Objective: To compare established associations between risk factors and mortality in UK Biobank – a study with an exceptionally low response rate to its baseline survey – against those from studies that have conventional response rates.

Design: Prospective cohort study alongside individual-participant meta-analysis of other cohort studies.

Setting: United Kingdom.

Participants: The analytical sample in the main analyses in UK Biobank comprised 499,701 people (response rate 5.5%) while the pooling of data from the Health Surveys for England (HSE) and the Scottish Health Surveys (SHS) included 18 studies and 89,895 individuals (mean response rate 68%). Both study populations were linked to the same nationwide mortality registries and, for the purposes of the present analyses, the baseline age range was aligned at 40-69 years.

Main outcome measure: Death from cardiovascular disease, selected malignancies, and suicide. To quantify the difference between hazard ratios in the two studies we computed a ratio of the hazard ratios using HSE-SHS as the referent.

Results: Risk factor levels and mortality rates were typically more favourable in UK Biobank participants relative to the HSE-SHS consortium. For the association between risk factor and mortality endpoints, however, there was close agreement between studies. Based on 14,288 deaths during an average of 7.0 years of follow up in UK Biobank, and 7,861 deaths over 10 years of mortality surveillance in HSE-SHS, for cardiovascular disease mortality, for instance, the age- and sex-adjusted hazard ratio (95% confidence interval) for ever having smoked cigarettes (versus never) was 2.04 (1.87, 2.24) in UK Biobank and 1.99 (1.78, 2.23) in HSE-SHS, yielding a ratio of hazard ratios close to unity (1.02, 0.88, 1.19). These conclusions were essentially unchanged when we compared results separately for men and women, and aligned baseline years and censoring dates across studies.

Conclusion: Despite a low response rate, risk factor associations in UK Biobank appear to be generalisable.