Minimum episode volume to reduce stochastic risk in the Enhancing Oncology Model.

89 Background: The Enhancing Oncology Model (EOM) is a Medicare alternative payment model assessing care quality & cost efficiency of participating physician group practices’ (PGP) total cost of care (TCOC) spend relative to a cancer-specific benchmark price for 7 cancers. However, low episode counts introduce/enhance case mix variability, resulting in financial risk and unstable performance assessments. This study examines the implications of case volume in the EOM. Methods: A retrospective analysis was conducted using EOM national de-identified baseline episode data. Predicted expenditures were estimated via the EOM price prediction model, with savings calculated as the ratio of the difference between predicted and actual expenditures to the predicted value. To assess variability, a bootstrap analysis with 1,000 resamples was performed across episode counts, computing the mean savings for each bootstrap iteration. The results were classified under Risk Arrangement 2 (RA2) of EOM, and the probability of mean savings falling into RA2’s risk corridors—including performance-based payment (PBP), recoupment (PBR), neutral zone (NZ), stop gain (SG), or stop loss (SL)—was determined for each episode count. Results: We note, from the bootstrap results, substantially higher savings variability at low episode counts, increasing exposure to stochastic financial extremes. The minimum episode counts at which the probability of extreme outcomes reaches 15%, 10%, 5%, 1%, and 0% are 94, 132, 199, 385, and 499, respectively. As episode counts rise, SG and PBP probabilities decline. Analysis of PGPs per baseline time period (n = 16,348) from the baseline episode data revealed two-thirds had < 72 episodes (mean: 93.8, median: 36), suggesting most PGPs face a heightened 1-in-6 chance of extreme financial outcomes due to low episode counts. PGPs with < 72 episodes had a median savings of $1,552 per episode, while those with > 360 episodes had a median loss of $498 per episode. Conclusions: PGPs with fewer episodes face greater financial risk and extreme outcomes, though performed disproportionately well in the baseline data. Further exploration is needed to assess the impact of cancer type mix. Establishing minimum thresholds or refining risk-adjustment techniques to mitigate stochastic risk could encourage PGP participation in EOM-like models. Future research should explore dynamic adjustments to TCOC models that account for provider case volume variations while preserving value-based care principles. Probability of outcomes in RA2 risk corridors Episode Count SG PBP NZ PBR SL 5 26.9% 18.7% 4.4% 10.3% 39.7% 36 4.9% 33.4% 12.3% 23.7% 25.7% 72 0.6% 34.2% 19.1% 29.6% 16.5% 94 0.5% 27.2% 21.6% 36.3% 14.4% 132 0.1% 27.9% 26.4% 36.3% 9.3% 199 0% 23.8% 28.5% 43.0% 4.7% 385 0% 16.5% 35.0% 47.6% 0.9% 499 0% 11.2% 41.6% 47.2% 0%