This article is focussed on the following two questions: first, whether socio-economic status (SES) differentials in health among the aged vary by the indicators of health and SES applied. Secondly it is investigated how SES differentials in health vary by age. The analyses are based on a telephone survey of 821 older people (> or = 60 years) living in a private household in Germany. In addition to the traditional indicators of SES (education, income and occupational status) two alternative indicators (assets and home ownership) are used in the study. Self-rated health, depression (CES-D) and functional status according to Advanced Activities of Daily Living are introduced as health indicators. Results of multiple regression analyses show that associations between income and the health indicators were comparably strongest. Education, occupational prestige, assets and home ownership were not consistently related to health, especially after controlling for the remainder of the SES indicators. Thus, home ownership and assets do not predict health significantly beyond the effect of the traditional indicators of SES. With regard to age-related variations of the social gradient results were not consistent: In terms of self-rated health associations do not differ between the age groups (60-65, 66-74, 75+), while associations between traditional SES indicators and depression tend to become stronger among those aged 75 years and older. With regard to functional status there were no significant associations with socio-economic status in the multivariate model among those aged 75 years and older. The present study shows that the magnitude of the association between social inequality and health among the aged as well as age related changes in the social gradient depend on the indicators of SES and health applied. Because the analyses are based on cross-sectional data it is not possible to identify causal relationships or to separate age effects from cohort effects. Therefore longitudinal studies are needed in which traditional as well as alternative indicators of SES are used that are appropriate for older adults in general and important for an explanation of differentials in health among the aged in particular.
