear relation of BMI with the need of IMV, which was most pronounced in females under the age of 50. Second, we observed a non-linear relation of BMI with 28-day all-cause mortality, which was increased in patients with severe obesity (BMI ≥ 40 kg/m2). With the ongoing pandemic and the obesity epidemic feeding each other (42), this close association between BMI and the severity of COVID-19 pneumonia should foster more drastic measures to limit the risk of COVID-19 infections in patients with obesity. The evidence presented here may also inform physio-pathological research to elucidate the relation between obesity and severe lung damage in COVID-19. This article is protected by copyright. All rights reserved Acknowledgment: Accepted Article Authors thank Mr Maxime Caillier, Mr Deraedt and Mr Amhed Bouzidi for their management of regulatory aspects of the study. Role of the Sponsor: The sponsors were represented by the corresponding authors, Francois Pattou and Merce Jourdain. The sponsors had a role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and review and approval to submit the manuscript for publication. The sponsors did not have the right to veto submission to any particular journal, but did participate in the writing group’s discussion when selecting an appropriate journal for submission. The corresponding authors had the final say in submitting the manuscript for publication. This article is protected by copyright. All rights reserved Accepted Article References: 1. Anon. (2020). WHO Coronavirus Disease (COVID-19) Dashboard. [WWW document]. URL https://covid19.who.int 2. Zhou F, Yu T, Du R, et al. Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study. The Lancet 2020;395:1054–1062. 3. Grasselli G, Zangrillo A, Zanella A, et al. Baseline Characteristics and Outcomes of 1591 Patients Infected With SARS-CoV-2 Admitted to ICUs of the Lombardy Region, Italy. JAMA 2020. 4. Richardson S, Hirsch JS, Narasimhan M, et al. Presenting Characteristics, Comorbidities, and Outcomes Among 5700 Patients Hospitalized With COVID-19 in the New York City Area. JAMA 2020;323:2052–2059. 5. Williamson EJ, Walker AJ, Bhaskaran K, et al. OpenSAFELY: factors associated with COVID-19 death in 17 million patients. Nature 2020:1–11. 6. Docherty AB, Harrison EM, Green CA, et al. Features of 20 133 UK patients in hospital with covid-19 using the ISARIC WHO Clinical Characterisation Protocol: prospective observational cohort study. BMJ 2020;369. This article is protected by copyright. All rights reserved 7. Caussy C, Pattou F, Wallet F, et al. Prevalence of obesity among adult inpatients with Accepted Article COVID-19 in France. The Lancet Diabetes & Endocrinology 2020;8:562–564. 8. Petrilli CM, Jones SA, Yang J, et al. Factors associated with hospital admission and critical illness among 5279 people with coronavirus disease 2019 in New York City: prospective cohort study. BMJ 2020;369. 9. Gao M, Piernas C, Astbury NM, et al. Associations between body-mass index and COVID-19 severity in 6·9 million people in England: a prospective, community-based, cohort study. The Lancet Diabetes & Endocrinology 2021;0. 10. Cariou B, Hadjadj S, Wargny M, et al. Phenotypic characteristics and prognosis of inpatients with COVID-19 and diabetes: the CORONADO study. Diabetologia 2020;63:1500–1515. 11. Deng S-Q, Peng H-J. Characteristics of and Public Health Responses to the Coronavirus Disease 2019 Outbreak in China. Journal of Clinical Medicine 2020;9:575. 