Senior Director Pfizer Inc. Brooklyn, United States
Background: Globally, awareness of the importance of diversity and equity in clinical trials is growing. In 2022, US President Joe Biden signed into law a bill that called on drug makers to submit “diversity action plans” that address equity with applications for new drug approvals. Understanding disease epidemiology using real world data (RWD) can help guide equity goals in diversity action plans. However, RWD in the US may underrepresent historically underserved racial and ethnic groups.
Objectives: We used a linear weighting method to adjust the underlying racial, ethnic, age, and sex distributions of patients with Type 2 Diabetes Mellitus (T2DM) in the Optum Electronic Health Records (EHR) to improve the representativeness of traditionally underserved racial and ethnic groups in these data.
Methods: Adult patients (≥18 years) with T2DM in the US Optum EHR were weighted to the demographics of the 2019 US Census. Specifically, weights were calculated as the proportion of each sex, age, race, and ethnicity strata in the US Census divided by the proportion of the sex, age, race, and ethnicity strata in the Optum EHR. Weights were then multiplied by stratified T2DM counts to generate weighted T2DM disease frequencies that reflected the sex, age, race, and ethnicity distributions of these populations in the 2019 US Census.
Results: According to the 2019 US Census, 16.41% of adults identify as Hispanic, but Hispanic patients accounted for only 7.72% of the T2DM adult patients in the Optum EHR. After weighting, the proportion of Hispanic patients among those with T2DM rose to 18.68%. Likewise, African Americans are 12.96% of the adult US population, 16.09% of Optum EHR T2DM patients before weighting, and 17.41% of T2DM patients after weighting. Finally, Asian adults are 6.12% of the US population, 2.13% of the Optum EHR T2DM patient population before weighting, and 5.46% of T2DM patients after weighting.
Conclusions: Before linear weighting, Hispanics accounted for a lower proportion of the T2DM patients in the Optum EHR than they do in the 2019 US Census. But, after linear weighting, Hispanics constituted a higher proportion of those with T2DM than in the Census data, which is consistent with literature suggesting that Hispanics have an elevated risk of T2DM compared to the majority White population. Weighting also increased the proportion of African Americans and Asian Americans with T2DM compared to the unweighted EHR data. Linear weighting is a simple, highly scalable method that may ameliorate the underrepresentation of some racial and ethnic groups in RWD. Thus, linear weighting may be a useful tool for setting representative, equitable goals for clinical trial recruitment.
