In 2017,
The Lancet published a large study by Swiss epidemiologist Silvia Stringhini and her collaborators, analysing the impact of the most important causes of preventable death in Western societies. They estimated the number of years of life lost for each risk factor at the individual level and its contribution to preventable death at the societal level (PAF = Population Attributable Fraction). The multicohort study and meta-analysis used individual-level data from 48 independent prospective cohort studies with information on socioeconomic status, high alcohol consumption, physical inactivity, current smoking, hypertension, diabetes and obesity, and mortality, for a total population of 1,751,479 from seven high-income WHO member countries. A limitation of many studies of health risk factors is
confounding bias: many risk factors are interrelated and cluster together in high-risk populations. For example, low physical activity and obesity go hand in hand. People who are physically inactive tend to gain weight, and people who are severely obese have difficulty exercising. The unique advantage of the huge amount of individual data in the Stringhini study is that it allows (estimation of) the relative contribution of each separate risk factor. The following table shows that, at an individual level, smoking is the single greatest risk of avoidable death, followed by diabetes and high alcohol consumption. At the population level, diabetes and high alcohol consumption have a low prevalence. Physical inactivity, smoking and low socioeconomic status (SES) are then the top three preventable causes of early death. Smoking, physical inactivity and low SES account for almost two thirds of all avoidable deaths. A puzzling finding is the small contribution of obesity as a cause of avoidable premature death. There are two reasons why obesity is not an important independent risk factor, as is often assumed. • Being overweight is a risk for early death without correcting for confounding risk factors. Overweight is usually measured by the body mass index (BMI = kg/m2), which is much easier to measure than physical activity. Most studies only measured BMI, not physical activity, and did not correct for confounding. • A major pitfall in many studies of weight and health is that "normal" and "healthy" are often confused. The
WHO definition of "normal" adult BMI (between 18.5 and 24.9 kg/m2) is based on a normal weight and height distribution of US citizens in the 1960s, not on the associated risk of death in 2023. A meta-analysis of the association between BMI and mortality in 230 cohort studies with 3.74 million deaths among 30.3 million participants found that the risk of death in adults is not increased between 23 and 30 kg/m2 (see Figure 2). An adult BMI of 18.5 kg/m2, considered 'normal' by WHO criteria, is associated with a 30% increase in all-cause mortality. However, this is a measure of correlation, not causation, so it does not disprove previously held notions of the relationship between health and weight. ==United States==