Suboptimal Insurance Linked to Worse Survival Outcomes in Positive Clinical Trials

Get Permission

Findings from a recent SWOG study could cast doubt on the generalizability of treatment effects observed in positive clinical trials, especially among underinsured patients. According to data presented at the 2019 ASCO Quality Care Symposium,1 patients with Medicaid or no insurance had no observed benefit from experimental treatment in a large set of trials with positive overall survival findings for experimental therapy. In contrast, positive treatment effects were observed for all other patient subgroups.

The analysis of patient-level clinical data over several decades found that this pattern was uniformly observed in all demographic, temporal, and clinical groups. Moreover, the influence of insurance status on both progression-free survival and overall survival was observed early within the first year of treatment and was sustained at least through 7.5 years.

Joseph M. Unger, PhD

Joseph M. Unger, PhD

“The magnitude of treatment benefits from experimental therapies may not be uniform among demographic and insurance status groups,” said lead author of the study, Joseph M. Unger, PhD, Assistant Member of the Cancer Prevention Program at Fred Hutchinson Cancer Research Center, in Seattle. “Patients with Medicaid or no insurance are likely at greater risk of experiencing competing risks of death, which can reduce power to identify the benefits of new experimental therapies. Trial designs going forward should account for the risk of noncancer deaths in important patient subgroups.”

As Dr. Unger explained, although phase III trials provide the highest level of evidence for demonstrating the efficacy of new treatments, only one in four randomized phase III cancer clinical trials is positive for the specified endpoint, and fewer still show a benefit for overall survival. What’s more, said Dr. Unger, single trials have insufficient statistical power to prove that the benefit of the new treatment applies to the major demographic, socioeconomic, and clinical subgroups.

“The generalizability of treatment effects to these subgroups is often simply assumed, and few, if any, distinctions about treatment efficacy within subgroups are made in cancer treatment guidelines,” he observed.

Insurance Status and Patient Outcomes

For this study, Dr. Unger and colleagues used data from the SWOG Cancer Research Network to identify all treatment trials completed in 1985 or later in which the experimental therapy was superior to standard therapy in terms of overall survival (P < .05). These data were then pooled across the different trials. The researchers categorized insurance status as private vs Medicaid or no insurance, which has been shown to be an indicator of both reduced access to health-care services and low socioeconomic status at the patient level, said Dr. Unger. He also noted that multivariable Cox regression models were used to identify whether survival patterns by arm differed by demographic and insurance status factors.

As Dr. Unger reported, researchers identified 19 trials that had positive overall survival in the experimental arm between 1984 and 2012, with a total sample size of more than 11,000 patients. These trials represented a diverse set of cancers, and patients were well balanced between standard and experimental arms, said Dr. Unger, who noted that 11% of patients were black and 18% of patients overall had Medicaid or no insurance.

“While overall survival treatment effects did not differ by age, sex, or race/ethnicity, patients with private insurance had a much greater benefit from the experimental therapy, and in each instance, this interaction value was highly statistically significant,” said Dr. Unger. “One key takeaway is that underinsured patients had no observed benefit from the experimental therapy in trials where everybody else essentially had that benefit.”

Suboptimal Insurance Impacts Outcomes

Dr. Unger and colleagues also examined progression-free survival to assess whether the patterns were due solely to non–cancer-related deaths and found similar discrepancies. Finally, the researchers truncated follow-up time—starting at 0.1 year out to 7.5 years—to determine when the interaction between insurance status and treatment became notable.

“The influence of insurance status manifests itself very quickly within the first year of follow-up,” said Dr. Unger. “From that point on, there’s nearly a constant influence of insurance status on treatment effects over time. Although this effect is more meaningful for overall survival than it is for progression-free survival, the effect is statistically significant for both endpoints within the first year.”

SWOG Trials, 1985–2014

This study of SWOG treatment trials completed between 1985 and 2014 has found that patients with Medicaid or no insurance had a smaller treatment benefit than private insurance patients (hazard ratio [HR] = 1.22 vs HR = 1.70, P = .03).

According to Dr. Unger, future research should examine whether reduced treatment impact is due to limited access to supportive services or worse adherence to protocol care in those with limited financial resources.

“Even in a setting in which patients are uniformly staged and treated under protocol care, suboptimal insurance influenced both cancer and noncancer outcomes,” Dr. Unger concluded. “These findings may inform policy regarding continued resource support for patients beyond their initial treatment.” 

DISCLOSURE: Dr. Unger reported no conflicts of interest. This study was sponsored by the National Cancer Institute.


1. Unger JM, Blanke CD, LeBlanc ML, et al: Patients enrolled in clinical trials with Medicaid or no insurance and overall survival benefits from new interventions. 2019 ASCO Quality Care Symposium. Abstract 119. Presented September 6, 2019.