Predictors of mortality among a national sample of elderly widowed people: analysis of 28-year mortality rates

Design and setting: face to face home interview survey across England. Measures: physical, psychological, social, and socio-economic status and circumstances. Results: excess risk of mortality, which had been noted up to six months post bereavement among males aged 75+, had disappeared. In contrast to findings up to 13 years post-bereavement, neither psycho-social factors, social circumstances nor social class independently predicted differentials in mortality when analysed up to 28 years post-bereavement. The most significant, independent predictors, up to the 28-year term, were, as would be expected, male sex, older age, poorer physical functioning, and expressed 'relief at the death of the spouse'. When the sample was split by duration of widow(er)hood male sex and older age retained significance. Conclusion: the increasing frailty of the sample overall, and reduced statistical power in split-sample analyses, may explain the loss of significance of physical functioning and 'expressed relief at the death' in the split-sample results. The psycho-social risk factors for mortality after bereavement reduce over time, although further examination of expressed relief would be worthwhile.