Intended for healthcare professionals


Understanding the relation between BMI and mortality

BMJ 2019; 364 doi: (Published 26 March 2019) Cite this as: BMJ 2019;364:l1219

Linked research

Body mass index and all cause mortality in HUNT and UK Biobank studies

  1. Kathryn E Bradbury, senior research fellow1,
  2. Benjamin J Cairns, senior statistical epidemiologist2
  1. 1National Institute for Health Innovation, School of Population Health, The University of Auckland, Auckland 1072, New Zealand
  2. 2MRC Population Health Research Unit, Nuffield Department of Population Health, University of Oxford, Oxford, UK
  1. Correspondence to: K E Bradbury k.bradbury{at}

Further evidence of causation, with a J shape that’s more pronounced in smokers

High body fatness is an important cause of ill health.1 In a linked paper, Sun and colleagues (doi:10.1136/bmj.l1042) use two cohort studies—the Norwegian HUNT Study and the UK Biobank—to investigate further the relation between body mass index (BMI) and mortality.2 BMI is often used as a simple proxy for body fatness, because most of the variation in BMI is due to variations in body fatness.3

Large observational studies consistently report J shaped associations between BMI and mortality.45678 Confounding by smoking and reverse causation (whereby diseases that lead to death may cause weight loss) are important sources of bias in this estimated relation. These biases shift the apparent optimum BMI upwards and exaggerate the increased mortality at low BMIs. Studies that have attempted to control for them have generally found an optimum BMI around 22-25 in most populations,4678 but important questions remain about the BMI-mortality relation, including: Is …

View Full Text

Log in

Log in through your institution


* For online subscription