Social class mortality differentials: artefact, selection or life circumstances?

Data from 10 years follow up of mortality in the OPCS Longitudinal Study are used to relate deaths of men in 1976-81 to their social class as recorded by the 1971 census. Explanations of social class mortality differentials are critically reviewed in the light of these new data. The similarity between the class differentials observed for men aged 15-64 years in this study and those reported in the 1970-2 Decennial Supplement on Occupational Mortality indicate that the published gradients were not in fact grossly distorted by numerator denominator biases. Distortions to gradients observed in the early years of the longitudinal study and ascribed to selective health related mobility out of employment from the principal social classes to the permanently sick had largely worn off after five years of follow up. Sharp gradients at ages over 75 years, similar to those at younger ages, suggest that, for men aged over 50 years, selective health related mobility between social classes does not contribute to differentials in mortality.