Diet and physical activity

Physical activity has the cleanest new evidence in behavioural epidemiology. Measured activity from wearable devices predicts mortality better than any questionnaire; the relationship is graded down to very low levels; and the largest gains, in every study, are for moving from almost nothing to a little — a few thousand steps a day, a few minutes of brisk movement. Inactivity carries an attributable fraction of the order of five to eight per cent of deaths in the published estimates, and rises with age and with sedentary work. The surveys that track it are being replaced by devices whose wearers are healthier and richer than average, which flatters the trend.

Diet is where the evidence is broadest and least settled. The associations between dietary pattern and mortality are consistent across the large cohorts — more plants, less processed meat, less sugar — and the attributable estimates are large; but almost all of it is observational, self-reported and confounded by income. The live argument is ultra-processed food: whether the processing itself harms, or the fat, sugar and salt it carries, and the cohorts cannot separate the two. What has been shown to change diet at scale is reformulation under pressure — sugar levies halved the sugar in soft drinks in the countries that imposed them — with effects on weight that are modest and mostly measured in children. Those evaluations file under What changes behaviour at scale; obesity and diabetes as outcomes file under their own driver.

The two behaviours share a gradient and a measurement problem. Both are steeply patterned by deprivation, both are self-reported in most official series, and both are moving in directions the surveys are slow to show: activity falling with sedentary work and rising with the wearable-wearing middle class at once.

Status

established

Direction

mixed

Would change our view

A population-representative cohort with device-measured activity and mortality follow-up; a causal study separating processing from composition in diet; or a national activity or diet series turning on a policy that can be evaluated.

Trackers

Gaps

Measured rather than reported consumption and activity almost anywhere; diet data of any quality outside a few cohorts; device data from Japan, Korea and the Nordic countries; children, where the trend matters most and is measured least.