Seven anthropometric indices of adiposity and incident cardiovascular disease in China and England: a harmonised two- cohort study

Background Body-mass index (BMI) does not capture body-fat distribution, and evidence comparing alternative measured adiposity indices for cardiovascular risk remains fragmented across single indices, single countries and self-reported measurements. We aimed to compare seven anthropometric indices head-to-head and to test whether the resulting risk signals and prediction models generalise between China and England. Methods Prospective analysis of two nationally representative ageing cohorts with measured anthropometry: CHARLS (China, n = 10,557) and ELSA (England, n = 5,231), yielding 15,788 adults aged ≥ 45 years free of cardiovascular disease (CVD) at baseline and 3,175 incident CVD events. Cox models with a pre-specified adjustment gradient estimated dose–response associations; mediation analysis, quantile g-computation, Fine–Gray competing-risk models and restricted-mean-time-lost analyses characterised mechanisms and absolute effects; and gradient-boosted models trained in one country were externally validated in the other with TreeSHAP interpretation. Results All seven indices showed independent, monotonic associations with incident CVD (fully adjusted top-versus-bottom tertile hazard ratios 1.13–1.49). Blood pressure and cardiometabolic disease jointly mediated 32.4% of the BMI–CVD association (China 40.6%; England 22.7%). A simultaneous one-tertile increase in the adiposity–blood-pressure mixture yielded a joint hazard ratio of 1.59 (95% CI 1.49–1.70). The highest BMI tertile lost 0.132 years of CVD-free time over 7 years. Prediction models transported bidirectionally (C-index 0.623 China→England; 0.618 England→China; combined-model 5-year time-dependent AUC 0.680 in China and 0.739 in England) with preserved discrimination but imperfect calibration. Adding the seven indices to conventional risk factors improved the C-index by 0.010 (95% CI 0.008–0.014). Conclusions Measured waist circumference and the weight-adjusted waist index should complement BMI in cardiovascular risk assessment. Adiposity signals generalise across populations with very different adiposity distributions, but absolute-risk deployment requires country-specific recalibration. Clinical trial registration Clinical trial number: not applicable.