12. Zhao Q, Meng M, Kumar R, et al. The impact of COPD and smoking history on the severity of COVID-19: A systemic review and meta-analysis. Journal of Medical Virology n/a. 13. Truog RD, Mitchell C, Daley GQ. The Toughest Triage — Allocating Ventilators in a Pandemic. New England Journal of Medicine 2020. 14. Simonnet A, Chetboun M, Poissy J, et al. High Prevalence of Obesity in Severe Acute Respiratory Syndrome Coronavirus-2 (SARS-CoV-2) Requiring Invasive Mechanical Ventilation. Obesity 2020;28:1195–1199. 15. Goyal P, Choi JJ, Pinheiro LC, et al. Clinical Characteristics of Covid-19 in New York City. New England Journal of Medicine 2020;382:2372–2374. 16. Bello-Chavolla OY, Bahena-López JP, Antonio-Villa NE, et al. Predicting Mortality Due to SARS-CoV-2: A Mechanistic Score Relating Obesity and Diabetes to COVID-19 Outcomes in Mexico. J Clin Endocrinol Metab 2020;105. 17. Kalligeros M, Shehadeh F, Mylona EK, et al. Association of Obesity with Disease Severity Among Patients with Coronavirus Disease 2019. Obesity 2020;28:1200–1204. 18. Palaiodimos L, Kokkinidis DG, Li W, et al. Severe obesity, increasing age and male sex are independently associated with worse in-hospital outcomes, and higher in-hospital mortality, in a cohort of patients with COVID-19 in the Bronx, New York. Metabolism 2020;108:154262. 19. Busetto L, Bettini S, Fabris R, et al. Obesity and COVID-19: an Italian snapshot. Obesity n/a. This article is protected by copyright. All rights reserved 20. Cai Q, Chen F, Wang T, et al. Obesity and COVID-19 Severity in a Designated Accepted Article Hospital in Shenzhen, China. Diabetes Care 2020. 21. Chao JY, Derespina KR, Herold BC, et al. Clinical Characteristics and Outcomes of Hospitalized and Critically Ill Children and Adolescents with Coronavirus Disease 2019 (COVID-19) at a Tertiary Care Medical Center in New York City. The Journal of Pediatrics 2020;0. 22. Buscemi S, Buscemi C, Batsis JA. There is a Relationship Between Obesity and Coronavirus Disease 2019 but More Information is Needed. Obesity n/a. 23. Kass DA, Duggal P, Cingolani O. Obesity could shift severe COVID-19 disease to younger ages. Lancet 2020;395:1544–1545. 24. Lighter J, Phillips M, Hochman S, et al. Obesity in patients younger than 60 years is a risk factor for Covid-19 hospital admission. Clin Infect Dis. 25. Ong SWX, Young BE, Leo Y-S, Lye DC. Association of higher body mass index (BMI) with severe coronavirus disease 2019 (COVID-19) in younger patients. Clin Infect Dis. 26. CDC. (2020). CDC Overweight & Obesity. [WWW document]. URL https://www.cdc.gov/obesity/index.html 27. DGOS_Michel.C, DGOS_Michel.C. (2020). Obésité et Covid-19. [WWW document]. URL https://solidarites-sante.gouv.fr/soins-et-maladies/prises-en-charge- specialisees/obesite/article/obesite-et-covid-19 28. Anon. (2020). COVID-19: infection prevention and control (IPC) - GOV.UK. [WWW document]. URL https://www.gov.uk/government/publications/wuhan-novel-coronavirus- infection-prevention-and-control 29. Anon. (2020). Obesity and COVID-19: Policy statement. [WWW document]. URL https://www.worldobesity.org/news/obesity-and-covid-19-policy-statement 30. The Lancet Diabetes & Endocrinology. COVID-19: underlying metabolic health in the spotlight. Lancet Diabetes Endocrinol 2020;8:457. 31. Anon. (2020). Clinical management of severe acute respiratory infection when COVID-19 is suspected. [WWW document]. URL https://www.who.int/publications- detail/clinical-management-of-severe-acute-respiratory-infection-when-novel-coronavirus- (ncov)-infection-is-suspected 32. Huang C, Wang Y, Li X, et al. Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China. The Lancet 2020;395:497–506. 33. Anon. (2020). Wlorld Health organization. Overweight and obesity. [WWW document]. URL This article is protected by copyright. All rights reserved http://www.who.int/gho/ncd/risk_factors/overweight_obesity/obesity_adults/en/ Accepted Article 34. Ding T, Zhang J, Wang T, et al. Potential Influence of Menstrual Status and Sex Hormones on female SARS-CoV-2 Infection: A Cross-sectional Study from Multicentre in Wuhan, China. Clin Infect Dis. 35. Kayem G, Lecarpentier E, Deruelle P, et al. A snapshot of the Covid-19 pandemic among pregnant women in France. Journal of Gynecology Obstetrics and Human Reproduction 2020:101826. 36. Freeman EW, Sammel MD, Lin H, Gracia CR. Obesity and reproductive hormone levels in the transition to menopause. Menopause 2010;17:718–726. 37. Li Y, Jerkic M, Slutsky AS, Zhang H. Molecular mechanisms of sex bias differences in COVID-19 mortality. Critical Care 2020;24:405. 38. Zhu L, She Z-G, Cheng X, et al. Association of Blood Glucose Control and Outcomes in Patients with COVID-19 and Pre-existing Type 2 Diabetes. Cell Metabolism 2020;31:1068-1077.e3. 39. Schetz M, De Jong A, Deane AM, et al. Obesity in the critically ill: a narrative review. Intensive Care Med 2019;45:757–769. 40. Acharya P, Upadhyay L, Qavi A, et al. The paradox prevails: Outcomes are better in critically ill obese patients regardless of the comorbidity burden. Journal of Critical Care 2019;53:25–31. 41. Rottoli M, Bernante P, Belvedere A, et al. How important is obesity as a risk factor for respiratory failure, intensive care admission and death in hospitalised COVID-19 patients? Results from a single Italian centre. European Journal of Endocrinology 2020;1. 42. Dicker D, Bettini S, Farpour-Lambert N, et al. Obesity and COVID-19: The Two Sides of the Coin. OFA 2020:1–9. This article is protected by copyright. All rights reserved Accepted Article Tables This article is protected by copyright. All rights reserved Accepted Article Table 1. Characteristics at baseline of patients admitted in ICU, in the whole cohort (n=1461) and in patients who did (n=1080) or did not require invasive mechanical ventilation (n=381) All patients Non IMV P- N IMV (n=1080) (n=1461) (n=381) valuea Characteristics at admission Male sex, No. (%) 1461 1070 (73.2) 246 (64.6) 824 (76.3) < 0.001 Age, median (IQR), y 1461 64 (56-73) 62 (53-75) 65 (56-72) 0.48 Age by classes, No. (%) 1461 < 50 y 210 (14.4) 74 (19.4) 136 (12.6) < 0.001 50 to 75 y 968 (66.3) 211 (55.4) 757 (70.1) ≥ 75 y 283 (19.4) 96 (25.2) 187 (17.3) BMI, median (IQR), kg/m2 1375 28.1 (25.4-32.3) 27.7 (24.7-31.2) 28.4 (25.4-32.6) 0.001 BMI, by WHO classes, No. (%) 1375 < 18.5 kg/m2 8 (0.6) 4 (1.1) 4 (0.4) 18.5 to 24.9 kg/m2 296 (21.5) 93 (25.7) 203(20.0) 25 to 29.9 kg/m2 557 (40.5) 147 (40.6) 410 (40.5) 0.005b 30 to 34.9 kg/m2 301 (21.9) 71 (19.6) 230 (22.7) 35 to 39.9 kg/m2 134 (9.8) 32 (8.8) 102 (10.1) ≥ 40 kg/m2 79 (5.8) 15 (4.1) 64 (6.3) Pre-existing conditions Diabetes, No. (%) 1461 426 (29.2) 99 (26) 327 (30.3) 0.11 Hypertension, No. (%) 1461 752 (51.5) 193 (50.7) 559 (51.8) 0.71 Hyperlipidemia, No. (%) 1461 423 (29) 99 (26) 324 (30) 0.14 Current smoker, No. (%) 1275 83 (6.5) 14 (4.1) 69 (7.4) 0.049 Cardiovascular disease, No. 1461 