Next-Generation Alternative Payment Models in Oncology—Will Precision Preclude Participation?

On June 30, 2022, the Oncology Care Model (OCM), the first cancer-specific alternative payment model (APM) from the US Center for Medicare and Medicaid Services (CMS), ended. The OCM was Medicare’s first chronic disease–specific APM. While voluntary, many practices participated; at one point, a quarter of all oncology patients in the US were served by an OCM-participating practice. However, the OCM was associated with a $315.6 million net loss to Medicare without meaningful improvements in quality. Reasons for this loss include overspending, such as potentially excessive monthly payments for care coordination for low-risk cancer episodes, and little risk-sharing by practices, with very few practices electing to incur downside risk for exceeding spending targets.¹

Learning from the modest OCM results, CMS recently announced its second voluntary oncology APM, the Enhancing Oncology Model (EOM), which will launch in July 2023.² The EOM includes several changes to the OCM, including methods that increases the predictability and accuracy of episode spending benchmarks (ie, how much a 6-month course of cancer treatment should cost). This is a welcome development. However, lower upfront payments, mandatory downside risk, and increased reporting requirements may disincentivize practices from participating in the EOM, and this selection bias could complicate interpretation of the eventual results. To engage most oncology practices in value-based reform, voluntary oncology APMs must couple careful attention to episode design with appropriate upfront incentives for participation.

Improving on the OCM

Although final details are pending, the EOM on the surface makes several improvements on the OCM. These are associated with improved granularity in bundle creation and price forecasting.

Methodological Improvements Associated With Improved Accuracy of the Expected Costs of Care for Episodes

The OCM set spending targets for discrete 6-month episodes of cancer treatment. One major criticism of the OCM was the heterogeneity of spending within episodes. For example, within a metastatic breast cancer episode, there could be a 100-fold price difference among approved cancer treatments based on receptor status and line of therapy.¹ The EOM includes a much smaller set of high-volume cancers, which should limit the effect of heterogeneous case-mix on smaller practices. The incorporation of clinical features, such as metastatic status and ERBB2 status, for breast cancer allows more granularity and precise estimation of cancer episode costs. Finally, the EOM incorporates anovel therapy adjustment when setting reimbursement that is calculated separately for each cancer, which accounts for differential trends in novel therapy availability among different cancers.

Narrowing of Systemic Treatments That Can Trigger an Episode

In the OCM, practices found it challenging to decrease spending for indolent cancers, such as hormone receptor–positive breast and castrate-sensitive prostate cancers, that have relatively low costs and complications. The exclusion of cancers treated only with endocrine therapies focuses clinician attention on cancers with higher rates of complications, such as cancers treated with intravenous chemotherapy.

Required Reporting of Social Determinants of Health and Electronic Patient-Reported Outcomes (ePROs)

The recognition that social factors beyond a patient’s cancer type have a significant association with health outcomes is an important part of EOM. Mandatory collection of patient-level sociodemographic and ePRO data during EOM may allow for a greater understanding and, if targeted with evidence-based interventions, management of socioeconomic and symptom burdens of cancer survivors.

Challenges With Selection Bias

Despite positive efforts to improve the specificity of spending targets, several elements of the EOM might dissuade participation. First, like OCM, the EOM is a voluntary model. Larger practices with existing care management infrastructure and prior exposure to APMs have an advantage.³ At the same time, practices that participated in the OCM and may have already started to reduce wasteful spending may be penalized because their benchmark is partially dependent on their own historical spend. Therefore, it is not clear whether on balance practices with experience in APMs will choose to participate again in EOM.

Second, compared with the OCM, the EOM reduces Monthly Enhanced Oncology Services (MEOS) payments by nearly half. While the EOM proffers an additional MEOS payment for dual-eligible patients, this allocation may withhold funds from practices situated in states that have not expanded Medicaid. Additionally, given that practices have a smaller list of included cancers and thus fewer eligible patients to work with, the reduction in MEOS payments sharply curtails the total additional payments to practices. The CMS rationalizes this decision by assuming that MEOS payments were associated with the net loss of the OCM. However, these MEOS payments allowed smaller practices to invest in expanded services, such as care management, palliative care, and after-hours infrastructure, that were key to the OCM’s relative success. Decreased ability to support such services may paradoxically decrease savings for participating EOM practices. One must spend money to earn money.

Third, the EOM not only introduces mandatory downside risk to participating practices, but it also involves a narrower financial safe zone, which is the amount of spending that leads to neither a spending reward or penalty. Whereas the OCM recouped monies from only practices that spent more than the benchmark, the EOM goes further and recoups monies from any practice that does not achieve, at minimum, a 2% reduction in spending less than the benchmark. This creates increased risk, particularly for first-time participants or smaller practices that lack existing practice management resources. Finally, reporting requirements to collect individual-level ePROs and social determinants of health are well-intentioned but may involve considerable upfront investment that community practices (which are still recovering from the financial shock induced by the COVID-19 pandemic) cannot bear.⁴

The result of all these factors? The EOM is unlikely to garner participation from smaller practices that lack sufficient infrastructure, practices that did not perform well in OCM due to high costs of care, and practices without experience in value-based care programs to determine their expected risk. The later years of the OCM foreshadow this: even with the higher MEOS payments and decreased reporting requirements in the OCM, one-third of practices withdrew from the program between 2016 and 2022.⁵ Without incentives for these groups to participate, the EOM’s selection bias may make its findings difficult to interpret.

Possible Solutions to Ameliorate Selection Bias and Broaden Effects in Oncology APMs

To reduce selection bias, we propose 3 changes to EOM that potentially balance more targeted episode construction with the appropriate incentives to participate. First, the EOM should restore MEOS payments back to original OCM levels. Alternatively, first-time APM participants, solo practices, and rural practices should receive higher one-time payments to induce participation while they build the data, reporting, and care management infrastructure necessary to succeed.

Second, rather than immediately launching with mandatory 2-sided risk, the EOM could consider establishing a 1-year to 2-year period with gradually increasing levels of risk sharing. This glide path would potentially allow time for unexpected challenges in the model (including recovery from COVID-19–related volume declines) to correct, as well as allow practices time to receive feedback and understand ways to improve.

Third, to satisfy ePRO and socioeconomic data reporting requirements, the EOM should specify evidence-based, narrow, 7 to 14–item assessments rather than forcing practices to develop their own tools. This could ensure that practices that do not have the infrastructure in place for data collection and abstraction can participate and collect longitudinal data.

Conclusions

Relatively straightforward changes to the EOM before implementation in 2023 may drive participation and align practice and Medicare incentives. This potentially represents the best strategy for more patients with cancer to benefit from value-based care.