National benchmarks and practice level variation on eight cross cutting quality measures.

255 Background: While there is increased attention on the importance of quality measurement in oncology, especially with the rise of value-based payment, limited data exist on national averages and practice level variation for proposed quality measures to establish benchmarks and targets for quality improvement initiatives or value-based contracts. Methods: UnitedHealthcare (UHC) developed peer comparison reports for eight cross cutting quality measures for practices with an active contract for at least one of its commercial, Medicare or Medicaid health plans and ≥1 provider from the following specialties: gynecologic oncology, hematology/oncology, pediatric hematology/oncology, radiation oncology, or surgical oncology. Adherence to the quality measures below was calculated using a mix of claims data, clinical data from a prior authorization for cancer therapy, and CMS MIPS data. Patients were attributed through an algorithm that selected the most probable physician responsible for the patient’s care - responsible prior authorization provider, servicing provider or most recent visited provider prior to the treatment, varying by each measure. Dates of service differ by measure, ranging from 1/1/2019 through 12/31/2020. Results: We identified 5,828 unique tax identification numbers (TINs) with UHC members with cancer attributed to them during 2019-20. The number of practices included in the measurement cohort per measure varied significantly from 301 to 4,120 (tobacco screening and performance status, respectively). 2,422 TINs met the minimum patient count for at least one measure (≥10 patients or events). Overall performance ranged from 13.5% to 77.3% (hospice admission and PS documented) for measures where higher adherence reflects better quality of care. For measures where lower scores represent higher quality of care the range was 11.4% to 22.6% (hospice < 3 days and ED admission, respectively). Observed adherence was statistically better than expected for 0.5%-5.8% and statistically less than expected for 0.9%-5.6% of TINs in UHC’s network; however, more than half of the practices had insufficient sample size to make a determination. Conclusions: We observed substantial variation in quality across a national cohort of oncology practices. However, even for a large national payer, small sample sizes limited the assessment of a substantial number of practices.[Table: see text]