Background/Purpose: The FDA places emphasis on studying clinically relevant trial populations but does not address how these might be defined. SLE disproportionately affects women and people of African descent. A marked underrepresentation of people of African descent in lupus nephritis clinical trials has been reported, but this has not been investigated in SLE more broadly. The study aims to compare the demographics of individuals with SLE from 2 distinct RWD sources with completed North America (NA)-only SLE trials and propose statistical parameters for cohort sizes to support trial planning.

Methods: Data were evaluated from the TriNetX Analytics Network containing electronic health records (EHRs) from >150 million individuals in the US. Demographics (gender, race, ethnicity) were assessed for living individuals with an ICD-10-CM code M32.9 SLE in the last 5 years that had received healthcare services. Published data from the US Centers for Disease Control and Prevention (CDC) National Lupus Registries were also evaluated. Binomial confidence intervals were calculated for demographic cohorts from each data source. Data were compared to average proportions of the same cohorts from completed industry-sponsored, NA-only, SLE trials reported in clinicaltrials.gov.

Results: RWD Sources: EHR – 154,760 individuals met eligibility (87.0% Female, 9.5% Male, 3.4% Unknown Gender, 55.8% White, 25.6% Black or African American [BAM], 3.1% Asian, 15.4% Other/Unknown Race, 9.6% Hispanic/Latino [HL]). Registries – 204,295 individuals were estimated as living with SLE in the US in 2018 (90.2% Female, 9.8% Male, 59.5% White, 31.1% BAM, 5.8% Asian, 3.5% Other Race, 20.9% HL).
Clinical Trials: Data from 10 trials were analyzed (1190 subjects [M = 119, SD = 127]). Gender was reported for 10 (93.1% Female, 6.9% Male). Race was reported for 9 (1160 subjects [M = 129, SD = 130], 52.4% White, 40.7% BAM, 1.8% Asian, 5.1% Other/Unknown. Ethnicity was reported for 3 (142 subjects [M = 47, SD = 12]), 20% HL.
Cohort Modelling: Based on an analysis of binomial confidence intervals, a hypothetical NA SLE trial of 100 subjects would be statistically representative (p < .05) of the EHR-derived SLE population if it included a range of 80-94 Females, 4-15 Males, 46-66 White, 17-34 BAM and 4-15 HL patients. The same trial would be representative of the registry-derived population if it included a range of 84-96 Female, 4-16 Male, 50-69 White, 22-40 BAM and 13-29 HL patients.

Conclusion: Registries and large EHR databases are potential data sources for the rapid identification of “real world” demographics and the development of diversity plans for clinical trials. However, the SLE registry-derived model would necessitate a higher range of BAM and HL patients vs. EHR. Both methods have their advantages and limitations. EHR data was obtained in “real-time”, but was restricted to administrative codes. The registry-derived population may be considered more clinically representative of US SLE patients, but are estimates ~5 years old.  Notably, our data shows a higher proportion of BAM patients in NA-only SLE trials compared to EHR and registry-derived populations. The proportion of HL trial patients was consistent with the registry data, but higher than the EHR-derived model.

« Back to ACR Convergence 2024

ACR Meeting Abstracts - https://acrabstracts.org/abstract/modeling-diversity-in-systemic-lupus-erythematosus-sle-clinical-trials-using-real-world-data-rwd-sources/