STUDY OBJECTIVES: To assess the relative contribution of age and social class to variations in the prevalence of a selection of self reported health problems. To examine the implications of observed variations for research on health inequalities. DESIGN: Secondary analysis of the Health Survey for England (1991-1997) using morbidities that are particularly prone to class effects. A statistical measure of the "relative class effect" is introduced to compare the effects of adjusting for social class and age. MAIN RESULTS: There is substantial variation in the relative importance of the age and class distributions of different diseases. Age effects often overshadow those of class even for conditions where an apparently strong social gradient exists. Only for self reported mental health among women does the social gradient exceed the age gradient. Within the context of a dominating age gradient, social gradients are relatively high for mental health and general health for both sexes. Variation in the relative strengths of the social gradients between the sexes are observed for angina symptoms. CONCLUSIONS: Given variations in the "relative class effect", analysis recognising the distinct contributions of age, sex, and social class to specific morbidities is advocated as a transparent and robust approach to the assessment of morbidity based inequality.
STUDY OBJECTIVES: To assess the relative contribution of age and social class to variations in the prevalence of a selection of self reported health problems. To examine the implications of observed variations for research on health inequalities. DESIGN: Secondary analysis of the Health Survey for England (1991-1997) using morbidities that are particularly prone to class effects. A statistical measure of the "relative class effect" is introduced to compare the effects of adjusting for social class and age. MAIN RESULTS: There is substantial variation in the relative importance of the age and class distributions of different diseases. Age effects often overshadow those of class even for conditions where an apparently strong social gradient exists. Only for self reported mental health among women does the social gradient exceed the age gradient. Within the context of a dominating age gradient, social gradients are relatively high for mental health and general health for both sexes. Variation in the relative strengths of the social gradients between the sexes are observed for angina symptoms. CONCLUSIONS: Given variations in the "relative class effect", analysis recognising the distinct contributions of age, sex, and social class to specific morbidities is advocated as a transparent and robust approach to the assessment of morbidity based inequality.