373 (25.5) 111 (29.1) 262 (24.3) 0.06 (%) Chronic kidney disease, No. 1265 133 (10.5) 35 (10.8) 98 (10.4) 0.86 (%) Chronic obstructive pulmonary 1461 146 (10) 45 (11.8) 101 (9.4) 0.17 disease, No. (%) Malignancy, No. (%) 1461 150 (10.3) 37 (9.7) 113 (10.5) 0.68 Immunosuppression, No. (%) 1275 88 (6.9) 19 (5.9) 69 (7.3) 0.39 Severity score SAPS-II 1135 39 (29-53) 29 (22-39) 43 (33-57) < 0.001 Treatments This article is protected by copyright. All rights reserved Hypoglycemic treatment 1362 Accepted Article No treatment, No. (%) Yes, but no insulin, No. (%) 990 (72.7) 222 (16.3) 282 (76.2) 58 (15.7) 708 (71.4) 164 (16.5) 0.09 b Insulin, No. (%) 150 (11) 30 (8.1) 120 (12.1) Renin–angiotensin–aldosterone system inhibitors ARBs, No. (%) 1153 192 (16.7) 59 (18.1) 133 (16.1) 0.41 ACEi, No. (%) 1153 190 (16.5) 55 (16.9) 135 (16.3) 0.82 Abbreviation: N., number, IQR, interquartile range, y, years, BMI, body mass index, WHO, World Health Organization, SAPS-II, simplified acute physiology score, ARBs, angiotensin receptor blockers, ACEi, angiotensin converting enzyme inhibitors, IMV, invasive mechanical ventilation. SAPS: simplified acute physiology score. a Mann-Whitney U test was used for continuous variables comparison, and χ2 test with Yates' continuity correction was used for categorical variables comparison if not specified. b Cochran-Armitage Trend Test was used for comparison of ordinal BMI categories. This article is protected by copyright. All rights reserved Accepted Article Table 2. Association of BMI categories with the need of invasive mechanical ventilation requirement and 28-day all-cause mortality BMI categories, kg/m2 P-value a / b < 25 25 to 29.9 30 to 34.9 35 to 39.9 ≥ 40 (n=319) (n=591) (n=323) (n=143) (n=85) IMV No. (%) 219 (68.5) 436 (73.8) 248 (76.7) 109 (76.0) 69 (81.4) Center-adjusted OR 1.20 (0.86 1.46 (0.98 1.53 (0.93 2.35 (1.21 1.00 (ref.) 0.07 / 0.004 (95% CI) to 1.66) to 2.15) to 2.52) to 4.52) Fully-adjusted OR 1.16 (0.82 1.63 (1.09 1.72 (1.02 3.06 (1.53 1.00 (ref.) 0.008 / < 0.001 (95% CI)c to 1.61) to 2.44) to 2.88) to 6.10) Death No. (%d) 84 (40.3) 134 (33.6) 72 (33.3) 34 (36.4) 31 (47.2) Unadjusted HR 0.79 (0.59 0.75 (0.53 0.73 (0.48 1.21 (0.77 1.00 (ref.) 0.10 / 0.81 (95% CI) to 1.06) to 1.06) to 1.12) to 1.87) Adjusted HR (95% 0.82 (0.61 0.96 (0.68 0.92 (0.60 1.68 (1.06 1.00 (ref.) 0.03 / 0.13 CI)c to 1.10) to 1.36) to 1.42) to 2.64) Values were calculated after handling missing values by multiple imputations. This article is protected by copyright. All rights reserved Accepted Article Abbreviation: IMV, invasive mechanical ventilation, No., number, OR, odds ratio, CI, confidence interval, HR, hazard ratio, BMI, body mass index, ref., reference. a p-value by treating BMI categories as categorical variable in regression model b p-value for by treating BMI categories as ordinal variable in regression model. c adjusted on center and pre-specified covariates (age, sex, diabetes, hypertension, hyperlipidemia and current smoking) d Kaplan-Meier estimate at 28-days This article is protected by copyright. All rights reserved Accepted Article Table 3. Association of continuous BMI with the need for invasive mechanical ventilation in multivariable analysis including age, sex and metabolic risk factors OR (95% CI) P-value BMI, per 5 kg/m2 increase 1.27 (1.12 to 1.45) < 0.001 Age, per 10-y increase 1.17 (1.05 to 1.31) 0.004 Male sex 1.82 (1.38 to 2.41) < 0.001 Hypertension 0.97 (0.72 to 1.30) 