Risk Stratification Of Bundled Payment Models Requires Clinical Data

March 16, 2021 - Incorporating clinical data into risk stratification could improve the accuracy of CMS bundled payment models like the Oncology Care Model (OCM), according to a new Avalere analysis.

The OCM is a bundled payment model that evaluates the total cost of care for episodes initiated by chemotherapy treatments within six-month performance periods (PP). At the end of each period, CMS measures practice performance by comparing total expenditures to a benchmark price predicted by factors like tumor type, comorbidities, member demographics, and utilization of certain services.

Participating practices that stay within the benchmark price earn a performance-based payment (PBP) and those that exceed the benchmark owe CMS recoupment.

Starting in PP7, which began in July 2019, CMS implemented a “metastatic adjuster” for breast, small intestine/colorectal, and lung cancers, which factors advanced disease into the benchmark price calculation.

Avalere researchers set out to assess whether adjusting for advanced disease will improve accuracy of benchmark prices. First, they grouped earlier-stage (stage I and II) and later-stage (stages III and IV) episodes from PP3 and PP4 using claims-based indicators. Then, the researchers conducted the OCM calculation to see if the metastatic adjuster accurately predicted episode expenditures for early and late-stage cancer.

While later-stage episodes had slightly higher costs compared to earlier-stage episodes, benchmark prices for earlier-stage episodes were significantly higher than later-stage episodes.

The researchers found that for later-stage episodes, expenditures exceeded benchmark prices by 10 percent, with a benchmark price of $36,930 and an episode expenditure of $40,670. On the other hand, average expenditures for earlier stage episodes ($39,290) were 17 percent lower than benchmark prices ($47,520).

The researchers noted that this disparity in benchmark prices is partially due to the fact that the prediction model adjusts for services commonly administered in the early stages of cancer treatment, such as chemotherapy, radiation, and surgery. However, the calculation does not take into account costs for late-stage care.

What’s more, treatment costs in earlier stage episodes make up a smaller ratio of overall expenditures compared to treatment costs in later stage episodes. Therefore, the metastatic adjuster may increase predictions for earlier-stage episodes and decrease predictions for later-stage episodes.

This finding reveals that during the first six PPs, a practice’s ability to earn PBPs may have been hindered if they had high rates of late-stage breast cancer episodes.

"Starting in PP7, the incorporation of a metastatic adjuster will mark the first time clinical registry data is factored into setting benchmark prices,” the report authors wrote. “Performance in metastatic episodes is likely to improve, and the adjuster could benefit practices with a higher proportion of later-stage breast, lung, and small intestine/colorectal episodes (the metastatic adjuster only applies to these 3 cancer types).”

As payers and providers strive to reduce patient out-of-pocket costs for fee-for-service (FFS) Medicare beneficiaries, it is important to note that bundled payment models in oncology may be beneficial for some practices and disadvantageous for others.

In order for bundled payment models to accurately measure practice performance, policy makers should leverage more clinical data that details how expenditures differ depending on level of advanced disease.

For instance, this analysis shows that if the metastatic risk calculation had better accounted for differing treatment costs throughout breast cancer progression, benchmark prices would have been more accurate, leading to more appropriate reimbursement.

In other words, while the goal of the metastatic risk calculation was to promote care coordination and proper reimbursement based on patient’s cancer stage, lack of clinical data use in the creation of the adjuster may sabotage its original purpose.

This finding will be especially important as the OCM evolves into its next iteration, the Oncology Care First model.