Inclusiveness and ethical considerations for observational, translational, and clinical cancer health disparity research

Abstract

Background Although general trends in cancer outcomes are improving, racial/ethnic disparities in patient outcomes continue to widen, suggesting disparity‐related shortcomings in cancer research designs. Methods Using convenience sampling, a total of 24 data sources, representing several research designs and 5 high‐burden tumor types, were included for analyses. The percentages of races/ethnicities across each design/tumor type were compared with those of the 2017 US Census data. The authors used a framework based on the Belmont principles to evaluate the ethical strengths and/or weaknesses of each design. Results In all designs, white individuals were found to be overrepresented. African American and Asian individuals were underrepresented, and Native Americans had consistently poor or no representation. In general, ethical concerns varied according to the study design. Clinical trials were high with regard to respect for persons and beneficence but low for equitable subject selection related to the inclusion of race/ethnicity. Observational study designs were more inclusive for race/ethnicity compared with clinical and translational studies, but their clinical usefulness was less. Conclusions The authors proposed that ethical concerns should vary according to the study design. Because observational designs have strengths in inclusiveness for race/ethnicity, their clinical usefulness can be improved by extending the Learning Health System in hospital catchment populations, the use of health records linked to biospecimens, and minority oversampling. Likewise, minority enrollment into clinical trials can be improved through Learning Health System linking and by using National Cancer Institute–mandated Community Outreach and Engagement Cores. This will allow precision medicine for otherwise overlooked minority subgroups.

Patterns of National Cancer Institute‐Sponsored Clinical Trial Enrollment in Black Adolescents and Young Adults

“…We were unable to account for number of clinical trials open for AYA enrollment during this time period so further work is needed. 40 , 41 Lower Black AYA and overall enrollment in 2014, when the NCI Clinical Trials Network was restructured, may have been secondary to a decrease in the overall number of available trials. As expected, we identified differences in CTT enrollment among the AYAs with those 15–19y enrolling more often than those 20–39y.…”

Addressing racial/ethnic disparities in cancer clinical trials: Everyone has a role to play

“…In a review of 24 data sources conducted in 2017, White individuals were found to be overrepresented compared to other racial/ethnic groups who were consistently underrepresented across all cancer research study designs for 5 high‐burden tumor types. Notably, equitable subject selection was higher in observational studies compared with interventional clinical trials 5 . This echoes the reality that clinical trial recruitment is complex and requires a deeper appreciation of the multiple factors that impact participation of racial/ethnic minorities.…”

Part 1: Artificial intelligence technology in surgery

“…Effective measures used to capture relevant information should prioritize the inclusion of minority races that often are under-represented in, for example, large clinical trials to avoid bias. 43 Data sets are the lifeblood of algorithms and statistical modelling on which AI systems are trained. Researchers and clinicians have a duty of care to ensure that the provenance and quality of the data used to train these algorithms are impeccable and curated in compliance with federal regulatory guidelines.…”

“…This requirement should be implicit for the design of AI systems in any surgical specialty with patient recruitment as the first step in the data acquisition process for clinical trials. Effective measures used to capture relevant information should prioritize the inclusion of minority races that often are under‐represented in, for example, large clinical trials to avoid bias 43 …”