Study population and setting
This large case series included patients aged 18 and older who were diagnosed with COVID-19 in 238 hospitals across the United States from March 1 to December 31, 2020. The data were drawn from the Premier Healthcare Database Special COVID-19 Release, a hospital-based electronic database that includes patients from all health insurance/payer groups; data from the subset of hospitals (238/~800) that recorded height and weight were used. The primary analytic sample included adults with ICD-10 code confirmed COVID-19 diagnosis in either the emergency department (ED) or inpatient setting, and with non-missing height and weight. Multivariable logit regression analyses evaluated the association between body mass index (BMI; weight in kg/height in m^2) and four outcomes: 1) inpatient hospitalization (reference group: those seen in ED only), 2) intensive care unit (ICU) admission, 3) invasive mechanical ventilation, and 4) death. The analysis for inpatient hospitalization included all adults diagnosed in the ED or inpatient, while the remaining 3 outcomes were assessed among hospitalized patients. BMI was assessed as both a continuous and categorical variable; the latter grouped by underweight (<18.5 kg/m^2), healthy weight (18.5-24.9 kg/m^2), overweight (25-29.9 kg/m^2), and four classes of obesity (30-34.9; 35-39.9; 40-44.9; ≥45 kg/m^2).
Summary of Main Findings
Among the 148,494 patients with COVID-19 in the hospital/ED setting (median age 55 years; 46% male; 51% non-Hispanic white), 51% were obese, 28% were overweight, 9% were healthy weight, and 2% were underweight. Approximately half—71,491—of these COVID-19 patients were hospitalized (median age 65 years; 51% male; 56% non-Hispanic white), with similar BMI category distributions. Hospitalization was elevated for adults with obesity and underweight relative to those of normal weight, and risk increased with category of obesity, independent of age, sex, race/ethnicity, payer type, geospatial characteristics of the hospital (urbanicity and US Census region), and admission month. Similarly, when assessed as a non-linear continuous variable, BMI demonstrated a J-shaped association with hospitalization, ICU admission, and death. Among hospitalized patients, the risk of ICU admission was significantly elevated only in the two highest obesity classes relative to healthy weight, while the risk of mechanical ventilation increased in a graded relationship across all BMI categories. The risk of death among hospitalized patients increased with obesity severity; patients with the highest obesity class had a 2-fold greater risk of mortality relative to those with healthy weight. Most of the inferences were observed both among those younger than aged 65 years and those aged 65 years and older, although the effect sizes and uncertainty around them were larger in the younger age group. The magnitude of the risks attenuated after accounting for comorbidities (hypertension, diabetes, chronic kidney disease, asthma, coronary atherosclerosis or other heart disease, chronic obstructive pulmonary disease, and cancer), but the ordering of risk was retained.
The large sample size provided sufficient power to assess BMI across seven categories (including four obesity classes) and to stratify the analyses above and below aged 65 years to assess potential differences in the risks of BMI by age group. The authors performed rigorous multivariable analysis, including sensitivity analyses to assess robustness to adjustment for comorbidities and to address missing BMI with a multiple imputation approach.
The hospitalization outcome in this study, assessed only among patients in the hospital or ED, is not comparable to hospitalization in studies with population-based cases of COVID-19, because those seeking care in the ED are likely to differ in significant ways from the overall population with COVID-19 (i.e. more likely to have severe symptoms; less likely to have a primary healthcare provider; more likely to be underweight or obese). The authors justified not adjusting for known risk factors for COVID-19 severity–such as heart disease, lung disease, and diabetes–because these conditions may be partially caused by obesity, but it might be more balanced to consider the effect of BMI as falling between the primary results and the more modest results from the sensitivity analyses that adjusted for these comorbidities.
This study assessed the relationship between BMI and four COVID-19 severity outcomes in more granularity than has been possible in previous smaller studies, demonstrating the relative effects of seven mutually exclusive BMI categories, and showing graded increases in risk with increasing obesity severity for most outcomes. The study also evaluated BMI as a non-linear continuous variable, identifying specific BMI values in the healthy to overweight range as inflection points with the lowest risk of severe COVID-19 events using polynomial spline models. The large sample size supported age-stratified analyses, and demonstrated that obesity remained a strong risk factor for mortality and invasive mechanical ventilation among patients aged 65 and older as well as the younger population.
This review was posted on: 23 March 2021