Study population and setting
This study included 770 adult patients (61% male, mean age 63.5 years) with laboratory-confirmed SARS-CoV-2 infection admitted to two hospitals in New York City between March 4 and April 9, 2020 (with clinical outcomes followed up to April 16, 2020u). Patient characteristics, including demographics, medical history, laboratory results, and outcomes were reviewed retrospectively from medical records. The exposure of interest was body mass index (BMI), categorized as normal, underweight (<18.5), or obese (>30), and the outcome was a composite of ICU admission or death. Risk ratios were estimated with multivariable generalized linear models using a Poisson distribution.
Summary of Main Findings
Patients with obesity were more likely to be younger, and white or African-American. Comorbidities such as hypertension, diabetes, cardiovascular disease, chronic kidney or liver disease were not associated with BMI category; underweight patients had higher concentrations of d-dimer and troponin I. Compared to normal weight patients, obese patients were more likely to present with fever, cough, and dyspnea. 196 patients were admitted to the ICU, 88 patients died, and 241 either died or were admitted to the ICU. In multivariable analysis controlling for age, race/ethnicity, racial category, and troponin I levels, obesity was associated with a greater risk of ICU admission or death compared to patients in the normal weight category (RR: 1.58, 95% CI: 1.18 to 2.13). Results were robust to several sensitivity analyses, including restriction to patients over 60 years of age.
There were a reasonably large number of admitted patients and outcomes, and the authors performed some sensitivity analyses.
Although multiple comorbidities such as diabetes and hypertension were not associated with BMI categories to a statistically significant degree, the multivariable analyses should have considered them as possible confounding variables, since: 1) they are known to be associated with obesity, and 2) they are known to be associated with COVID-19 severity. Characterization of comorbidities also relied on medical records and did not involve assessment at time of admission (e.g., history of diabetes vs. current hemoglobin A1C levels). The broad categorization of BMI corresponds to established cutoffs, but may obscure more nuanced relationships between risk and BMI. Follow-up was as short as one week for some patients and thus may miss subsequent ICU admissions or deaths.
This study provides some evidence for obesity as an independent risk factor for COVID-19 severity.
This review was posted on: 10 July 2020