Validating the use of body mass index with computed tomography in a racially and ethnically diverse cohort of patients admitted with SARS-CoV-2.

Abstract

Body mass index (BMI) is criticized for being unjust and biased in relatively healthy racial and ethnic groups. Therefore, the current analysis examines if BMI predicts body composition, specifically adiposity, in a racially and ethnically diverse acutely ill patient population. Patients admitted with SARS-CoV-2 having an evaluable diagnostic chest, abdomen, and/or pelvic computed tomography (CT) study (within 5 days of admission) were included in this retrospective cohort. Cross-sectional areas (centimeters squared) of the subcutaneous adipose tissue (SAT), visceral adipose tissue (VAT), and intramuscular adipose tissue (IMAT) were quantified. Total adipose tissue (TAT) was calculated as sum of these areas. Admission height and weight were applied to calculate BMI, and self-reported race and ethnicity were used for classification. General linear regression models were conducted to estimate correlations and assess differences between groups. On average, patients (n = 134) were aged 58.2 (SD = 19.1) years, 60% male, and racially and ethnically diverse (33% non-Hispanic White [NHW], 33% non-Hispanic Black [NHB], 34% Hispanic). Correlations between BMI and SAT and BMI and TAT were strongest revealing estimates of 0.707 (0.585, 0.829) and 0.633 (0.534, 0.792), respectively. When examining the various adiposity compartments across race and ethnicity, correlations were similar and significant differences were not detected for TAT with SAT, VAT, or IMAT (all P ≥ 0.05). These findings support the routine use of applying BMI as a proxy measure of total adiposity for acutely ill patients identifying as NHW, NHB, and Hispanic. Our results inform the validity and utility of this tool in clinical nutrition practice.