Practice-Level Spending Variation for Radiation Treatment Episodes Among Older Adults With Cancer.

Importance: Radiation treatments are an essential but expensive component of cancer care. Objective: To elucidate trends in radiation spending and identify factors associated with practice-level variations across the US health care system to inform alternative payment model design. Design, Setting, and Participants: This population-based cross-sectional study analyzed fee-for-service Medicare beneficiaries from 2009 to 2020. Patients were continuously enrolled in fee-for-service Medicare Parts A and B during the 1 year prior through 28 days after a radiation treatment episode. Data were analyzed from January 2023 to September 2024. Exposures: Medicare beneficiaries with cancer who received radiation therapy. Main Outcomes and Measures: Radiation treatment-specific standardized spending and utilization during 90-day treatment episodes were examined and characterized by radiation type (conformal, intensity modulated, stereotactic, proton, or brachytherapy) and number of fractions. Linear regression models with practice random effects to understand practice-level variation in standardized radiation spending were estimated. Variables were added to adjust for year, patient demographics, cancer type, geography, radiation technology, and number of fractions per episode. Results: From 2009 to 2020, 1 898 864 beneficiaries with cancer (mean [SD] age, 74 [8.4] years; 48.5% female) initiated 2 149 385 radiation treatment episodes at 2150 practices. Mean (SD) 90-day standardized radiation treatment-specific spending was $13 683 ($8628). Practice-level per-episode radiation-specific spending variation was high (SD after adjusting for year, $4121). It remained high even after adjusting for patient demographic characteristics, cancer type, geography, radiation technology, and number of fractions (SD, $1487). From 2009 to 2020, unadjusted per-episode standardized radiation-specific spending increased slightly from $12 978 to $13 689 (P = .04). During this time, the median (IQR) number of fractions per episode decreased from 25 (10-33) to 16 (5-29) (P < .001), while the proportion of radiation episodes using intensity-modulated or proton radiation treatment increased (from 5% to 18% and 0.4% to 2%, respectively [both P < .001]) and use of conformal radiation treatment decreased from 61% to 38% (P < .001). Conclusions and Relevance: In this cross-sectional study, there was substantial variation in practice-level radiation spending and number of fractions for older patients with cancer undergoing radiation treatment both within and across health care markets. This practice-level variation suggests that there may be opportunities for savings under population-based payment models.