Six anthropometric indices and all-cause mortality in ageing populations: an observational cohort study in six longitudinal studies

BACKGROUND: Body-mass index (BMI) does not capture fat distribution and may contribute to the apparent obesity paradox in older adults. We compared six anthropometric indices in relation to all-cause mortality across multiple ageing cohorts. METHODS: We analyzed data from six longitudinal studies in 25 countries (CHARLS, HRS, ELSA, MHAS, SHARE, KLoSA). Adults aged 45 years or older with baseline anthropometry and mortality follow-up were evaluated (n = 89,745). Exposures were BMI, waist-to-height ratio (WHtR), body roundness index (BRI), A Body Shape Index (ABSI), weight-adjusted waist index (WWI), and conicity index (CI). Cox models estimated hazard ratios (HRs) per 1-SD increase with adjustment guided by a prespecified directed acyclic graph. Non-linear associations were examined with floating absolute risks and restricted cubic splines. RESULTS: During a median follow-up of 10 years, 16,593 deaths occurred. BMI was weakly inversely associated with all-cause mortality (HR per 1-SD 0.96, [95% CI 0.93-0.98]), attenuating after exclusion of early deaths (0.97, [0.94-1.00]). ABSI (1.15, [1.11-1.19]) and CI (1.09, [1.06-1.13]) showed a consistent positive association. Dose-response patterns differed: BMI, BRI, and WHtR followed a shallow U-shape, whereas ABSI, WWI, and CI increased approximately linearly without a threshold. CONCLUSIONS: In older adults, indices capturing central adiposity-particularly ABSI-are more consistently associated with higher all-cause mortality than BMI. Routine assessment of waist and height, and communication of simple shape metrics, could improve risk stratification and counselling in clinical and public health practice. CLINICAL TRIAL NUMBER: Not applicable.