Background: Health benefits of marriage have long been recognised and extensively studied but previous research has yielded inconsistent results for older people, particularly older women. At older ages accumulated benefits or disadvantages of past marital experience, as well as current marital status, may be relevant, but fewer studies have considered effects of marital history. Possible effects of parity, and the extent to which these may contribute to marital status differentials in health, have also been rarely considered.
Methods: We use data from the Office for National Statistics Longitudinal Study, a large record linkage study of 1% of the population of England & Wales, to analyse associations between marital history 1971-1991 and subsequent self-reported limiting long-term illness and mortality in a cohort of some 75,000 men and women aged 60-79 in 1991. We investigate whether prior marital status and time in current marital status influenced risks of mortality or long term illness using Poisson regression to analyse mortality differentials 1991-2001 and logistic regression to analyse differences in proportions reporting limiting long-term illness in 1991 and 2001. Co-variates included indicators of socio-economic status at two or three points of the adult life course and, for women, number of children borne (parity).
Results: Relative to men in long-term first marriages, never-married men, widowers with varying durations of widowerhood, men divorced for between 10 and twenty years, and men in long-term remarriages had raised mortality 1991-2001. Men in long-term remarriages and those divorced or widowed since 1971 had higher odds of long-term illness in 1991; in 2001 the long-term remarried were the only group with significantly raised odds of long-term illness. Among women, the long-term remarried also had higher odds of reporting long-term illness in 1991 and in 2001 and those remarried and previously divorced had raised odds of long-term illness and raised mortality 1991-2001; this latter effect was not significant in models including parity. All widows had raised mortality 1991-2001 but associations between widowhood of varying durations and long-term illness in 1991 or 2001 were not significant once socio-economic status was controlled. Some groups of divorced women had higher mortality risks 1991-2001 and raised odds of long-term illness in 1991. Results for never-married women showed a divergence between associations with mortality and with long-term illness. In models controlling for socio-economic status, mortality risk was raised but the association with 1991 long-term illness was not significant and in 2001 never-married women had lower odds of reporting long-term illness than women in long-term first marriages. Formally taking account of selective survival in the 20 years prior to entry to the study population had minor effects on results.
Conclusions: Results were consistent with previous studies in showing that the relationship between marital experience and later life health and mortality is considerably modified by socio-economic factors, and additionally showed that taking women's parity into account further moderated associations. Considering marital history rather than simply current marital status provided some insights into differentials between, for example, remarried people according to prior marital status and time remarried, but these groups were relatively small and there were some disadvantages of the approach in terms of loss of statistical power. Consideration of past histories is likely to be more important for later born cohorts whose partnership experiences have been less stable and more heterogeneous.