(SLB-060) A Cross-Sectional Study of Race and Ethnicity Reporting and Representation among Cancer Patients in Select US Real-World Data Sources

Background: Understanding the demographic characteristics of patients contributing health information to real-world data (RWD) is critical to ensure equitable representation in healthcare research. However, race and ethnicity remain insufficiently captured in cancer patients across RWD sources.

Objectives: This study aimed to characterize the capture and representativeness of race and ethnicity in select US real-world databases across prevalent cancers.

Methods: This study explored the reporting rate and distribution of race and ethnicity data for select cancers between 2017 and 2021 across the Merative® MarketScan® Multi-State Medicaid Database (MDCD), Optum’s de-identified Clinformatics® Data Mart Database socioeconomic (SES), and Optum® de-identified Electronic Health Record data set (Optum® EHR), as well as the Surveillance, Epidemiology, and End Result (SEER) database. MDCD captures the healthcare service utilization of individuals covered by select state Medicaid programs. Optum CDM SES is derived from a database of administrative health claims for members of large commercial and Medicare Advantage health plans. Optum® EHR is a longitudinal EHR repository derived from US healthcare provider organizations aggregates data from integrated delivery networks and healthcare provider organizations. SEER database utilizes cancer registry data and covers approximately 48% of the US population with rigid capture of race and ethnicity data.

Results: The reporting rates of race and ethnicity varied across cancers, which were 38-40% and 34-35% in Optum CDM SES, and 86-93% and 84-90% in Optum® EHR, respectively. MDCD aggregated race and ethnicity into a single variable, with reporting rates between 78 and 87%. Compared with SEER data, the proportion of Black patients was higher in MDCD and Optum CDM SES for most cancers, with the exception of prostate cancer and multiple myeloma in Optum CDM SES. Conversely, in Optum® EHR Black patients were underrepresented across all cancers except acute lymphocytic leukemia and acute myeloid leukemia. The proportions of Asian patients were lower in all cancers in Optum CDM SES and Optum® EHR yet higher in prostate cancer and chronic lymphocytic leukemia in Optum CDM SES compared with SEER data. Notably, data for Asian population were not available in MDCD. The proportions of Hispanic/Latino were consistently lower for all cancers across all databases compared to SEER data.

Conclusions: Despite considerable effort in planning and implementation of programs to mitigate health disparities in healthcare, the capture of race and ethnicity data varies across RWD sources, and the corresponding distributions among patients whose data are captured is inconsistent with actual disease distribution in the US cancer registry data.