Does unemployment cause mortality?

Previous observational follow-up studies of individuals have shown that mortality rates among the unemployed are higher than among the employed (Moser et al. 1984, Iversen et al. 1987, Martikainen, 1990, Morris et al. 1994). Two mechanisms may explain this excess mortality. (1) Causal effects of unemployment: Becoming jobless and prolonged redundancy have negative effects on health and increase the risk of premature death. The causal effects of unemployment are mainly assumed to be mediated through increased psychosocial stress, tobacco and alcohol consumption, as well as loss of income and material deprivation. (2) Selection: Persons likely to become unemployed, or to have difficulty in re-employment, have pre-existing ill-health and/or "lifestyle" (e.g. tobacco and alcohol consumption, diet), socioeconomic (e.g. social class, housing tenure) or personal characteristics (e.g. age, sex, physical weakness, psychological characteristics, and early life experiences) that increase the risk of future ill-health and mortality. In the following I will present three approaches that have been used to determine whether the relationship between unemployment and mortality is causal. All analyses are based on the 1980, 1985 or 1990 census records of all Finnish men which were linked to information on unemployment and death records in the period 1981-1993 by the Statistics Finland. Only about 0.3 per cent of registered death could not be matched to a census record All data are analysed by means of Poisson regression analysis, and the GLIM statistical package is used in fitting the models (Aitkin et al. 1989).