Patients Enrolled In Clinical Trials With Medicaid Or No Insurance And Overall Survival Benefits From New Interventions.

119 Background: Very few new treatments tested in randomized phase III cancer clinical trials show an overall survival (OS) benefit. Understanding whether the benefits are consistent among all patient groups is critical for informing guideline care. However, within trials, there is limited power to examine demographic or insurance subsets, especially in under-represented groups. Here we systematically examine whether positive treatment effects apply to all demographic and insurance groups. Methods: We examined all SWOG treatment trials completed from 1985-2014, and pooled data from trials with a statistically significant (p < .05) OS benefit for the experimental treatment. We conducted multivariable Cox regression analyses adjusting for clinical risk level and stratifying by study-specific 2-year OS probability ( < 25% vs. 25-75% vs. > 75%). Interaction tests determined whether OS hazard ratios (HRs) differed by 5 prespecified factors including age ( < 65 vs. ≥65), race (black vs. other), sex, ethnicity (Hispanic vs. non-Hispanic), and insurance status in patients < 65 years (private insurance vs. Medicaid/no insurance). Results: We identified 16 trials, containing 9,328 patients, with statistically significantly better OS in the experimental arm. The OS HR did not differ by sex (p = .60), race (p = .68), or ethnicity (p = .66). However, patients ≥65 years had a smaller treatment benefit than patients < 65 (HR = 1.24 vs HR = 1.37, p = .03), and patients with Medicaid/no insurance had a smaller treatment benefit than private insurance patients (HR = 1.22 vs. HR = 1.70, p = .03). Conclusions: The magnitude of the OS benefit from experimental treatments may not be uniform among demographic and insurance status groups. Older age patients and those with suboptimal insurance are at greater risk of experiencing competing risks of death, which can reduce power to identify the benefits of new experimental therapies. Trial designs should account for the risk of non-cancer deaths in important patient groups. Future research should examine whether reduced treatment impact is due to a higher comorbid burden in older patients, and to worse adherence to protocol care in those with limited financial resources.