0.84 Diabetes 1.21 (0.89 to 1.65) 0.21 Hyperlipidemia 1.08 (0.78 to 1.48) 0.64 Current smoking 1.25 (0.66 to 2.35) 0.48 OR were calculated using multivariable mixed logistic regression model by taking into account center as random effect and after handling missing values by multiple imputations. Abbreviation: BMI, body mass index, y, year, OR, odds ratio, CI, confidence interval. This article is protected by copyright. All rights reserved Accepted Article Table 4. Association of BMI ≥ 40 kg/m2 with the 28-day all-cause mortality in multivariable analysis including age, sex and metabolic risk factors HR (95% CI) P-value BMI ≥ 40 kg/m2 1.84 (1.23 to 2.75) - Age, per 10-y increase 1.74 (1.54 to 1.95) < 0.001 Male sex 1.24 (0.96 to 1.60) 0.10 Hypertension 0.93 (0.72 to 1.19) 0.56 Diabetes 1.25 (0.98 to 1.58) 0.07 Hyperlipidemia 1.00 (0.77 to 1.29) 0.97 Current smoking 1.00 (0.59 to 1.69) 0.99 HR were calculated using frailty model by taking into account center as random effect and after handling missing values by multiple imputations. Abbreviation: BMI, body mass index, y, year, HR, hazard ratio This article is protected by copyright. All rights reserved Accepted Article Figure 1. Linear association of BMI categories with the need of invasive mechanical ventilation requirement (Figure 1a) and non-linear association of BMI with 28-day all-cause mortality (Figure 1b) Odds and hazards ratio were calculated using the lean category (BMI < 25 kg/m2) as reference and adjusted on center and pre-specified covariates (age, sex, diabetes, hypertension, hyperlipidemia and current smoking) after handling missing values by multiple imputations. This article is protected by copyright. All rights reserved Accepted Article P-value were calculated by treating BMI categories as categorical variable in regression model. Abbreviation: CI, confidence interval. Figure 2. Association of continuous body mass index (BMI) with the need for invasive mechanical ventilation, in the overall study population after handling missing values by multiple imputations and on the complete cases population (sensitivity analysis) according to sex, and sex/age subgroups. Odds ratios (OR) are expressed per 5 kg/m2 increase, with 95% confidence intervals (CI) and calculated using mixed logistic regression models including center as random effect and adjustments for prespecified known risk factors (fixed effects) as diabetes, hypertension, hyperlipidemia and current smoking. This article is protected by copyright. All rights reserved Accepted Article This article is protected by copyright. All rights reserved Tables and Figures titles Accepted Article Table 1. Characteristics at baseline of patients admitted in ICU, in the whole cohort (n=1461) and in patients who did (n=1080) or did not require invasive mechanical ventilation (n=381) Table 2. Association of BMI categories with the need of invasive mechanical ventilation requirement and 28-day all-cause mortality Table 3. Association of continuous BMI with the need for invasive mechanical ventilation in multivariable analysis including age, sex and metabolic risk factors Table 4. Association of BMI ≥ 40 kg/m2 with the 28-day all-cause mortality in multivariable analysis including age, sex and metabolic risk factors Figure 1. Linear association of BMI categories with the need of invasive mechanical ventilation requirement (Figure 1a) and non-linear association of BMI with 28-day all-cause mortality (Figure 1b) Figure 2. Association of continuous body mass index (BMI) with the need for invasive mechanical ventilation, in the overall study population after handling missing values by multiple imputations and on the complete cases population (sensitivity analysis) according to sex, and sex/age subgroups. Odds ratios (OR) are expressed per 5 kg/m2 increase, with 95% confidence intervals (CI) and calculated using mixed logistic regression models including center as random effect and adjustments for prespecified known risk factors (fixed effects) as diabetes, hypertension, hyperlipidemia and current smoking. This article is protected by copyright. All rights reserved Accepted Article Supplementary material Table S1: Association of BMI categories with the need of invasive mechanical ventilation requirement and 28-day all-cause mortality in available case analysis. Table S2: Association of continuous BMI with the need for invasive mechanical ventilation in multivariable analysis including age, sex and metabolic risk factors, in available case analysis. Table S3: Association of BMI ≥ 40 kg/m2 with the 28-day all-cause mortality in multivariable analysis including age, sex and metabolic risk factors, in available case analysis. Figure S1: Flow diagram of the 1461 study participants patients. A total of 28 centers in 8 countries (Belgium, Brazil, China, France, Israel, Italy, Spain, and USA,) were initially contacted to participate. Abbreviation: ICU, intensive care unit, RT-PCR, reverse transcription polymerase chain reaction. This article is protected by copyright. All rights reserved Figure S2: Distribution of body mass index in the study population of critically ill patients Accepted Article with COVID-19, and in general population, adjusted for age and sex in each country. Figure S3: Description of patients enrolled in each in 21 centers, by population size, male sex, age, body mass index and known metabolic risk factors. Patients are described for each center and plotted as a radial bar chart. Centers are sorted by population size (the largest at the outer edge of the circle). Proportion (%) of patients from the cohort (A); proportion (%) of male patient (B); mean age in years (C); mean body mass index in kg/m2 (D); proportion of patients (%) with diabetes (E), hypertension (F), hyperlipidemia (G), and current smoking (H). Study Groups and Collaborators This article is protected by copyright. All rights reserved Accepted Article oby_23223_f1.tiff This article is protected by copyright. All rights reserved Overall population oby_23223_f2.pdf Available cases Accepted Article OR (95%CI) P OR (95%CI) P Overall (n=1461) 1.21 ( 1.07 to 1.36) 0.002 (n=1375) 1.25 ( 1.09 to 1.42) <0.001 Sex Men (n=1070) 1,21 (1.02 to 1.42) 0.02 (n=1005) 1,17 (0.98 to 1.38) 0.070 Women (n=391) 1.30 (1.08 to 1.56) 0.004 (n=370) 1.29 (1.07 to 1.56) 0.006 Sex / Age Men <50 years (n=162) 1.11 ( 0.83 to 1.49) 0.47 (n=150) 1.05 ( 0.78 to 1.42) 0.73 Women <50 years (n=48) 1.65 ( 0.97 to 2.79) 0.061 (n=45) 1.86 ( 1.07 to 3.22) 0.026 Men 50-75 years (n=719) 1.23 ( 0.99 to 1.52) 0.054 (n=669) 1.20 ( 0.95 to 1.51) 0.11 Women 50-75 years (n=249) 1.22 ( 0.97 to 1.53) 0.081 (n=233) 1.19 ( 0.94 to 1.50) 0.13 Men ≥75 years (n=189) 1.18 ( 0.79 to 1.76) 0.40 (n=186) 1.20 ( 0.76 to 1.87) 0.43 Women ≥75 years (n=94) 1.27 ( 0.87 to 1.84) 0.21 (n=92) 1.24 ( 0.85 to 1.78) 0.26 0.5 1 2 4 0.5 1 2 4 This article is protected by copyright. All rights reserved