Approaches to measuring health disparities that are unaffected by the prevalence of an outcome

As the overall prevalence of an outcome changes, binary measures of differences between rates of experiencing or avoiding the outcome tend also to change. As an adverse outcome declines in prevalence, relative differences between rates of experiencing it tend to increase while relative differences between rates of avoiding it tend to decline. As an outcome changes from being nearly universal to being rare, absolute differences will increase for a time and then decline, while differences measured in odds ratios exhibit a pattern that is the opposite of the absolute difference pattern. These tendencies have been universally overlooked in analysis of the size of racial and socioeconomic disparities in health and healthcare outcomes. But they create serious obstacles to the use of binary measures to determine whether disparities are changing over time or are otherwise larger in one setting than another. Relying on actual and hypothetical data on rates at which groups experience certain outcomes, this session would explain the tendencies and the obstacles they create for measurement of group differences. It would also explore two approaches for appraising the sizes of group differences in different settings that avoid these obstacles. The first involves identifying departures from the standard patterns of changes in various measures as an outcome increases or decreases in overall prevalence. The second involves the identification of the size of the disparity in each of two settings based on the difference between means of a hypothesized underlying normal distribution of factors associated with an outcome.