Sociodemographic and Clinical Factors Associated With Radiation Treatment Nonadherence and Survival Among Rural and...

Cancer treatment nonadherence is associated with higher rates of cancer recurrence and decreased survival. Rural patients with cancer experience a 10% higher mortality rate compared with their nonrural counterparts; geographic differences in nonadherence may contribute to this increased mortality. The goal of this study was to assess for geographic disparities and determine sociodemographic and clinical factors associated with radiation therapy (RT) nonadherence and survival among rural and nonrural patients with cancer.

We examined cancer registry, medical records, and billing claims data at a safety net academic medical center. Geographic residence was defined as rural versus nonrural by US Department of Agriculture 2013 Rural-Urban Continuum Codes. Other factors assessed were age, sex, race, marital status, insurance type, employment, area median household income, residential distance to cancer treatment center, clinical stage, cancer type, treatment modality, total radiation dose received, and radiation dose per fraction. We used Cox proportional hazards modeling to examine 7 ways of operationalizing nonadherence and selected the definition that resulted in the best model fit statistics and prediction of mortality. Overall survival rates were estimated with the Kaplan-Meier method. We then examined nonadherence as the main exposure along with additional covariates in least absolute shrinkage and selection operator penalized survival analyses and as the outcome in our multivariable generalized linear regression analyses predicting nonadherence. We considered 2-way interaction terms with the main exposure, geographic residence.

We identified 3,077 patients with cancer who averaged 62 years old, were 59% female, 34% Black, and 14% rural. Twenty-two percent of patients missed at least 2 fractions and missed an average of 10% of their treatment plan. Rural patients experienced a higher mortality rate than nonrural patients (53% vs 42%, P < .0001). Survival was assessed through December 31, 2021, with a mean follow-up of 4.5 years. Proportion of missed fractions as the indicator of nonadherence provided the best model fit statistics and prediction of survival. Marital status, employment status, tumor, nodes, metastases stage, cancer type, and age at diagnosis significantly affected survival, in addition to a treatment delay by geographic residence interaction effect. Specifically, patients residing in rural areas who experienced a treatment delay were more than twice as likely to die as nonrural residents who also experienced a treatment delay, and nearly twice as likely to die as rural residents who did not experience a treatment delay. The 2-year survival rate was 76% for nonrural residents who did not experience a treatment delay versus 27% for rural residents who experienced a treatment delay. Patients who were widowed, had stage 4 cancer, or lung cancer were more likely to be nonadherent. Finally, patients residing in rural areas who experienced a treatment delay were more likely to subsequently be nonadherent.

In a geographically and racially diverse population, RT nonadherence is a significant concern that affects survival, yet it is a modifiable risk factor. We demonstrated that rural residence was associated with both RT nonadherence and poorer overall survival. Rural patients with a treatment delay had the lowest overall survival, compared with both nonrural survivors and rural survivors without delay. Rural residents who are delayed in starting treatment are at heightened risk for poor outcomes and should receive targeted support to mitigate the observed disparities. Additional patient populations that may benefit from targeted treatment adherence support include widowed patients and those with stage 4 cancer or lung cancer.