Cancer clinical trial participation in socioeconomically vulnerable patients; A risk model to aid in targeted interventions

Background

In patients with cancer, those with lower incomes are less likely to participate in clinical trials. A broad-based evaluation of variables that could contribute to this disparity has not been conducted.

Methods

We used data from Health Information National Trends Survey (HINTS) databases for survey years 2014, 2017, and 2020, the survey years that included questions about whether patients with cancer participated in a clinical trial. We examined 21 demographic, socioeconomic, behavioral, geographic, and health information questions. We derived a risk model to predict clinical trial participation using a training/validation approach with best subset selection and k-fold cross-validation. Logistic regression was used.

Results

We examined N = 1023 participants with household income <$75,000 (the U.S. median). In the training dataset (n = 614), a 5-variable model was identified including race/ethnicity, education, trust, anxiety/depression, and geographic locale. A quartile-level risk score was constructed based on the sum of adverse risk factors. In the validation cohort (n = 409), each increase in quartile level was associated with a 73 % increase in the odds of trial nonparticipation (OR = 1.73, 95 %-CI, 1.19–2.53, p = 0.004), indicating successful model validation. Among all individuals, trial participation rates decreased from 18.6 % to 7.5 % to 4.6 % to 2.8 %, respectively, as the number of adverse risk factors increased from 0 to 1 to 2 to 3 to 4–5.

Conclusions

We developed and validated a 5-variable risk model that identified a large set of lower-income individuals at lower risk of trial participation. These findings could aid in identification of patients who may benefit from additional support to navigate the treatment trial decision-making process.