Proposed Article Structure
1. Highlights
2. Background: Why systemic therapy initiation matters in episode-based oncology payment
3. Study design and methods
4. Key results
5. Clinical and policy interpretation
6. Strengths and limitations
7. Implications for oncology practice and payment reform
8. Funding, registration, and citation
Highlights
The Oncology Care Model (OCM), a Medicare episode payment model triggered by systemic cancer therapy, was not associated with an increase in treatment initiation among patients with newly diagnosed cancer.
Among patients with poor-prognosis cancers, OCM participation was associated with a modest but statistically significant reduction in systemic therapy initiation, suggesting less treatment at the end of life or in settings where benefit may be limited.
Spending decreased significantly in the poor-prognosis cohort and trended downward in the incident cancer cohort, raising the possibility that prior evaluations may have underestimated savings by not explicitly examining treatment initiation.
The study addresses a central concern in health policy: whether a payment model triggered by a billable service can unintentionally create incentives to increase episode volume.
Background
Alternative payment models in oncology are intended to improve quality, coordination, and efficiency in cancer care. Yet they also create a familiar policy tension: if payment depends on the occurrence of a specific service, does the model inadvertently reward more of that service? The OCM, launched by the Centers for Medicare & Medicaid Services (CMS) in July 2016, is one of the most important U.S. oncology payment reforms to date. It used six-month episodes of care that were triggered by initiation of systemic therapy. Participating practices received monthly care management payments and later could face performance-based incentives tied to total episode spending and quality.
The concern addressed by Keating and colleagues is therefore highly relevant. If systemic therapy itself triggers an episode and its associated payments, practices might have a financial incentive to start treatment in marginal cases, especially during the early period when all practices were under one-sided risk and therefore shared in potential gains without exposure to downside losses. In oncology, where treatment decisions near the boundaries of expected benefit can be complex, even small distortions in treatment initiation could have important clinical, ethical, and financial consequences.
This issue is especially salient for patients with poor-prognosis cancers. In such populations, decisions around chemotherapy or other systemic therapy often require weighing limited survival benefit, toxicity, logistical burden, patient goals, and palliative alternatives. If payment reform encourages more treatment in these settings, it may worsen low-value care. Conversely, if it supports more deliberate care planning and symptom management, it could reduce unnecessary treatment and spending while better aligning care with patient preferences.
Study Design and Methods
This investigation was a quasi-experimental study using a matched difference-in-differences design based on serial cross sections of Medicare beneficiaries from January 2010 through December 2019. The analysis compared patients treated at OCM-participating practices with patients treated at matched nonparticipating practices, examining changes before and after OCM implementation in July 2016.
The authors identified beneficiaries with an index visit for cancer and followed them for one year. Two clinically distinct populations were analyzed. The first was an incident cancer population, defined as patients with newly diagnosed cancer. The second was a poor-prognosis cohort, intended to capture patients whose cancers carry limited life expectancy and for whom the appropriateness of systemic therapy may be particularly uncertain or preference-sensitive.
The primary outcome was initiation of systemic therapy in the year after the index visit. This outcome is conceptually important because OCM episodes begin only when systemic treatment is started. A secondary outcome was total Medicare payments in the year after the index visit.
The matched difference-in-differences approach is well suited to policy evaluation when randomization is not feasible. By comparing changes over time between intervention and comparison practices, it aims to isolate the association of OCM from secular trends. As with all such analyses, the validity of the approach depends heavily on the plausibility of parallel preintervention trends and on the adequacy of the matching strategy.
Study Population
The incident cancer analysis included 754,182 patient episodes representing 750,483 patients. Mean age was 74.1 years, and 62.2% were female. The poor-prognosis cohort included 517,858 patients with a mean age of 72.4 years, of whom 52.2% were female. The comparison was conducted across 197 OCM practices and 197 matched non-OCM practices.
The size of the study is a major strength. It offers sufficient power to detect modest changes in treatment initiation and spending, and it reflects real-world care across a broad Medicare population. However, because the analysis is limited to Medicare beneficiaries, generalizability to younger commercially insured patients or Medicaid populations remains uncertain.
Key Results
Primary outcome: systemic therapy initiation in incident cancer
Among patients with incident cancer, OCM participation was not associated with a statistically significant increase in systemic therapy initiation. The differential change was -0.9 percentage points, with a 95% CI from -2.2 to 0.3 percentage points; P = .14.
