How do we assess a racial disparity in health? Distribution, interaction, and interpretation in epidemiological studies

Abstract

Identifying the exposures or interventions that exacerbate or ameliorate racial health disparities is one of the fundamental goals of social epidemiology. Introducing an interaction term between race and an exposure into a statistical model is commonly used in the epidemiologic literature to assess racial health disparities and the potential viability of a targeted health intervention. However, researchers may attribute too much authority to the interaction term and inadvertently ignore other salient information regarding the health disparity. In this article, we highlight empirical examples from the literature demonstrating limitations of overreliance on interaction terms in health disparities research; we further suggest approaches for moving beyond interaction terms when assessing these disparities. We promote a comprehensive framework of three guiding questions for disparity investigation, suggesting examination of the group-specific differences in (1) outcome prevalence, (2) exposure prevalence, and (3) effect size. Our framework allows for better assessment of meaningful differences in population health and the resulting implications for interventions, demonstrating that interaction terms alone do not provide sufficient means for determining how disparities arise. The widespread adoption of this more comprehensive approach has the potential to dramatically enhance understanding of the patterning of health and disease and the drivers of health disparities.

Keywords: Health disparities; Interaction; Interpretation; Modification; Race; Regression.

Figure 1

A hypothetical scenario demonstrating an equal exposure distribution by racial group but different relationships between exposure and outcome for each racial group. The exposure is indicated along the x-axis and the outcome along the y-axis. The simplified, dichotomous race classification is denoted by a solid line for the Black sample and a dashed line for the White sample. The circles represent the proportion of each racial group at each level of exposure so that equally sized circles indicate that equal proportions of the racial group are exposed and unexposed. Larger circles represent a larger proportion exposed (or unexposed).

Figure 2

Four example interaction analysis scenarios highlighting two important axes: columns indicating the presence of an interaction (yes or no), and rows indicating whether the exposure contributes to a disparity in the outcome (yes or no). The four quadrants depict scenarios representing: A) interaction is present, but the exposure does not contribute to the disparity; B) interaction is not present with an exposure distribution that contributes to the disparity; C) interaction is present, but the exposure does not contribute to the disparity; and D) interaction is not present, and the exposure does not contribute to the disparity. The exposures are indicated along the x-axes and the outcomes along the y-axes. The simplified, dichotomous race classification is denoted by a solid line for the Black sample and a dashed line for the White sample. The circles represent the proportion of each racial group at each level of exposure so that equally sized circles indicate that equal proportions of the racial group are exposed and unexposed. Larger circles represent a larger proportion exposed (or unexposed).

Figure 3

Case study of a racial health disparity interaction analysis found in the literature, examining the relation between diabetes mellitus and uncontrolled blood pressure and whether this association varies by race. This represents a scenario where the interaction term indicates that the exposure-outcome relationship is stronger among the more advantaged group (captured by the steeper slope of the dashed line), yet the high exposure prevalence among the more disadvantaged group suggests that the exposure may be disparity-producing nonetheless. The exposure is indicated along the x-axis and the outcome along the y-axis. The simplified, dichotomous race classification is denoted by a solid line for the Black sample and a dashed line for the White sample. The circles represent the proportion of each racial group at each level of exposure so that equally sized circles indicate that equal proportions of the racial group are exposed and unexposed. Larger circles represent a larger proportion exposed (or unexposed).