Radiotherapy Deserts and the Radiation Oncology Alternative Payment Model: Risk to and Underrepresentation of Regions with Increased Palliative Need in Late-Stage Prostate Cancer

Purpose: Our previous work across all stages of prostate cancer has demonstrated that “radiotherapy (RT) deserts”--US counties demonstrating mismatch between cancer-specific outcomes and RT resource density—are located in disproportionately rural and Black communities. Given that the Radiation Oncology Alternative Payment Model (RO-APM) may particularly impact palliative RT through intensification of pre-existing barriers to access and delivery, we sought to investigate the interplay between late-stage prostate cancer (PC), RT resource variables and sociodemographic factors in relationship to counties participating in the model.

Methodology: At the US county level, we linked US HRSA Area Health Resources, USDA Economic Research Service, AMA Physician Masterfile, Medicare Provider and Service Files, and NIH Cancer State Profile datasets to build an ecologic study model. Provider and utilization per person at risk densities (PPR and UPR respectively) were defined as the number of attending RT physicians and numbers of billed Medicare beneficiaries per 100,000 persons at risk (males >/= 35 years). Associated counties were identified by zip codes designated within the final RO-APM. Logistic regressions evaluated relationships between late-stage PC incidence and metro (vs. nonmetro) status per rural-urban continuum definitions, %poverty, %race, PPR and UPR densities, and participating RO-APM counties.

Results: Analysis of 3141 counties revealed a mean (standard deviation) late-stage PC incidence rate per 100,000 of 23.1 (6.7), with PPR and UPR densities of 2.1 (SD 5.9) and 192.6 (557.6), respectively. Increasingly rural status, more racially homogeneous communities, and lower UPR density (β coefficients 1.6, 2.4-2.6, and -0.001, respectively, p<0.05), was associated with higher late-stage PC incidence on univariable analysis, but not %poverty or PPR density. Participating RO-APM counties exhibited greater odds of metro status and higher %Black, as well as higher PPR and UPR densities (OR 4.37, 3.39, 2.50, 2.34 respectively, p<0.05). However, RO-APM counties had lower incidence of late-stage PC in comparison to non-participating counties (OR 0.39, p<0.05). On multivariable analysis controlling for rural-urban continuum and %poverty, RO-APM participation remained associated with higher %Black population and metro status.

Conclusions: Characterization of participating RO-APM counties shows greater inclusion of those with higher RT resources and utilization, with decreased representation of counties with higher late-stage PC incidence. This mismatch highlights underrepresentation within the RO-APM of a patient population with elevated need for and likely benefit from palliative RT, raising further concern for the impact of the model on this important form of RT, accounting for nearly 50% of treatment delivered.

References: