key: cord-0854206-yiiz9s53
authors: Czernichow, Sébastien; Beeker, Nathanael; Rives‐Lange, Claire; Guerot, Emmanuel; Diehl, Jean‐Luc; Katsahian, Sandrine; Hulot, Jean‐Sébastien; Poghosyan, Tigran; Carette, Claire; Sophie Jannot, Anne
title: Obesity doubles mortality in patients hospitalized for SARS‐CoV‐2 in Paris hospitals, France: a cohort study on 5795 patients
date: 2020-08-20
journal: Obesity (Silver Spring)
DOI: 10.1002/oby.23014
sha: 0d6a50ea587b5e917078213091c7fe64d8a603b3
doc_id: 854206
cord_uid: yiiz9s53

BACKGROUND: Preliminary data from different cohorts of small sample size or with short follow‐up indicate poorer prognosis in people with obesity compared to other patients. This study aims to precisely describe the strength of association between obesity in patients hospitalised with Covid‐19 and mortality and clarify the risk according to usual cardiometabolic risk factors in a large cohort. METHODS: This is a prospective cohort study including 5795 patients aged 18‐79 years hospitalized from 1(st) February 2020 to 30 April 2020 in Paris area, with confirmed infection by SARS‐CoV‐2. Adjusted regression models were used to estimate the odds ratios (OR) and 95% confidence intervals (95% CI) for mortality rate at 30 days across BMI classes, without and with imputation for missing BMI. RESULTS: 891 deaths occurred at 30 days. Mortality was significantly raised in people with obesity with the following OR in BMI 30‐35, 35‐40 and >40 kg/m(2): 1.89 (95%CI 1.45‐2.47), 2.79 (1.95‐3.97) and 2.55 (1.62‐3.95), respectively (18.5‐25 kg/m(2), as the reference class). This increase holds for all age classes. CONCLUSION: Obesity doubles mortality in patients hospitalized with Covid‐19.

Since the end of 2019 and its emergence from China, the pandemic of Covid-19 has become the main worldwide public health threat responsible of lockdown measures for million people. Covid-19 symptoms range from a wide variety of clinical presentation: from none to severe respiratory symptoms leading to death. 1

This article is protected by copyright. All rights reserved Obesity, and especially its most extreme forms, is a source of stigma 2 , high emergency care utilization 3 , higher morbidity and increased mortality. 4 In the context of infectious disease, high body mass index (BMI) has been recognized as a risk factor for nosocomial, skin, as well as, respiratory disease infections. 5 About ten years ago, against the backdrop of the H1N1 influenza epidemic, it was clearly pointed out in a meta-analysis on more than 3000 individuals, that people with severe obesity had a two-fold increased risk for intensive care unit (ICU) admission and mortality, compared to counterparts without obesity. 6 In two single-center studies, the risk for need of invasive mechanical ventilation in patients with Covid-19 related severe acute respiratory syndrome was higher in patients with severe obesity, compared to normal weight patients, 7 but not for those with class I obesity (BMI 30 to 35 kg/m²). 8 Preliminary data from different cohorts of patients infected by COVID-19 of small sample size (lower than 400 patients), with short follow-up or with poorly described BMI indicate poorer prognosis in people with obesity compared to other patients. For instance one study shows higher mortality frequency in people with severe obesity admitted to ICU compared to people with less severe obesity 9 . However, it is not possible to conclude from these results that obesity is an independent factor of mortality for patients infected with COVID-19 due to the small sample sizes of these studies, neither to have a precise estimate of obesity size effect due to the absence of BMI categories and incomplete follow-up. These results need therefore to be confirmed in a large cohort, with available BMI and adequate follow-up. To further investigate the topic, we conducted an analysis of the association between BMI and risk for mortality at 30 days after hospitalization for Covid-19 in all Paris area-based public University hospitals.

This article is protected by copyright. All rights reserved

The Assistance Publique -Hôpitaux de Paris (AP-HP) is the largest hospital entity in Europe with 39 hospitals (22,474 beds) mainly located in the Greater Paris area with 1.5 M hospitalizations per year (10% of all hospitalizations in France). Since 2014, AP-HP is building an analytics platform based on a clinical data repository (CDR), aggregating day-to-day clinical data from 8.8 million patients captured by clinical databases. 10 The CDR has received the authorization of the French Data Protection Authority (Commission Nationale de l'Informatique et des Libertés, CNIL, n°1980120). From the beginning of COVID-19 epidemics, the EDS-COVID database stemmed from this initiative. The later database retrieved electronic health records from all AP-HP facilities and aggregates them into a clinical data warehouse following OMOP common data model. 11 Our analysis follows recommendations provided by the REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) Statement. 12 This study was approved by the institutional review board (authorization number IRB 00011591) from the Scientific and Ethical Committee from the AP-HP. All subjects included in this study were informed about the reuse of their data for research and subjects that objected to the reuse of their data were excluded from this study, in accordance to French legislation.

This article is protected by copyright. All rights reserved Smoking status was defined as being current smoker or having an history of smoking using the formerly mentioned "Covid19-APHP-NLP pipeline". Comorbidities were extracted from ICD-10 codes of previous and current hospitalization, I10 for hypertension, N18 for chronic kidney disease, G473 for sleep apnea, E78 for dyslipidemia, C00 to D48 for malignancies.

Heart failure was defined as having an I50 ICD-10 code in a previous hospitalization. Diabetes was defined as having a E11 ICD-10 codes of diabetes or having a HbA1C greater than 6.5% in any previous hospitalization.

Indirect information concerning BMI value was also retrieved for BMI imputation in patients with missing BMI. Using 4-digits E66 ICD-10 codes, the following variables were created: 

This article is protected by copyright. All rights reserved class, E6606, E6616, E6626, E6686, E6696 as ICD-10 [40; 50] BMI class, E6607, E6617, E6627, E6687, E6697 as ICD-10 >50 BMI class. Malnutrition was extracted using E43 to E46 ICD-10 codes. Mentions of obesity in free-text reports were also retrieved using the formerly mentioned "Covid19-APHP-NLP pipeline".

Age at admission, sex, ICU admission and death during hospitalisation were extracted from hospital administrative data.

Considered outcome was death during hospitalisation at 30 days after positive Covid PCR.

Outcome was retrieved through administrative hospital data.

Patients' characteristics were defined according to BMI classes and sex using median and interquartile range for continuous variables and proportion for binary variables both before and after missing BMI imputation.

We imputed missing BMI category using predictive mean matching considering as the following as explaining variables: comorbidities (hypertension, diabetes, sleep apnea, dyslipidemia, chronic kidney disease, heart failure, cancer), smoking status, sex, age and indirect information regarding BMI value (obesity from free-text reports, variables extracted from 4-digits E66 ICD-10 codes and malnutrition ICD-10 codes). To assess the predictive ability of these variables, we performed a regression analysis on BMI using the same

This article is protected by copyright. All rights reserved explaining variables on the complete dataset. To account for imputation variability, we generated five imputed samples.

Multivariate OR (95% CI) were estimated according to BMI classes, with adjustment for comorbidities, smoking status, age and sex using logistic regressions, both including and excluding patients with missing BMI and with stratification on age class. For analysis including patients with imputed BMI, variation across imputed datasets was taken into account by incorporating samples variability in the estimated confidence intervals. All analyses were performed using R 4.0.2 software and MICE package was used for multiple imputations process.

During the period of February 1 st through April 30, 2020, a total of 8671 patients with a PCRconfirmed Covid-19 infection were hospitalised in one of the thirty nine hospitals (Fig. 1 ).

Among them, 5795 patients were between 18 and 80 years old and had available bioclinical data of whom 4056 had available BMI (2597 extracted from free-text reports and 1459 extracted from clinical signs). Mean (SD) age was 58.9 y (14.7) in women (n=2004) and 60.3 y (13.0) in men (n=3791) ( Table 1) . Mean BMI was 29.3 (7.5) and 27.2 (6) kg/m² in women and men, respectively. Comorbidities were frequent and increased with BMI classes. People with class III obesity aggregated the most risk factors. Admission in intensive care unit (ICU) increased with BMI classes. Use of mechanical ventilation did not follow an obvious trend across BMI classes. BMI was imputed for 1739 patients and main BMI predictor in the imputation model were variables derived from indirect information on BMI from

This article is protected by copyright. All rights reserved hospitalization reports and ICD-10 codes (See regression coefficients and significance on   supplementary table 2) . Correlation coefficient for the regression model used to assess the ability to predict BMI was 63%, therefore the available indirect information on BMI were relevant to predict BMI.

Mortality was significantly higher in people with obesity taking into account age groups, sex, Figure S1 ).

This large study investigates the role of obesity on mortality risk at 30 days in patients hospitalized with Covid-19 infection in any of the 39 university public hospitals in Paris Region (France). We have shown that obesity was a major prognostic factor, independently of known chronic comorbidities.

Several hypotheses can be made to explain a worst survival rate in people with obesity compared to people without obesity. First, obesity is characterized by an increased low-

This article is protected by copyright. All rights reserved grade inflammatory state that relates to a dysfunctional adipose microenvironment. 13 The adipose cells are responsible for the secretion of pro-inflammatory adipokines, such as alpha-TNF, interleukine-6, lower adiponectin and increased leptin. The dysregulated cytokinic environment may be the early biological step that mediates multiple organ failure. 14 Second, obesity accumulates several respiratory disorders such as obstructive sleep apnea syndrome, asthma, restrictive respiratory syndrome and obesity hypoventilation syndrome. 15 People with obesity are at particular risk of acute respiratory distress syndrome (ARDS), whatever the aetiology of the syndrome. 16 One explanation for the high prevalence of ARDS in people with obesity may be the very specific pulmonary mechanics of such patients, characterized mainly by excessively high pleural pressures with generally preserved chest wall compliance. Such a pattern leads to the frequent occurrence of negative transpulmonary pressures favouring a greater incidence of atelectasis. 17 One suggested mean to counteract such phenomenon is to use high positive end-expiratory pressure (PEEP) settings, ideally based on oesophageal monitoring. 18 An important result of the study is the poorer vital prognosis observed in people with obesity and with Covid-19. Such a result is contrasting with the general findings of similar or even better prognosis than in the ARDS population without obesity. 19 However, one should keep in mind the worse vital prognosis previously observed in people with obesity with H1N1 infection. 6 A specific detrimental influence of the viral insult in people with obesity is therefore conceivable. In addition, the design of the study didn't allow to precisely assessing the ventilator settings used in people with obesity with Covid-19, compared to those without

