key: cord-0755425-pwue0zz3
authors: Silva, N. J.; Ribeiro-Silva, R. C.; Ferreira, A. J.; Teixeira, C. S.; Rocha, A. S.; Alves, F. J. O.; Falcao, I. R.; Pinto, E. J.; Santos, C. A. S.; Fiaccone, R. L.; Ichihara, M. Y. T.; Paixao, E. S.; Barreto, M. L.
title: Combined association of obesity and other cardiometabolic diseases with severe COVID-19 outcomes: a nationwide cross-sectional study of 21,773 Brazilian adult and elderly inpatients
date: 2021-05-18
journal: nan
DOI: 10.1101/2021.05.14.21257204
sha: 1dad6948dfe438a3bccf0d3acce7f89a8ca31469
doc_id: 755425
cord_uid: pwue0zz3

Objective: To investigate the combined association of obesity, diabetes mellitus (DM), and cardiovascular disease (CVD) with severe COVID-19 outcomes in adult and elderly inpatients. Design: Cross-sectional study based on registry data from Brazil's influenza surveillance system. Setting: Public and private hospitals across Brazil. Participants: Eligible population included 21,942 inpatients aged 20 years or older with positive RT-PCR test for SARS-CoV-2 until Jun 9th, 2020. Main outcome measures: Severe COVID-19 outcomes were non-invasive and invasive mechanical ventilation use, ICU admission, and death. Multivariate analyses were conducted separately for adults (20-59 years) and elders (>=60 years) to test the combined association of obesity (without and with DM and/or CVD) and degrees of obesity with each outcome. Results: A sample of 8,848 adults and 12,925 elders were included. Among adults, obesity with DM and/or CVD showed higher prevalence of invasive (PR 3.76, 95%CI 2.82-5.01) and non-invasive mechanical ventilation use (2.06, 1.58-2.69), ICU admission (1.60, 1.40-1.83), and death (1.79, 1.45-2.21) compared with the group without obesity, DM, and CVD. In elders, obesity alone (without DM and CVD) had the highest prevalence of ICU admission (1.40, 1.07-1.82) and death (1.67, 1.00-2.80). In both age groups, obesity alone and combined with DM and/or CVD showed higher prevalence in all outcomes than DM and/or CVD. A dose-response association was observed between obesity and death in adults: class I 1.32 (1.05-1.66), class II 1.41 (1.06-1.87), and class III 1.77 (1.35-2.33). Conclusions: The combined association of obesity, diabetes, and/or CVD with severe COVID-19 outcomes may be stronger in adults than in elders. Obesity alone and combined with DM and/or CVD had more impact on the risk of COVID-19 severity than DM and/or CVD in both age groups. The study also supports an independent relationship of obesity with severe outcomes, including a dose-response association between degrees of obesity and death in adults. These findings suggest important implications for the clinical care of patients with obesity and severe COVID-19 and support the inclusion of people with obesity in the high-risk and vaccine priority groups for protection from SARS-CoV-2.

• The study was based on registry data of a large nationwide sample of patients admitted, due 48 to severe SARS-CoV-2 infection, to public and private hospitals across the country. 49

• The large sample size and data availability allowed us to analyze the combined association of 50 obesity, diabetes and cardiovascular disease with severe COVID-19 outcomes, separately by 51 age groups and controlled by important confounding variables, e.g. underlying comorbidities. 52

• The cross-sectional study design does not allow causal inference, and generalization of results 53 must be cautious since only hospitalized cases of severe COVID-19 were included. 54

• As the study used routinely collected data, which has not been designed primarily for research 55 purposes, it may bring well-known limitations related to missing, underestimation, and 56 potential misclassification. 57

The coronavirus disease 2019 pandemic, caused by the severe acute respiratory 59 existence of diabetes and any chronic cardiovascular disease was obtained from dichotomous 108 questions (yes/no), which were answered based on patient or family's report or medical diagnosis. 109 We created a polytomous four-category variable to evaluate the separate and combined exposure of 110 obesity, diabetes and cardiovascular disease: none/reference (no existence of obesity, diabetes and 111 cardiovascular disease), OB (only existence of obesity), OB + DM and/or CVD (existence of obesity 112 with diabetes and/or cardiovascular disease), and DM and/or CVD (existence of diabetes and/or 113 cardiovascular disease). We also analyzed obesity in adults according to the following degrees of 114 severity based on WHO reference 21 : no obesity (<30 kg/m 2 ), obesity class I (≥ 30-34.9 kg/m 2 ), obesity 115 class II (≥ 35-39.9 kg/m 2 ), and obesity class III (≥ 40 kg/m 2 ). Due to the unavailability of BMI cutoff 116 points to classify the degree of obesity in elders, this analysis was only performed for adults. 117

