P36 Is it good, or bad, for health to stop working at older ages? Evidence from the health and employment after fifty (HEAF) study

Abstract

20). An index of social isolation (0-7), with higher scores meaning greater isolation, was generated at baseline and each follow-up from seven indicators, each worth 1 point: living alone, less than monthly face-to-face or telephone contact with children/family/friends, not being a member of any organisations, not working, not volunteering. The completecase sample was restricted to those with IC and isolation scores at baseline and no missing data on covariates. Sequential growth curve models included the predictors age and sex, isolation, and then covariate blocks of socioeconomic factors, health behaviours, health; a fully-adjusted model included all predictors and covariates. Results In the unconditional model, average IC score at baseline was 7.06, decreasing 0.23 units over each follow-up to 6.38 at the final wave. In a model including isolation, age, and sex, higher isolation was associated with lower baseline IC (beta=-0.23, 95% CI=-0.28 --0.18) but not associated with the rate of change of IC. This effect remained after adjusting for socioeconomic factors, health behaviours and self-rated health. Conclusion These results suggest social isolation has a detrimental effect on IC level, but not on the rate of decline over time. However, this initial complete-case analysis should be expanded with imputation of missing data and more complex modelling of IC trajectories. Nevertheless, the study highlights the potential of this novel IC model to monitor IC over time and explore factors detrimental to healthy ageing.