The role of selective mortality in the dynamics of SES-related health inequality across the lifecycle

The life course literature on the social gradient in health has been dominated by the cumulative advantage and age-as-leveller hypotheses, with selective mortality also recognised as a potentially important confounder in older cohorts. The main contribution of this paper is to establish a unified framework to fully account for the changing social gradient in terms of a sufficient set of mobility indices characterising the co-evolution of the joint distribution of socioeconomic status and health within any particular cohort. The main innovation is to identify selective mortality effects using a counterfactual health distribution for the start of the study period in the absence of those who are known to die before the end, rather than for the end of the period if there had been no deaths since the start which requires the imputation of the ‘would be’ health of non-survivors. Using longitudinal data for Great Britain, selective mortality is found to be an important driver of social gradient changes within older cohorts, contrary to the findings of a number of previous studies. We explain this contrast by demonstrating how estimates of selective mortality effects are affected by the choice of counterfactual health distribution and socioeconomic status measure.