This is the central negative finding of the study and an important one. It directly counters the concern that an episode model triggered by treatment would lead to more treatment starts overall in newly diagnosed cancer. In other words, during the early years of the OCM, when practices were effectively protected from downside financial risk, there is no evidence from this analysis that they responded by expanding systemic therapy initiation among incident cases.
Primary outcome: systemic therapy initiation in poor-prognosis cancer
In the poor-prognosis cohort, the association went in the opposite direction. OCM participation was associated with a statistically significant decrease in the likelihood of systemic therapy initiation. The reported differential change was 1.5 percentage points, but the confidence interval clearly indicates a reduction: 95% CI, -2.8 to -0.2 percentage points; P = .03.
Clinically, this finding may be more important than the modest absolute effect size suggests. In poor-prognosis settings, small shifts away from treatment initiation may reflect better advance care planning, more selective use of therapy, improved palliative integration, or a greater focus on supportive care when expected benefit is limited. These are plausible pathways through which a care-management-focused oncology model could improve value.
Secondary outcome: Medicare spending
For patients with incident cancer, OCM was associated with a non-statistically significant relative decrease in spending of -$898.26 in the year after the index diagnosis (95% CI, -$1890.31 to $93.80; P = .08). Although this did not reach conventional statistical significance, the direction of effect aligns with the model’s intended goal of lowering spending without increasing unnecessary treatment.
In the poor-prognosis cohort, spending fell significantly by -$2192.15 relative to comparison practices (95% CI, -$3559.66 to -$833.63; P = .002). This is a clinically and policy-relevant result. When coupled with lower systemic therapy initiation, it suggests that at least some of the savings under OCM may have arisen not merely from reducing prices or downstream service use within episodes, but from changing whether episodes were initiated in the first place.
Clinical and Policy Interpretation
The findings speak to a longstanding design challenge in alternative payment models: the difference between per-episode efficiency and episode-generation incentives. A model can succeed in reducing spending conditional on an episode occurring, yet still increase total spending if it encourages more episodes. Keating and colleagues tested exactly that concern and found no evidence of increased episode triggering through greater systemic therapy initiation. Indeed, in poor-prognosis disease, the model was associated with less initiation.
For clinicians, this is reassuring. It suggests that OCM did not broadly push oncologists toward overtreatment simply because systemic therapy started the payment clock. For policymakers, it offers a more nuanced lesson: payment models that combine episode incentives with care management support may not necessarily induce volume expansion, particularly when the targeted population includes patients for whom careful treatment selection is central to high-value care.
The reduction in treatment initiation among poor-prognosis cancers merits careful interpretation. It should not be assumed to represent underuse. In context, it may indicate more appropriate avoidance of low-benefit therapy. The OCM included monthly enhanced oncology services payments intended to support care coordination, access, and navigation. These resources may have enabled better conversations about prognosis, goals of care, and alternatives to systemic treatment. If so, the mechanism would be consistent with contemporary oncology quality priorities, including earlier palliative care integration and reduction of aggressive end-of-life care that does not improve outcomes patients value.
The study also raises an important point about prior OCM evaluations. If analyses focus only on spending among triggered episodes, they may miss savings attributable to fewer treatment initiations. This paper argues that omission could underestimate the full effect of the model. That is a methodologic insight with implications far beyond oncology. Evaluators of episode-based payment reforms should consider both within-episode efficiency and effects on episode incidence.
Expert Commentary
From a health services research perspective, this study is notable because it evaluates a behavioral response that many payment reforms are theoretically vulnerable to but that is often difficult to measure. In oncology, treatment initiation is not a purely administrative event; it is also a deeply clinical decision. Demonstrating no overall increase in starts therefore helps address concerns that financial design overwhelmed clinical judgment.
The results are also broadly consistent with the direction of evidence supporting less aggressive care near the end of life when palliative approaches are emphasized. American Society of Clinical Oncology guidance has long encouraged early palliative care integration for advanced cancer, particularly where prognosis is limited and symptom burden is high. A payment model that supports coordination and patient-centered decision-making could reasonably reduce treatment use in poor-prognosis settings without compromising quality.
Still, interpretation should remain cautious. Reduced systemic therapy initiation is not inherently beneficial unless it reflects appropriate selection and preserved or improved patient-centered outcomes. The present analysis did not report symptom control, quality of life, hospice timing, patient preferences, or survival consequences. Those outcomes are critical for determining whether less treatment represented better care, delayed care, or foregone beneficial treatment in some subgroups.
