Socio-economic inequalities in mortality. Methodological problems illustrated with three examples from Europe.

BACKGROUND: Studies from most European countries have been able to demonstrate that lower socioeconomic groups have higher risks of disease, disability and premature death. Uncertain is, however, whether these studies have also been able to estimate the precise magnitude of these inequalities, their patterns and their trends over time. The purpose of this paper is to illustrate the extent to which results of descriptive studies can be biased due to problems with the data that are commonly available to European countries.
METHODS: Three illustrations are presented from a project on socio-economic inequalities in premature morbidity and mortality in Europe. These illustrations concern three problems often encountered in data on social class differences in mortality among middle aged men: the numerator/denominator bias in cross-sectional studies (illustrated for France), the exclusion of economically inactive men (illustrated for 4 countries), and the use of approximate social class schemes (illustrated for Sweden).
RESULTS: In each illustration, inequalities in mortality among middle aged men could be demonstrated, but data problems appeared to bias estimates of the precise magnitude of inequalities in mortality, their patterns by social class and cause of death, and their trends over time. The bias was substantial in most cases. Usually, it was difficult to predict in which ways and to what extent inequality estimates would have been biased.
CONCLUSIONS: When the aim of a study is to determine the precise magnitude, patterns or time trends of health inequalities, the results should be evaluated carefully against a number of potential data problems. Investments are needed, e.g. in data sources and in the measurement of socio-economic status, to secure that future studies can describe socio-economic inequalities in health in Europe in more detail and with more reliability.