3724 - Medicare Advantage Representation at U.S. Proton Therapy Centers Relative to Catchment-Weighted Benchmarks

Medicare Advantage (MA) enrollment exceeds half of Medicare beneficiaries, yet MA representation at proton beam therapy (PBT) centers relative to local market benchmarks remains poorly characterized. We quantified MA representation among PBT patients, assessed whether under-representation is concentrated at NCI-designated centers, and examined the role of community socioeconomic vulnerability.

Using a national all-payer claims database linked to the CDC's Social Vulnerability Index (SVI), we identified 28,686 unique Medicare beneficiaries treated with PBT at 38 US centers (2021-2024). Centers were stratified by NCI-designation status (NCI and non-NCI) and SVI of their respective catchment areas. To account for geographic variation in MA penetration, we defined a 100-mile catchment area for each center (50/150-mile sensitivity analyses were similar). Expected MA share was estimated using CMS Medicare enrollment data, volume-weighted by year. Analyses were limited to patients whose vendor-defined origination ZIP code fell within the catchment to mitigate bias from regional/national referrals. The primary endpoint was the center’s Representation Ratio (Rc), defined as the observed MA share divided by the catchment-expected MA share (Rc < 1 indicates under-representation). We compared Rc values between NCI-designated and non-NCI centers and assessed "access dyads" (center pairs within the same Metropolitan Statistical Area [MSA]) to compare centers with shared catchments and similar MA penetration.

The median center-level Rc was 0.56 (IQR 0.10-0.98; Wilcoxon signed-rank P<0.01 vs 1). The median Rc for NCI-designated centers (0.39; IQR 0.08–0.85) was numerically lower than non-NCI centers (0.72; IQR 0.39–1.32), but between-group difference did not reach statistical significance (Mann-Whitney P=0.07). Catchment-area SVI was not associated with MA representation (Spearman rho= -0.01, P=0.94). Dyad analyses revealed discordance (absolute difference in Rc >1) between centers sharing the same market, indicating that local MA penetration alone does not explain variation.

MA beneficiaries are substantially under-represented at PBT centers relative to catchment benchmarks, with a non-significant trend toward higher gaps at NCI-designated centers. Socioeconomic vulnerability did not explain between-center variation, and discordance within shared MSAs indicates variation is unlikely to be driven by geography alone. Together, these findings suggest that variation in MA access is unlikely to be driven by geography or community vulnerability alone and may reflect institution and payer-level factors. Achieving equitable access to advanced radiation technologies may require reforms to payer administrative and coverage practices, not infrastructure expansion alone.