Strengths and Limitations
Strengths
The study has several major strengths. It uses a large national Medicare dataset spanning nearly a decade, includes substantial numbers of practices and patients, and applies a matched difference-in-differences design that is appropriate for evaluating a large-scale policy intervention. The dual focus on treatment initiation and total spending is especially valuable because it addresses both utilization behavior and financial consequences.
Limitations
As with any nonrandomized policy analysis, residual confounding remains possible. Practice participation in OCM was voluntary, so unmeasured differences between participating and comparison practices may have influenced observed trends despite matching. The paper summary provided does not detail subgroup heterogeneity by cancer type, line of therapy, biomarkers, or oral versus infused systemic treatment, all of which could shape treatment initiation patterns.
The outcome measure captures whether systemic therapy was initiated, but not whether treatment was evidence-based, concordant with patient goals, or of high clinical value. Nor does the abstract provide cancer-specific mortality or quality-of-life data. In addition, Medicare claims are not ideal for granular clinical variables such as performance status, frailty, symptom burden, or disease burden, which are highly relevant when judging treatment appropriateness in poor-prognosis cancers.
Finally, the study period includes important secular changes in oncology, including rapid growth in immunotherapy, targeted therapy, and evolving supportive care practices. Although the difference-in-differences framework is designed to account for secular trends, treatment innovation can affect specialties unevenly and may complicate attribution.
Implications for Oncology Practice and Payment Reform
For practicing oncologists, the study suggests that value-based payment models need not necessarily distort front-end treatment decisions. The OCM appears not to have increased initiation of systemic therapy overall and may have reduced potentially low-value starts in poor-prognosis settings. That is a favorable signal for future models that aim to reward coordination and appropriateness rather than volume.
For health policy experts, the findings support more comprehensive evaluation frameworks. Episode-based models should be assessed at two levels: what happens within episodes, and whether the number of episodes changes. Ignoring the latter can miss clinically meaningful behavioral responses and produce incomplete estimates of savings or harms.
For CMS and successors to OCM, such as newer oncology payment reforms, the data suggest that supportive care infrastructure and care management payments may influence decision quality at treatment initiation. Future models should preserve safeguards against undertreatment while encouraging shared decision-making, advance care planning, and palliative care access. Incorporating patient-reported outcomes, hospice use, and disease-specific quality measures would strengthen interpretation of changes in therapy initiation.
Conclusion
In this large Medicare analysis, the Oncology Care Model was not associated with increased systemic therapy initiation among patients with incident cancer, easing concerns that a treatment-triggered episode payment model would encourage more treatment starts. Among patients with poor-prognosis cancers, OCM was associated with lower systemic therapy initiation and lower Medicare spending, findings that may reflect more selective and potentially more appropriate use of treatment.
The study’s broader message is methodological as well as clinical: evaluations of episode payment models should examine not only spending per episode but also whether the model changes the likelihood that episodes occur at all. In oncology, where treatment initiation can be both high-cost and highly preference-sensitive, that distinction is essential.
Funding, Registration, and Citation
Funding information was not provided in the abstract supplied here. This was a quasi-experimental observational study and not a clinical trial; no ClinicalTrials.gov registration number was reported in the abstract.
Citation: Keating NL, Lam MB, Landrum MB, McWilliams JM, Wright AA, Brooks GA, Zubizarreta JR, Buzzee B, Landon BE. The Oncology Care Model and Initiation of Systemic Therapy for Cancer. JAMA Internal Medicine. Published online April 27, 2026. PMID: 42043828. URL: https://pubmed.ncbi.nlm.nih.gov/42043828/
Selected References
Keating NL, Lam MB, Landrum MB, McWilliams JM, Wright AA, Brooks GA, Zubizarreta JR, Buzzee B, Landon BE. The Oncology Care Model and Initiation of Systemic Therapy for Cancer. JAMA Internal Medicine. 2026. PMID: 42043828.
Centers for Medicare & Medicaid Services. Oncology Care Model. CMS Innovation Center program materials and public methodology documents.
Smith TJ, Temin S, Alesi ER, et al. American Society of Clinical Oncology provisional clinical opinion: the integration of palliative care into standard oncology care. Journal of Clinical Oncology. 2012;30(8):880-887. PMID: 22312101.
Schnipper LE, Davidson NE, Wollins DS, et al. American Society of Clinical Oncology statement: a conceptual framework to assess the value of cancer treatment options. Journal of Clinical Oncology. 2015;33(23):2563-2577. PMID: 26101248.
