Body-mass index and all-cause mortality: individual-participant-data meta-analysis of 239 prospective studies in four continents

Abstract

Background: Overweight and obesity are increasing worldwide. To help assess their relevance to mortality in different populations we conducted individual-participant data meta-analyses of prospective studies of body-mass index (BMI), limiting confounding and reverse causality by restricting analyses to never-smokers and excluding pre-existing disease and the first 5 years of follow-up.

Methods: Of 10 625 411 participants in Asia, Australia and New Zealand, Europe, and North America from 239 prospective studies (median follow-up 13·7 years, IQR 11·4-14·7), 3 951 455 people in 189 studies were never-smokers without chronic diseases at recruitment who survived 5 years, of whom 385 879 died. The primary analyses are of these deaths, and study, age, and sex adjusted hazard ratios (HRs), relative to BMI 22·5-<25·0 kg/m(2).

Findings: All-cause mortality was minimal at 20·0-25·0 kg/m(2) (HR 1·00, 95% CI 0·98-1·02 for BMI 20·0-<22·5 kg/m(2); 1·00, 0·99-1·01 for BMI 22·5-<25·0 kg/m(2)), and increased significantly both just below this range (1·13, 1·09-1·17 for BMI 18·5-<20·0 kg/m(2); 1·51, 1·43-1·59 for BMI 15·0-<18·5) and throughout the overweight range (1·07, 1·07-1·08 for BMI 25·0-<27·5 kg/m(2); 1·20, 1·18-1·22 for BMI 27·5-<30·0 kg/m(2)). The HR for obesity grade 1 (BMI 30·0-<35·0 kg/m(2)) was 1·45, 95% CI 1·41-1·48; the HR for obesity grade 2 (35·0-<40·0 kg/m(2)) was 1·94, 1·87-2·01; and the HR for obesity grade 3 (40·0-<60·0 kg/m(2)) was 2·76, 2·60-2·92. For BMI over 25·0 kg/m(2), mortality increased approximately log-linearly with BMI; the HR per 5 kg/m(2) units higher BMI was 1·39 (1·34-1·43) in Europe, 1·29 (1·26-1·32) in North America, 1·39 (1·34-1·44) in east Asia, and 1·31 (1·27-1·35) in Australia and New Zealand. This HR per 5 kg/m(2) units higher BMI (for BMI over 25 kg/m(2)) was greater in younger than older people (1·52, 95% CI 1·47-1·56, for BMI measured at 35-49 years vs 1·21, 1·17-1·25, for BMI measured at 70-89 years; pheterogeneity<0·0001), greater in men than women (1·51, 1·46-1·56, vs 1·30, 1·26-1·33; pheterogeneity<0·0001), but similar in studies with self-reported and measured BMI.

Interpretation: The associations of both overweight and obesity with higher all-cause mortality were broadly consistent in four continents. This finding supports strategies to combat the entire spectrum of excess adiposity in many populations.

Funding: UK Medical Research Council, British Heart Foundation, National Institute for Health Research, US National Institutes of Health.

Figure 1

Association of body-mass index with all-cause mortality, by geographical region Boxes are plotted against the mean BMI in each group. The HR per 5 kg/m² higher body-mass index (BMI) and its 95% CI are calculated only for BMI more than 25·0 kg/m². Analyses restricted to never-smokers without pre-existing chronic disease, excluding the first 5 years of follow-up. The reference category is shown with the arrow and is 22·5–<25·0 kg/m². CIs are from floating variance estimates (reflecting independent variability within each category, including reference). Areas of squares are proportional to the information content (ie, inverse of the floating variance). HR=hazard ratio.

Figure 2

Association of body-mass index with all-cause mortality, by baseline age group The HR per 5 kg/m² higher body-mass index (BMI) and its 95% CI are calculated only for BMI more than 25·0 kg/m². Analyses restricted to never-smokers without pre-existing chronic disease, and excluding the first 5 years of follow-up, and include data from all geographical regions. The reference category is shown with the arrow and is 22·5–<25·0 kg/m². CIs are from floating variance estimates (reflecting independent variability within each category, including the reference category). Areas of squares are proportional to the information content. Analyses by baseline age and the three main geographical regions are in the appendix (p 38). HR=hazard ratio.

Figure 3

Association of body-mass index with all-cause mortality, by sex The HR per 5 kg/m² higher body-mass index (BMI) and its 95% CI are calculated only for BMI more than 25·0 kg/m². Analyses restricted to never-smokers without pre-existing chronic disease, excluding the first 5 years of follow-up, and include data from all geographical regions. The reference category is shown with the arrow and is 22·5–<25·0 kg/m². CIs are from floating variance estimates (reflecting independent variability within each category, including reference). Areas of squares are proportional to the information content. Analyses by sex and the three main geographical regions (east Asia, Europe, and North America) are in the appendix (p 39). HR=hazard ratio.

Figure 4

Association of body-mass index with mortality, by major underlying cause The HR per 5 kg/m² higher body-mass index (BMI) and its 95% CI are calculated only for BMI more than 25·0 kg/m². Analyses restricted to never-smokers without pre-existing chronic disease, excluding the first 5 years of follow-up, and include data from all geographical regions. The reference category is shown with the arrow and is 22·5–<25·0 kg/m². CIs are from floating variance estimates (reflecting independent variability within each category, including reference). Areas of squares are proportional to the information content. Analyses of cause-specific mortality by three geographical regions (east Asia, Europe, and North America) are in the appendix (pp 41, 42).