|Home||<< 1 >>|
Carslake, D., Davey Smith, G., Gunnell, D., Davies, N., Nilsen, T. I. L., & Romundstad, P. (2017). Confounding by ill health in the observed association between BMI and mortality: evidence from the HUNT Study using offspring BMI as an instrument. Int J Epidemiol, .
Abstract: Background: The observational association between mortality and body mass index (BMI) is U-shaped, leading to highly publicized suggestions that moderate overweight is beneficial to health. However, it is unclear whether elevated mortality is caused by low BMI or if the association is confounded, for example by concurrent ill health. Methods: Using HUNT, a Norwegian prospective study, 32 452 mother-offspring and 27 747 father-offspring pairs were followed up to 2009. Conventional hazard ratios for parental mortality per standard deviation of BMI were estimated using Cox regression adjusted for behavioural and socioeconomic factors. To estimate hazard ratios with reduced susceptibility to confounding, particularly from concurrent ill health, the BMI of parents' offspring was used as an instrumental variable for parents' own BMI. The shape of mortality-BMI associations was assessed using cubic splines. Results: There were 18 365 parental deaths during follow-up. Conventional associations of mortality from all-causes, cardiovascular disease and cancer with parents' own BMI were substantially nonlinear, with elevated mortality at both extremes and minima at 21-25 kg m-2. Equivalent associations with offspring BMI were positive and there was no evidence of elevated parental mortality at low offspring BMI. The linear instrumental variable hazard ratio for all-cause mortality per standard deviation increase in BMI was 1.18 (95% confidence interval: 1.10, 1.26), compared with 1.05 (1.03, 1.06) in the conventional analysis. Conclusions: Elevated mortality rates at high BMI appear causal, whereas excess mortality at low BMI is likely exaggerated by confounding by factors including concurrent ill health. Conventional studies probably underestimate the adverse population health consequences of overweight.