The severe COVID-19 outcomes were mechanical ventilation use, ICU admission, and death. 119

Information on the use of mechanical ventilation by the patient was obtained and analyzed as a 120 polytomous three-category variable (no use/ use of non-invasive ventilation/ use of invasive 121 ventilation). ICU admission was obtained and analyzed as a dichotomous variable (no/ yes). Death 122 was analyzed as a dichotomous variable based on the patient's endpoint outcome (cure/ death). 123

Covariates 124

Demographic and comorbidity information were selected as descriptive and confounding variables. 2 125 Age in years was calculated from birth and notification dates. Sex was obtained as a dichotomous 126 variable (female/ male). The preexistence of each comorbidity was also obtained as a dichotomous 127 variable (no/ yes): chronic pulmonary disease, asthma, chronic kidney disease, chronic hematologic 128 disease, neurological disease, chronic liver disease, and immunodeficiency/ immunosuppression. 129

All analyses were subdivided into adults (≥ 20 and < 60 years) and elders (≥ 60 years). For descriptive 131 analyses, absolute and relative frequencies were calculated for the demographic and comorbidity 132 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted May 18, 2021. ; https://doi.org/10.1101/2021.05.14.21257204 doi: medRxiv preprint variables according to the main exposure variable. Multinomial logistic regression models were 133 conducted to test the association of obesity (without and with diabetes and/or CVD) with non-invasive 134 and invasive mechanical ventilation use . To test the association of this exposure variable with ICU  135 admission and death, simple logistic regression models were performed. Same models were analyzed 136 considering the degree of obesity as the main exposure variable for adults. Crude and adjusted 137 estimates were interpreted based on the prevalence ratio (PR) and 95% confidence intervals (95%CI). 138

These estimates were obtained from logistic models using delta method, function 'prLogisticDelta', 139 which is implemented in R and available in the package 'prLogistic'. Adjusted models included the 140 following list of confounding variables: sex, age (years), and the preexistence of chronic pulmonary 141 disease, asthma, kidney disease, hematologic disease, neurological disease, liver disease, and 142 immunodeficiency/ immunosuppression. The models that tested the degrees of obesity were also 143 adjusted for DM and CVD. All analyses were performed using Stata version 15.1 (Stata Corporation, 144

College Station, USA) and R version 3.6.1 (R Foundation for Statistical Computing, Austria). 145

As the study exclusively used publicly available de-identified data, it was not possible to involve 147 patients or the public in the design, or conduct, or reporting, or dissemination plans of our research. 148

During the study period, 21,942 individuals registered in the SIVEP-Gripe were ≥ 20 years old, 150 hospitalized, tested positive for SARS-CoV-2, and had complete demographic and comorbidity 151 information (Figure 1 ). Of these, 169 (0.8%) were excluded due to implausible values of BMI. Of 152 the 21,773 individuals included in the study, 8,848 (40.3%) were adults aged between 20-59 years, 153 and 12,925 (59.6%) were elders aged 60 years or older. Since some patients were still hospitalized 154 on the study endpoint date, information for some outcomes were incomplete. The study samples 155 included in the analysis of each outcome were 8,075 adults and 11,829 elders for mechanical 156 ventilation, 8,414 adults and 12,222 for ICU admission, and 6,565 adults and 9,943 elders for death. 157 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted May 18, 2021. ; https://doi.org/10.1101/2021.05.14.21257204 doi: medRxiv preprint Based on demographic and clinical characteristics, the analytical samples in each outcome were very 158 similar to the overall study population and the excluded samples (Supplementary Table 1) . 159

The prevalence of obesity was 9.7% in adults and 3.5% in elders. The frequency of obesity without 160 and with DM and/or CVD was respectively 4.6% and 5.1% in adults and 0.7% and 2.8% in elders. 161

Non-invasive and invasive mechanical ventilation were respectively required by 45. 0% subgroup of adults with DM and/or CVD showed in general the lowest prevalence ratios for all 172 analyzed outcomes than the subgroups with the presence of obesity alone or combined ( Table 3) . 173

