Is self-selection the main driver of positive interpretations of general health checks? The Inter99 randomized trial

Abstract

Objective: To investigate if the lower mortality among participants of a health check followed by lifestyle intervention of high risk persons is explained by self-selection.

Methods: All persons residing in the study area (Copenhagen; Denmark) were randomized to intervention (n=11,629) or control group (n=47,987). Persons in the intervention group were invited for a health check and individual lifestyle counselling. At baseline, 52.5% participated. Differences between participants and control group in 10-year all-cause and disease specific mortality was assessed. In survival analyses we controlled for socio-demography and mental and physical health.

Results: Mortality rates were highest among non-participants and lowest among participants in the intervention group, whereas mortality rates of controls were approximately the average of those of participants and non-participants. In adjusted analyses the difference in both all-cause mortality (HR 0.73, 95% CI 0.62-0.85) and disease specific mortality was diminished between participants and controls, and turning insignificant as regard cancer mortality. In total, 39% of the lower mortality among participants in the intervention group was explained by self-selection according to socio-demography and health.

Conclusion: A substantial part of the lower mortality among participants in a preventive health check can be explained by confounding. The remaining lower mortality is most probably due to residual confounding, as the lower mortality was seen both for lifestyle related and non-lifestyle related diseases.

Keywords: Life style intervention; Mortality; Participation rate; Population level; RCT; Selection bias.