Physician Practice Variability in the Use of Extended-Fraction Radiation Therapy for Bone Metastases: Are We Choosing Wisely?

PURPOSE: Routine use of extended-fraction (> 10 fractions) radiation therapy (RT) for palliation of bone metastases is recognized as a low-value intervention by the American Society for Radiation Oncology. We examined contemporary practice patterns of, and physician characteristics associated with extended-fraction RT use.MATERIALS AND METHODS: We conducted a retrospective cohort study using Medicare fee-for-service data. We included patients who underwent 2- or 3-dimensional external-beam RT for bone metastases between January 1, 2016, and December 31, 2018. Physicians treating > 10 patients over the study period were analyzed for their individual practice. Hierarchic logistic regression modeling was used to identify patient- and physician-level factors associated with extended-fraction RT use.RESULTS: A total of 12,221 patients (median age, 75.6 years; 40.9% women, 87.6% white) were included. The rate of extended-fraction RT was 23.4%. A total of 1,432 physicians treated any patient. Among the 382 physicians treating > 10 patients, 127 (33.2%) used extended-fraction RT > 30% (consensus threshold). Physician factors associated with decreased odds of extended-fraction RT were years since medical school graduation (≤ 10 years and 11-20 years v ≥ 31 years: adjusted odds ratio [aOR], 0.32 [95% CI, 0.20 to 0.51] and 0.64 [95% CI, 0.44 to 0.93]) and practicing in the Northeast or Midwest versus the South (aOR, 0.36 [95% CI, 0.22 to 0.58] and 0.48 [95% CI, 0.31 to 0.74]). Physicians treating > 20 patients ( v 11-14 patients) over the study period had increased odds of delivering extended-fraction RT (aOR, 1.53 [95% CI, 1.10 to 2.12]).CONCLUSION: In this study, almost one fourth of patients received extended-fraction RT, and one third of physicians had an extended-fraction RT use rate of > 30%. Personalized feedback of performance data, clinical pathways and peer review, and updated reimbursement models are potential mechanisms to address this low-value care.