Abstract

STUDY OBJECTIVES: (1) To assess the extent to which death of a spouse causes excess mortality by controlling for the effects of confounding and other sources of bias. Three possible sources of bias are considered: accidents common to spouses, common socioeconomic environment, and common lifestyles. (2) To assess the duration specific effects of death of a spouse on mortality. DESIGN AND SETTING: Prospective study of mortality in Finland among all 35-84 year old married Finnish men and women (1,580,000 people). Baseline sociodemographic measurement from the 1985 census records. Follow up by computerised record linkage to death certificate registers for the period 1986-91 (about 116,000 deaths, of which almost 10,000 among the bereaved) using personal identification codes. RESULTS: (1) After controlling for confounding effects, excess mortality was 17% in men and 6% in women. (2) Excess mortality was higher for short durations than long durations of bereavement. (3) Excess mortality after bereavement was higher in men than women. CONCLUSIONS: Controlling for confounding does not seem to have a crucial modifying effect on the relationship between spousal bereavement and mortality. It seems that death of a spouse has a causal effect on mortality. However, although spousal bereavement is a major stressful life event, this causal effect seems to be relatively small and short lived.