Among elders, obesity without DM and CVD increased independently the prevalence of ICU 174 admission by 40% (95%CI 1.07-1.82) and death by 67% (1.00-2.80). To a lesser extent, obesity with 175 DM and/or CVD was also associated with an increased prevalence of invasive mechanical ventilation 176 need (PR 1.66, 95%CI 1.22-2.27), ICU admission (PR 1.37, 95%CI 1.19-1.59), and death (PR 1.39, 177 95%CI 1.07-1.80). Elders with DM and/or CVD had the lowest prevalence ratios for the analyzed 178 outcomes than the subgroups of elders with obesity alone or combined ( Table 3) . 179

In the analyses by the degree of obesity, we did not observe much difference in the prevalence of 180 adverse outcomes, except for the prevalence of death that increased with the severity of obesity: Class 181 I 1.32 (95%CI 1.05-1.66), Class II 1.41 (1.06-1.87), and Class III 1.77 (1.35-2.33) ( Table 4) . 182 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 18, 2021. ; https://doi.org/10.1101/2021.05.14.21257204 doi: medRxiv preprint

This is the first study that describe the relationship of obesity and COVID-19 in Brazil, based on a 184 large nationwide sample of adults and elders tested positive for SARS-CoV-2 and admitted to public 185 and private hospitals. Our results highlights that obesity with DM and/or CVD was associated with 186 higher rates of invasive mechanical ventilation use, ICU admission, and death in adults, while obesity 187 alone (without DM and CVD) was associated with higher rates of ICU admission and death among 188 elders. In both age groups, obesity alone and obesity combined with DM and/or CVD had more 189 impact on the risk of all severe COVID-19 outcomes than the subgroup with DM and/or CVD. The 190 study also supports the independent association of obesity with the analyzed outcomes and a dose-191 response association between degrees of obesity and death in adults. It is important to note that the COVID-19 pandemic imposes a double burden of disease, especially 205 among the elderly individuals, since the prevalence of diabetes, hypertension, cardiovascular 206 diseases, and other comorbidities associated with COVID-19 severity increases with age. 3,27 207 However, our study suggests that obesity combined with diabetes and/or cardiovascular disease may 208 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 18, 2021. ; https://doi.org/10.1101/2021.05.14.21257204 doi: medRxiv preprint offer higher risk of COVID-19 severity for adults although the overall prevalence of diseases and 209 rates of ICU admission and mortality were higher in elders. Obesity alone seemed to provide higher 210 risk of severe outcomes, especially death, in elders. 211

Few studies to-date have explored the combined and additional effect of obesity on COVID-19 212 severity. 13, 28 A study investigated the patterns of multimorbidity among fatal cases of COVID-19 in 213

Colombia. 28 Similar to our study, the authors found that obesity alone or with other diseases was 214 associated with a higher risk of COVID-19 fatality among young people. Furthermore, a population-215 based study in Mexico observed that the addition of obesity to any number of comorbidities 216 significantly increased the risk of COVID-19 lethality. 13 Using a causally ordered mediation analysis, 217 this study also found that 49.5% of the effect of diabetes on COVID-19 lethality was mediated by 218 obesity, particularly in early-onset cases < 40 years of age. 219

Other studies also suggest that obesity is independently associated with severe outcomes of COVID-220 19, regardless of age and other associated comorbidities. [11] [12] [13] [14] A large study in Mexico 13 showed that 221 patients with obesity had higher rates of ICU admission and were more likely to be intubated in 222 relation to patients without obesity. This study also found a five-fold increased risk of mortality due 223 to COVID-19 in patients with obesity. 13 In a hospital-based study in France, it was observed that BMI 224 > 35 kg/m 2 was associated with the need for invasive mechanical ventilation. 14 225

Few studies to-date have similarly found a dose-response association between degrees of obesity and 226 COVID-19 death. 29 Based on care records of 17,278,392 UK adults, the study showed that the risk 227 of COVID-19 death increases independently with the degree of obesity: 30-34.9 kg/m 2 (HR 1.05), 228 35-39.9 kg/m 2 (1.40), and ≥40 kg/m 2 (2.66). 29 Other studies have evidenced the association of obesity 229 with COVID-19 complications and death among adults. 12,30 A hospital-based study in New York City 230 showed that morbid obesity (BMI ≥ 40 kg/m 2 ) is strongly and independently associated with death in 231 hospitalized patients younger than 50 years. 30 Another study in New York City found a similar dose-232 response relationship between degrees of obesity and acute and critical care. 12 Patients less than 60 233 years old with BMI between 30 and 34.9 kg/m 2 (obesity class I) were 2.0 and 1.8 times more likely 234 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 18, 2021. ; https://doi.org/10.1101/2021.05.14.21257204 doi: medRxiv preprint to be respectively admitted for acute care (general hospital admission) and critical care (ICU 235 admission or invasive ventilator) compared to individuals with BMI < 30 kg/m 2 . Patients of the same 236 age group with BMI ≥ 35 kg/m 2 (obesity class II and III) showed 2.2 and 3.6 more chances of being 237 hospitalized for acute and critical care, respectively. 12 238

