Body mass index, waist circumference and waist-hip ratio: which is the better discriminator of cardiovascular disease mortality risk?: evidence from an individual-participant meta-analysis of 82 864 participants from nine cohort studies

Abstract

Few studies have examined both the relative magnitude of association and the discriminative capability of multiple indicators of obesity with cardiovascular disease (CVD) mortality risk. We conducted an individual-participant meta-analysis of nine cohort studies of men and women drawn from the British general population resulting in sample of 82 864 individuals. Body mass index (BMI), waist circumference (WC) and waist-to-hip ratio (WHR) were measured directly. There were 6641 deaths (1998 CVD) during a mean of 8.1 years of follow-up. After adjustment, a one SD higher in WHR and WC was related to a higher risk of CVD mortality (hazard ratio [95% CI]): 1.15 (1.05-1.25) and 1.15 (1.04-1.27), respectively. The risk of CVD mortality also increased linearly across quintiles of both these abdominal obesity markers with a 66% increased risk in the highest quintile of WHR. In age- and sex-adjusted models only, BMI was related to CVD mortality but not in any other analyses. No major differences were revealed in the discrimination capabilities of models with BMI, WC or WHR for cardiovascular or total mortality outcomes. In conclusion, measures of abdominal adiposity, but not BMI, were related to an increased risk of CVD mortality. No difference was observed in discrimination capacities between adiposity markers.

Figure 1. Derivation of the analytic sample

HSE: Health Survey for England, SHS: Scottish Health Survey

Figure 2. Age and sex-adjusted associations between anthropometric variables and mortality Upper row of figures is for all-cause mortality, lower for CVD mortality

For the association between each anthropometric variable and the outcome of interest, boxes are for the effect estimate (hazard ratio) and the vertical bars about are for the 95% confidence interval (CI). On the horizontal axis, the hazard ratio for each quintile is plotted against the quintile specific median of the anthropometric variable. For each figure, the hazard ratio (HR) and the 95% CI for a standard deviation for a one higher standard deviation in each anthropometric variables are shown. The p-value is for the log-linearity of the association. The vertical axis is on log scale to allow equal distribution of the CI about the estimates.

Figure 3. Multivariable adjusted associations between anthropometric variables and mortality

Conventions are as per figure 2. Covariates in the models are age, sex, cohorts, diabetes status, systolic blood pressure, history of CVD, social class, physical activity, smoking status, total cholesterol