The contribution of common causes of death to social class differences in life expectancy, 1993-99

Myer Glickman and Chris White, Office for National Statistics, and Alex Scott-Samuel, University of Liverpool

[Project number 20088]

Substantial social class differences in life expectancy and mortality in England and Wales persisted at the end of the 20th century. Analyses of differences between social classes in cause of death can help to improve understanding of the causal factors that lead to observed inequalities in overall mortality. For example, male social class gradients in ischaemic heart disease and cerebrovascular disease increased over the nineties as rapid reductions in the higher social classes outpaced trends in the lower social classes. Such findings are valuable for health inequalities policy formulation and evaluation.

Although methods for analysis of differences in mortality by major cause are straightforward, meaningful attribution of a "share" of the social class gap in mortality to specific causes is problematic on grounds of competing or substitute mortality.

In this analysis, we will use simple demographic methods to estimate the contribution of each major cause of death to overall social class differences in life expectancy in England and Wales in 1993-96 and 1997-99. Abridged life tables will be constructed for each social class, men and women separately, in each time period. To estimate the "crude" contribution of a cause of death to the overall inequality in life expectancy, we will subtract the probability of death from that cause from the total probability of death at each age. We will then redistribute the "saved" group across the probabilities at that age of survival, or of death from any other condition, based on the observed age, sex and social class-specific mortality rates. The resulting changes in life expectancy for each social class will be used to produce corrected estimates of the contribution in years of that cause to the total inequality in life expectancy. From this, we will identify key implications of the problem of competing mortality for health inequalities policy, and also discuss the limitations of the methods used, and potential for further research.

The LS is the only longitudinal data source representative of the England and Wales population that contains linked census and death events, and enables social class assignment from census records. This overcomes the problems of proxy reporting of occupation and employment status at time of death registration leading to potential bias in the analysis.