Race and ethnicity representation in phase 2/3 oncology clinical trial publications in the United States: A systematic review.

20 Background: The five 1997 Office of Management and Budget (OMB) races in the US include American Indian or Alaskan Native (AI/AN), Asian, Black or African American, Native Hawaiian or Other Pacific Islander (NHPI), and White, with Hispanic ethnicity. Despite the Affordable Care Act (ACA) mandating OMB-based collecting/reporting standards, race and ethnicity publishing in medical journals is inconsistent, despite being necessary to achieve health equity. We aimed to quantify race and ethnicity reporting rates and calculate representation quotients (RQs) in published oncology clinical trials. Methods: In this systematic review, PubMed/Embase were queried for phase 2/3 clinical trials of the 6 most common noncutaneous solid malignancies, published between 2012-2022, in 4 high-impact journals. Trial characteristics were recorded. The RQs for each race and ethnicity were calculated by dividing the percent of representation in clinical trial publications by the percent of year-matched, site-specific incident cancers in the US, compared with Kruskal-Wallis tests with Bonferroni Correction (BC). Reporting was compared between journal publications and ClinicalTrials.gov. Results: Among 1,202 publications evaluated, 364 met inclusion criteria: 16 JAMA , 241 JCO , 19 Lancet , and 88 NEJM . Publications included 268,209 patients (171,132 women 64%), with a median of 356 (IQR, 131-800) patients per publication. Reported race and ethnicity included AI/AN in 52 (14%) publications, Asian in 196 (54%), Black or African American in 215 (59%), Hispanic in 67 (18%), NHPI in 28 (8%), and White in 254 (70%). Median RQ varied across race (P < .001 BC) with 1.04 (IQR, 0.09-4.77) for Asian, 0.98 (IQR, 0.86-1.06) for White, 0.42 (IQR, 0.12-0.75) for Black or African American, and 0.00 (IQR, 0.00-0.00) for both AI/AN and NHPI patients. Sensitivity analyses showed similar findings upon subset analysis for US-only clinical trials. There was significantly less race and ethnicity reporting in the clinical trial publications compared with ClinicalTrials.gov documentation for AI/AN (14% vs 45%, P < .001 per McNemar χ 2 with continuity correction MC) and NHPI (7.7% vs 43%; P < .001 MC). Conclusions: While most phase 2/3 oncology clinical trials published in high-impact journals report race and ethnicity, most did not report AI/AN and NHPI racial categories. Our findings support a call to action for consistent journal policies and transparent race and ethnicity reporting, in alignment with ACA-concordant race and ethnicity federal reporting requirements.