Subjective life expectancy and actual mortality: results of a 10-year panel study among older workers

Hanna van Solinge, Kène Henkens

This research examined the judgemental process underlying subjective life expectancy (SLE) and the predictive value of SLE on actual mortality in older adults in the Netherlands. We integrated theoretical insights from life satisfaction research with existing models of SLE. Our model differentiates between bottom-up (objective data of any type) and top-down factors (psychological variables). The study used data from the first wave of the Netherlands Interdisciplinary Demographic Institute Work and Retirement Panel. This is a prospective cohort study among Dutch older workers. The analytical sample included 2278 individuals, assessed at age 50-64 in 2001, with vital statistics tracked through 2011. We used a linear regression model to estimate the impact of bottom-up and top-down factors on SLE. Cox proportional hazard regression was used to determine the impact of SLE on the timing of mortality, crude and adjusted for actuarial correlates of general life expectancy, family history, health and trait-like dispositions. Results reveal that psychological variables play a role in the formation of SLE. Further, the results indicate that SLE predicts actual mortality, crude and adjusted for socio-demographic, biomedical and psychological confounders. Education has an additional effect on mortality. Those with higher educational attainment were less likely to die within the follow-up period. This SES gradient in mortality was not captured in SLE. The findings indicate that SLE is an independent predictor of mortality in a pre-retirement cohort in the Netherlands. SLE does not fully capture educational differences in mortality. Particularly, higher-educated individuals underestimate their life expectancy.