An epidemiological application of Popper's method.

The strong logicodeductive method of science is applied in an investigation of the mortality of a large population of communications workers (338 306 white men) to explain the mortality deficits so often observed in occupational studies. A theory for the healthy worker effect is proposed, and its predictions are tested in comparisons of mortality curves fitted to Weibull functions using regression techniques. The theory, based on the effect of selection on prevalent chronic disease, successfully predicts that as age increases the mortality rates of an employed population soon after hire will diverge from those of a general population. This new finding was observed for mortality from all causes, all causes minus accidents, cardiovascular diseases, and cancers. The theory successfully predicts the rate of decrease of the healthy worker effect as duration of service increases and the relative magnitudes of the effect by cause. Having survived initial empirical testing, this explanation for the effects of selection on occupational mortality meets at least two criteria (only one being sufficient) for deciding whether an explanation is to be preferred: it has made more precise predictions, and it has made new predictions not previously suggested.