Measuring Health Disparities and Communicating Findings

In Healthy People 2010 (HP2010), health disparities were assessed using univariate analyses within the socio-demographic domains of race/ethnicity, sex, education, income, geographic location, and disability status. Further, disparities were calculated as percent (i.e., relative) differences between the best group rate and each of the other group rates within a domain. For comparability across objectives, objectives that were expressed in terms of favorable health outcomes were re-expressed using the adverse outcome for the purpose of computing disparity. For Healthy People 2020 (HP2020), we are considering alternative methods to measure disparities. For example, the odds ratio, which is widely used in epidemiology, can reveal the multivariate nature of health disparities (e.g., by race/ethnicity and sex). A related measure is the parity achievement exponent (PAE), the exponent that would need to be applied to the odds against the favorable health outcome in any given population group to reach the odds in the group with the best rate. Though less commonly used, the PAE eliminates the need to express objectives in terms of adverse outcomes. When considering methods to measure disparities, the challenge remains to develop accompanying language that concisely yet accurately conveys findings to general audiences, including policy-makers and the public.