Statistical methods to examine contributors to racial disparities in perinatal outcomes

Abstract

Longstanding racial disparities in preterm birth in the US remain incompletely understood. Often investigators use interaction terms or stratify epidemiologic analyses by race and ethnicity to examine disparities. However, these approaches presume differential susceptibility to similar exposures. However, American life remains largely racially segregated. As such, vastly different doses of harmful and beneficial exposures exist across racial and ethnic groups. Differences in exposure patterns and their sources likely explain a larger proportion of racial health disparities (mediation) than differential responses to exposures by race (effect modification). Thus, recently developed, user-friendly mediation analysis may be a more relevant and powerful tool to quantify the contribution of specific exposures to racial disparities. Such statistical methods coupled with evaluation of real-world reduction of exposures to toxicants may be used to focus policymakers’ efforts to improve perinatal health equity through targeted interventions.