One of the greatest strengths of the study was the use of SIVEP-Gripe dataset. Because severe acute 240 respiratory syndrome is a condition of compulsory notification in both public and private hospitals, 31 241 we have a nationwide representative sample of patients hospitalized for severe COVID-19 in Brazil. 242

In addition, the large sample sizes allowed us to analyze adults and elders separately, as well as the 243 degrees of obesity which dose-response association with death was evidenced. The availability of 244 important confounding variables (sex, age, and preexisting comorbidities) to control the estimated 245 associations, as well as hospital outcomes and mortality of COVID-19, was another differential of 246 the study. Only patients with positive RT-PCR test for SARS-CoV-2 and final diagnosis for COVID-247 19 were included which gives greater precision on the studied population. The availability and use of 248 data from health surveillance systems may be a lesson from Brazil that other countries can learn for 249 obtaining routine and timely data to guide health systems and research in preparing and responding 250 to pandemics before and during their course. 251

The study also has some limitations that must be considered. Because this is a cross-sectional study, 252 a causal association cannot be inferred. As we used routinely collected data, which has not been 253 designed primarily for research purposes, it may bring well-known limitations related to missing, 254 underestimation, and potential misclassification. Obesity prevalence may have been underestimated 255 due to the completeness of obesity and BMI data. Previous studies using SIVEP-Gripe data have also 256 found a low prevalence of obesity in this population. 32, 33 Better routine collection of height and weight 257 data is still needed in clinical practice. Also, we believe that health professionals have adopted more 258 the one method to collect weight and height information for BMI calculation, such as the patient's 259 self-report and direct measure. Therefore, in addition to BMI which implausible values were checked 260 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 18, 2021. ; https://doi.org/10.1101/2021.05.14.21257204 doi: medRxiv preprint and excluded, the classification of obesity was also confirmed from a dichotomous variable on the 261 presence of obesity (no/yes). Although it is known that BMI does not distinguish between fat and 262 lean body mass, and thus may lead to misclassification bias, BMI has been shown as a strong predictor 263 of excess body fat and has been widely used in epidemiological studies. 15 Information for some 264 outcomes were incomplete because some patients were still hospitalized on the study endpoint date. 265

However, that did not represent a potential selection bias to our study. The analytical samples in each 266 outcome had similar demographic and clinical characteristics than the overall study population and 267 the excluded samples (Supplementary Table 1 ). Data on ethnicity/race was very incomplete, and thus 268 was not included in the analysis. Additional studies are needed to further explore the relationship 269 between socioeconomic characteristics and obesity in severe disease. Finally, the generalization of 270 results must be cautious since the study included only hospitalized cases of COVID-19. 271

The combined association of obesity, diabetes, and/or cardiovascular disease with severe COVID-19 273 outcomes, especially ICU admission and death, may be stronger in adult than in elderly inpatients. In 274 both age groups, obesity alone and obesity combined with DM and/or CVD had more impact on the 275 risk of all severe COVID-19 outcomes than the subgroup with DM and/or CVD. The study also 276 supports an independent relationship of obesity with the severe outcomes, including a dose-response 277 association between degrees of obesity and death in adults. These findings suggest important 278 implications for the clinical care of patients with obesity and severe COVID-19, such as the increased 279 need of critical care and higher risk of death among these patients. Our study also supports the 280 inclusion of people with obesity, independently of other preexisting comorbidities and age, in the 281 high-risk and vaccine priority groups for protection from SARS-CoV-2 infection. 282

The authors thank the members of Rede CoVida's Epidemiology & Information Group for the work 284 of identifying and collecting data related to . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 

Data is freely available without restriction at https://opendatasus.saude.gov.br/dataset/bd-srag-2020. 305

