,
9 tweets,
2 min read
Read on Twitter
Our paper on body mass index (BMI) and all-cause mortality using Mendelian randomization with non-linear methods is out in BMJ today: bmj.com/content/364/bm…. Thread follows...
Overall, we saw a J-shaped relationship between BMI and mortality, with genetically increased BMI leading to average reduction in mortality for underweight individuals, and an average increase in mortality for overweight and obese individuals with nadir around BMI 22-25. But...
But when we stratified on smoking status, a J-shaped relationship was only observed in ever-smokers. For never-smokers, the relationship between BMI and mortality was always-increasing (ie monotone increasing), with no evidence of harm of being underweight.
Additionally, when splitting by cause of death (cardiovascular/cancer/non-cardiovascular non-cancer), a J-shaped relationship was only observed for non-cardiovascular non-cancer mortality - the main cause of death in this category was respiratory disease.
This suggests that previous observational analyses suggesting a benefit of increased BMI (or even of being slightly overweight) have been driven by confounding and reverse causation.
Two potential explanations for increased harm in underweight smokers: 1) underweight smokers are smoking themselves thin and so are more unhealthy than average, or 2) there is genuine benefit of having slightly increased BMI for respiratory disease.
Technical point: to minimize influence of confounding and reverse causation, we as far as possible compare like-with-like - all comparisons are between participants who, if they had the same genotype, would be in the same centile of the BMI distribution.
Within each centile of this distribution, we consider the impact on mortality of having genetic variants that predispose individuals to having increased BMI - this allows us to consider the average causal effect of a lifetime increase in BMI within strata of the BMI distribution.
In conclusion, our analyses suggest that, for middle-aged/early late-aged never-smokers, increased BMI leads to increased mortality risk across the whole distribution of BMI. /END
