Abstract
BACKGROUND: The prevalence of obesity and diabetes is rising. The aim of this study was to determine the association of body mass index (BMI) and waist circumference (WC) with type 2 diabetes mellitus (T2DM) in the elderly and to compare the discriminatory abilities of BMI, WC and other anthropometric indicators, including waist-to-height ratio (WHtR), body adiposity estimator (BAE) and body roundness index (BRI) for T2DM. METHODS: This cross-sectional study included 69,388 subjects aged ≥ 60 years living in Xinzheng, Henan Province, from January to December 2020. The data came from the residents' electronic health records of the Xinzheng Hospital Information System. Logistic regression was used to examine the relationships. Fully adjusted models adjusted for age, sex, place of residence, alcohol consumption, smoking, physical exercise, SBP and RHR. The area under the receiver operating characteristic curve (AUC) was used to compare the discriminatory ability of different anthropometric indicators for T2DM under the influence of potential risk factors. RESULTS: After adjusting for multiple covariates, compared with the first BMI quintile, the odds ratios (ORs) and 95% confidence intervals (CIs) from the second to fifth quintile for T2DM were 1.416 (1.335-1.502), 1.664 (1.570-1.764), 1.879 (1.774-1.990) and 2.156 (2.037-2.283), respectively. Compared with the first WC quintile, the ORs and 95% CIs from the second to fifth quintiles for T2DM were 1.322 (1.244-1.404), 1.549 (1.459-1.643), 1.705 (1.609-1.807) and 2.169 (2.048-2.297), respectively. Among men, compared with other anthropometric indicators (BMI, WHtR, BAE and BRI), WC showed the highest AUC (AUC: 0.629; 95% CI: 0.622-0.636). Among women, the AUCs of BMI (AUC: 0.600; 95% CI: 0.594-0.606), WC (AUC: 0.600; 95% CI: 0.593-0.606) and BAE (AUC: 0.600; 95% CI: 0.594-0.607) were similar, and the AUCs of BMI, WC and BAE were higher than WHtR, BRI. CONCLUSIONS: All anthropometric indicators were positively associated with T2DM. In men, WC with the strongest positive association with T2DM was the best predictor of T2DM. In women, BMI was most strongly associated with T2DM, and the predictive powers of BMI, WC and BAE were similar. After adjusting the potential confounding factors including age, sex, place of residence, alcohol consumption, smoking, physical exercise, SBP and RHR, the effect of these factors was eliminated, the findings were independent of the covariates considered.