This article is protected by copyright. All rights reserved obesity. BMI data was missing in about a third of included patients and was therefore imputed when missing. Of note, we benefitted from a large number of indirect information regarding BMI missing values using free-text reports and ICD-10 codes, but it was not sufficient to accurately predict BMI. However, odds ratios before and after imputation were similar values.

This study was considerably facilitated by the EDS-COVID database, which retrieved electronic health records from all AP-HP facilities and aggregates them into a clinical data warehouse. This clinical data warehouse allowed retrieving in real time a large set of data to deeply characterize our study population. This approach was secured by a data quality program ensuring high standard of quality for this database. 10 Furthermore, even if we were able to collect a large sample size, BMI does not capture body composition or even variations in weight. Indeed, our data indicate poorer prognosis with aging, both for undernutrition and severe obesity, which strongly relates to muscle mass loss and sarcopenia in the context of an hypercatabolic state related to Covid-19 infection. 23 It has been previously shown that sarcopenic obesity is associated with a longer hospital stay and a worst recovery after ICU. 24 Our results might be limited by the fact that only mortality during hospitalization was considered. However, it is unlikely that patients hospitalized for Covid-19 died from their infection after being discharged from hospital. Therefore, the subsequent underestimation of mortality due to this potential bias is likely to be limited.

This article is protected by copyright. All rights reserved

In summary, our data show for the first time in a large multicentre setting that obesity is related to mortality in patients hospitalized with Covid-19. The presence or absence of cardiometabolic risk factors did not modify the increased mortality risk. In the context of a global Covid-19 pandemic lockdown, the detrimental effect of accumulating sedentary lifestyle and increased food intake will worsen quality of life, depression risk, 25 and global mortality in fragile patients with severe obesity. Thus, people with obesity in Covid-19 pandemic context require a personalized management.

Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study

Joint international consensus statement for ending stigma of obesity

Obesity and emergency care in the French CONSTANCES cohort

Body-mass index and mortality among adults with incident type 2 diabetes

Obesity and site-specific nosocomial infection risk in the intensive care unit

Obesity is associated with higher risk of intensive care unit admission and death in influenza A (H1N1) patients: a systematic review and metaanalysis

Obesity is associated with severe forms of COVID-19

High prevalence of obesity in severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) requiring invasive mechanical ventilation

Covid-19 in Critically Ill Patients in the Seattle Region -Case Series

Initializing a hospital-wide data quality program. The AP-HP experience

Observational Health Data Sciences and Informatics (OHDSI): Opportunities for Observational Researchers

The REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) statement

Inflammatory processes in obesity: focus on endothelial dysfunction and the role of adipokines as inflammatory mediators

Obesity a Risk Factor for Severe COVID-19 Infection: Multiple Potential Mechanisms

Obesity in the critically ill: a narrative review

Body mass index is associated with the development of acute respiratory distress syndrome

Obesity and ARDS: Opportunity for Highly Personalized Mechanical Ventilation?

A lung rescue team improves survival in obesity with acute respiratory distress syndrome

ARDS in People with obesity: Specificities and Management

Sarcopenic obesity in the ICU

Effect of obesity on intensive care morbidity and mortality: a meta-analysis

We would like to acknowledge the authors thank the EDS APHP Covid consortium integrating the APHP Health Data Warehouse team as well as all the APHP staff and volunteers who contributed to the implementation of the EDS-COVID database and operating solutions for this database (list in the Supplementary Table 1) .

S. Czernichow reports honorarium from Novonordisk for board participation and conferences, as well as participation to Mygoodlife. All other authors declared no conflict of interest

This article is protected by copyright. All rights reserved

SC and ASJ designed the study and had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. SC drafted the paper with the help of ASJ, CRL, TP, EG, JSH, JLD, SK, NB and CC. ASJ and NB did the analyses. Data were collected from all Assistance Publique -Hôpitaux de Paris. All authors critically revised the manuscript for important intellectual content and gave final approval for the version to be published.

The study was funded by Assistance Publique hôpitaux de Paris. The Study sponsor had no role in study design, data collection, data analysis, data interpretation, or writing of this report. The corresponding author (SC) had full access to all the data in the study and had final responsibility for the decision to submit for publication.

This article is protected by copyright. All rights reserved

This article is protected by copyright. All rights reserved ccepted Article This article is protected by copyright. All rights reserved