Code book and analytic code will be made available upon request from the corresponding author. 306 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 18, 2021. ; https://doi.org/10.1101/2021.05.14.21257204 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 18, 2021. ; https://doi.org/10.1101/2021.05.14.21257204 doi: medRxiv preprint Table 3 . Combined association of obesity, diabetes, and/or cardiovascular disease with non-invasive and invasive mechanical ventilation use, intensive care unit admission, and death in adult and elderly patients hospitalized with severe COVID-19. OB: obesity (BMI≥30 kg/m 2 ), DM: diabetes mellitus, CVD: cardiovascular disease, ICU: intensive care unit, PR: prevalence ratio, 95%CI: 95% confidence interval. * Crude and adjusted multinomial logistic regression models for mechanical ventilation use in adults (n=8075) and elders (n=11829). ** Crude and adjusted logistic regression models for ICU admission in adults (n= 8414) and elders (n=12222). *** Crude and adjusted logistic regression models for death in adults (n=6565) and elders (n=9943). # Adjusted for sex, age in years, pulmonary disease, asthma, kidney disease, hematologic disease, neurological disease, liver disease, and immunosuppression. Degrees of obesity defined by the WHO cutoff points. PR: prevalence ratio, 95%CI: 95% confidence interval. * Crude and adjusted multinomial logistic regression models for mechanical ventilation use (n=8075). ** Crude and adjusted logistic regression models for ICU admission (n=8414) and mortality (n=6565). # Adjusted for sex, age in years, diabetes mellitus, cardiovascular disease, pulmonary disease, asthma, kidney disease, hematologic disease, neurological disease, liver disease, and immunosuppression.

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) 

World Health Organization. WHO Coronavirus Disease (COVID-19) Dashboard

People Who Are at Increased Risk for Severe Illness

Comorbid Chronic Diseases are Strongly Correlated with Disease Severity among COVID-19 Patients: A Systematic Review and Meta-Analysis

Obesity as a risk factor for COVID-19: an overview

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

Obesity is Associated with Severe Forms of COVID-19

Early Transmission Dynamics in Wuhan, China, of Novel Coronavirus-Infected Pneumonia

Baseline Characteristics and Outcomes of 1591 Patients Infected with SARS-CoV-2 Admitted to ICUs of the Lombardy Region

Prevalence of obesity among adult inpatients with COVID-19 in France

Obesity is the comorbidity more strongly associated for Covid-19 in Mexico. A case-control study

Obesity is associated with worse outcomes in COVID-19: Analysis of Early Data From

Obesity in patients younger than 60 years is a risk factor for Covid-19 hospital admission

Predicting Mortality Due to SARS-CoV-2: A Mechanistic Score Relating Obesity and Diabetes to COVID-19 Outcomes in Mexico

High Prevalence of Obesity in Severe Acute Respiratory Syndrome Coronavirus-2 (SARS-CoV-2) Requiring Invasive Mechanical Ventilation

Worldwide trends in body-mass index, underweight, overweight, and obesity from 1975 to 2016: a pooled analysis of 2416 population-based measurement studies in 128.9 million children, adolescents, and adults

Health Effects of Overweight and Obesity in 195 Countries over 25 Years

Vigitel Brasil 2018: vigilância de fatores de risco e proteção para doenças crônicas por inquérito telefônico

Burden of disease in Brazil, 1990-2016: a systematic subnational analysis for the Global Burden of Disease Study

COVID-19 and obesity-lack of clarity, guidance, and implications for care

Severe Acute Respiratory Syndrome dataset -including COVID-19 data

World Health Organization. Physical status: the use and interpretation of anthropometry. Geneva: WHO

Encuesta sobre Salud, Bienestar y Envejecimiento (SABE) en América Latina el Caribe: Informe Preliminar (Internet)

Guidelines for the Collection and Analysis of Anthropometric Data in Health Services. Technical Standard for the Food and Nutrition Surveillance System

Body mass index and all-cause mortality among older adults

Impact of Obesity on Influenza A Virus Pathogenesis, Immune Response, and Evolution

COVID-19 and endocrine diseases. A statement from the

Features of 20 133 UK patients in hospital with covid-19 using the ISARIC WHO Clinical Characterisation Protocol: prospective observational cohort study

Multimorbidity patterns among COVID-19 deaths: proposal for the construction of etiological models

Factors associated with COVID-19-related death using OpenSAFELY

Morbid Obesity as an Independent Risk Factor for COVID-19 Mortality in Hospitalized Patients Younger than 50. Obesity (Silver Spring)

Influenza in Brazil: surveillance pathways

Epidemiological and clinical characteristics of the COVID-19 epidemic in Brazil

Ethnic and regional variations in hospital mortality from COVID-19 in Brazil: a cross-sectional observational study