Presentation (CTI)

OBJECTIVES: FDA diversity requirements over time have resulted in numerous published studies regarding diversity planning; these use a range of sources. We compared three different data sources for diversity planning in female breast cancer: U.S. census data, Clarivate’s U.S. open claims data, and cancer registry data from Surveillance, Epidemiology, and End Results Program (SEER).
METHODS: We derived the US incidence of breast cancer by presence of key receptors (HER2+; HR-/HER2- [triple negative breast cancer; TNBC]) stratified by race/ethnicity using SEER data. We compared the results from SEER with the corresponding patient populations from US claims-linked EHR data for all breast cancer patients who had at least two claims in the given year and population census data.
RESULTS: US census data by race/ethnicity differed significantly (chi-squared p<0.05) from SEER data in patients with breast cancer, with notable differences in key receptors, except in non-Hispanic Asians with TNBC. Claims-linked EHR data also differed significantly from SEER data in patients with breast cancer (chi-squared p<0.05).
CONCLUSIONS: It is critical to validate epidemiological data sources prior to use in diversity planning. SEER is considered to be the gold standard for cancer epidemiology. Diversity planning in breast cancer should not be limited to census data; population data does not account for differences in risk of disease by race/ethnicity. The nuances of the claims and EHR datasets and their contributing components must be assessed for their suitability for diversity planning. Treatment setting should also be considered; claims prescription data may be an ideal source for addressing disparity by treatment setting relevant to the clinical trial, and should be explored in future work.