Discordance Between BMI-Defined Overweight/Obesity and Waist-to-Height Ratio-Defined Central Obesity Among Adults with Type 2 Diabetes in Malawi: A Multicenter Cross-Sectional Study.

Abstract

Background: Body mass index (BMI) does not capture abdominal fat distribution and may classify adiposity differently from central-obesity indicators. We quantified discordance between BMI-defined overweight/obesity and waist-to-height ratio (WHtR)-defined central obesity among adults with type 2 diabetes mellitus (T2DM) in Malawi. Methods: This multicenter cross-sectional study included 1356 adults attending diabetes clinics at three tertiary hospitals. Overweight/obesity was BMI ≥ 25 kg/m². Central obesity was defined by waist circumference (WC ≥ 94/80 cm, men/women), waist-to-hip ratio (WHR ≥ 0.90/0.85), and WHtR ≥ 0.50. BMI and WHtR were cross-classified into four phenotypes, and agreement between the two classifications was summarized using observed agreement and Cohen's kappa. Logistic and multinomial logistic regression models estimated (adjusted) odds ratios with 95% CIs; sensitivity analyses added study site as a fixed effect and modified Poisson models with robust variance provided adjusted prevalence ratios for the common binary outcomes. Results: Prevalence of BMI-defined overweight/obesity was 60.8%; central obesity was 67.5% (WC), 65.3% (WHR), and 80.1% (WHtR). Cross-classification showed 23.7% had central obesity only and 56.3% had combined overweight/obesity and central obesity; observed agreement between the two classifications was 71.8% and Cohen's kappa was 0.35 (95% CI 0.30-0.40). Overweight/obesity was associated with older age (aOR 2.95), female sex (aOR 2.19), and physical inactivity (aOR 1.93). WC-defined central obesity was strongly associated with female sex (aOR 10.16), although the corresponding adjusted prevalence ratio was considerably smaller (aPR 2.25). WHR-defined central obesity was associated with older age, and anxiety. WHtR-defined central obesity was associated with older age, female sex, ever-married status, and late-evening eating behavior (aOR 1.36), although the association did not persist when study site was added. The combined phenotype was associated with age, sex, marital status, meal frequency, and physical inactivity. Conclusions: Overweight/obesity and central obesity were highly prevalent, with WHtR identifying the greatest burden and a substantial central-obesity-only phenotype that BMI alone would not detect. Incorporating WHtR alongside BMI may improve anthropometric assessment of adiposity in diabetes care, particularly among older and female patients.