Central obesity indices and the risk of heart disease, type 2 diabetes mellitus, and stroke: findings from a Chinese prospective cohort

Obesity is a major risk factor for cardiometabolic disease, but the relative value of different anthropometric indices for risk remains uncertain. Body mass index (BMI) is widely used but does not reflect fat distribution, whereas central obesity measures may better capture visceral adiposity. However, their associations with major cardiometabolic diseases have not been comprehensively compared in Asian populations. Using data from a large prospective cohort of Chinese adults, we compared the associations of BMI, waist circumference (WC), waist-to-height ratio (WHtR), body roundness index (BRI), a body shape index (ABSI), and weight-adjusted waist index (WWI) with incident heart disease, type 2 diabetes mellitus (T2DM), and stroke. Machine learning–assisted effect estimates were derived using a doubly robust Double Machine Learning framework and were complemented by Cox proportional hazards models. Restricted cubic spline analyses were used to explore nonlinear exposure–response relationships. Sensitivity analyses were conducted to assess the robustness of the findings. Most indices were positively associated with the outcomes, although the strength and statistical significance of the associations varied by index, outcome, and adjustment model. Waist-based central obesity indices, particularly WC, WHtR, and BRI, generally showed larger estimated effects and stronger associations with incident T2DM and stroke than BMI, whereas BMI showed associations comparable to those of central obesity indices for heart disease. In the primary confounder-adjusted model, the highest quartiles of BRI, WC, and WHtR were associated with hazard ratios above 4.0 for T2DM, whereas the corresponding hazard ratios ranged from 1.79 to 2.74 for heart disease and stroke. Associations were generally attenuated after additional adjustment in the clinical pathway-adjusted sensitivity analyses, with weaker and less stable findings for ABSI and WWI. Machine learning–assisted effect estimation and conventional survival analyses suggested that waist-based central obesity indices were more consistently associated with incident T2DM and stroke than BMI. Among the evaluated indices, WC, WHtR, and BRI showed broadly comparable performance and may provide useful complementary information for cardiometabolic risk stratification in Asian populations.