At a glance
A pair of new studies indicates that Medicare’s Oncology Care Model, a major test of value‑based payment in cancer care, was more successful than early evaluations suggested.
The new analyses show that participating practices achieved growing savings over time, largely through smarter use of supportive drugs, while measured quality of care remained stable.
OCM practices were less likely to begin chemotherapy in patients with poor‑prognosis cancers — in whom benefits of chemotherapy are often uncertain — suggesting more selective use of treatment, which resulted in lower spending.
The future of federally funded research at Harvard Medical School — supported by taxpayers and done in service to humanity — remains uncertain. Learn more.
In the United States, most health care is paid for one service at a time, where every visit, scan, and treatment generates a separate bill. While straightforward, this approach tends to reward the volume of care delivered, not the overall outcomes that patients experience.
Over the past decade, policymakers have been testing a different idea: value‑based care.
Instead of paying only for individual services, value‑based models tie payment to the total cost, quality, and outcomes of care for patients over a defined period. The goal is to align payment more closely with better care and better results for patients, not simply more care.
In 2016, Medicare introduced the Oncology Care Model (OCM) to apply value‑based payment to cancer care. Launched by the Centers for Medicare & Medicaid Services Innovation Center, OCM asked participating oncology practices to rethink how they delivered care to patients receiving systemic therapy. In return, Medicare offered monthly care‑coordination payments and the chance to share in savings if practices could hold down overall spending while maintaining quality.
The model ran for six years and ended in 2022, with a prevailing judgment that it had not saved enough to justify its cost.
Two new studies published in JAMA and JAMA Internal Medicine suggest that earlier evaluations may have underestimated the model’s impact.
Harvard Medicine News spoke with Nancy Keating, professor of health care policy in the Blavatnik Institute at Harvard Medical School and HMS professor of medicine at Brigham and Women’s Hospital, who collaborated with large multi-institutional teams of colleagues from HMS, Dartmouth Cancer Center, Abt Global, and others on this research, about what new data from the OCM experiment reveal about both the promise and the limits of value-based care in oncology.
Harvard Medicine News: How did the Oncology Care Model work for practices and patients?
Nancy Keating: OCM was a payment and delivery experiment to try to control spending while preserving or improving the quality of oncology care. Participation was voluntary, with just over 200 practices signing up. When a patient at one of those practices initiated chemotherapy — traditional chemotherapy, immunotherapy, targeted therapy, or hormonal therapy — that triggered a six‑month “episode” of care.
For those six months, the practice could bill an additional $160 per patient per month. That money was to be used to improve and redesign care delivery.
Medicare then looked at all the spending on that patient’s care during the episode and compared it to what similar care had cost for that practice in the past, adjusted for inflation and changes in patient mix. If, over a six-month performance period, the practice met certain quality benchmarks and had savings, they could share in those savings.
Early on, all practices were in one‑sided risk, where there was upside if the practice had savings, but if spending was over the target, nothing happened. Later, some practices moved into two‑sided risk, where they could potentially earn more but also had to pay money back if spending was higher than targets.
HMNews: The first of your two new papers, in JAMA, evaluates OCM overall. What were the key findings?
Keating: Both studies were very much a team effort. For the JAMA study, we were part of an evaluation team contracted by the Centers for Medicare & Medicaid Services (CMS) to study the model. We looked at spending, utilization, and quality over the six years the program ran, comparing OCM practices to other oncology practices that were similar in size, patient mix, and other characteristics but did not participate in OCM.
We found a modest improvement in spending overall: Total payments decreased by an average of $616 per six-month episode over the study period.
To me, the most important finding is that savings increased substantially over time, suggesting it takes some time for practices to learn how to deliver care in this new world. By the later years of the model, practices were saving well over $1,000 for a six‑month episode — enough to offset most of what Medicare was paying them in monthly care-coordination and performance-based payments. This means that practices were figuring out how to be more efficient.
Most of the savings came from prescribing lower‑cost supportive‑care drugs and adopting biosimilars faster, not from changing which chemotherapy drugs they used. Practice leaders told us during site visits that they preferred not to tell doctors which chemotherapy agents to prescribe.
We also looked at several measures of quality — patient experiences, emergency department visits, hospice use, chemotherapy‑associated ED visits and hospitalizations, timeliness of treatment after surgery for curative cancers, supportive‑care quality, and end‑of‑life care. Basically, everything was flat. Nothing was better, but nothing was worse. Given the big shift in incentives, it’s at least a relief that quality didn’t get worse.
One important point is that, averaged over the entire six‑year model, the $616 in savings per episode did not make up for all the $160‑per‑month care‑coordination payments and performance‑based bonuses. Overall, CMS lost money on OCM, even though by the end of the model it was essentially breaking even.
HMNews: The second paper, in JAMA Internal Medicine, which was not initiated or funded by CMS, looked at a different concern: that paying per episode might push doctors to start more treatment. What did you find?
Keating: That concern was especially relevant in a model with one‑sided risk and generous monthly payments. You can imagine it could be easy to keep treating patients with chemotherapy, which brings in extra practice revenue, even if you’re not sure it will benefit the patient.
We asked whether there was a change in the propensity to start chemotherapy. We identified patients who looked to be newly diagnosed with cancer or with newly recurrent cancers, and we had two groups: all comers and those with poor‑prognosis cancers. The poor‑prognosis cancers are more aggressive, where chemotherapy is unlikely to cure and the ratio of benefits to harms is more uncertain.
We found that, particularly in the poor‑prognosis group, there was a drop in the likelihood of initiating chemotherapy in OCM relative to comparison practices. That decrease was associated with a substantial decrease in spending — about $2,000 less per patient in the year that followed the initial diagnosis. Our hypothesis going in was that we would see more chemotherapy for these patients. Instead, we saw less, which we think reflects more thoughtful decisions about who is most likely to benefit.
Here’s how that connects back to the money. In our first paper, we looked only at people who started chemotherapy and entered an OCM episode, and we found that spending per six‑month episode was about $616 lower on average. That reduction did not fully cover all the extra payments Medicare was making to practices — the $160‑per‑month care‑coordination payments plus performance bonuses — so overall CMS lost money on OCM, even though in the later years savings per episode were much larger.
What this second paper shows is that practices weren’t just lowering spending once treatment started; for some patients, especially those with poor‑prognosis cancers, they were deciding not to start chemotherapy at all. Because chemotherapy is so expensive, those avoided starts likely generated additional savings that were not counted in the OCM evaluation. So OCM probably saved more money than CMS realized and may even have been cost saving if chemotherapy initiation had been considered.
HMNews: Do these findings change how you think about whether OCM “worked”? What do they say about value-based care?
Keating: The perception in the oncology community is that OCM didn’t work. Our evaluation concluded that although the model didn’t produce net savings for CMS and therefore could not be implemented broadly as a permanent payment model, by the end of the six years, practices were breaking even. Our new paper suggests that had CMS taken a broader view — including the decision about whether to start chemotherapy in poor-prognosis patients — they might have found net savings.
I think OCM was not a failure, and our research suggests that value-based payment in oncology has potential. At the same time, value‑based care is challenging to implement. The follow‑up model to OCM, the Enhancing Oncology Model, was necessarily designed several years ago, before we fully understood OCM’s impacts. It focuses on a smaller subset of cancer types and requires practices to accept two‑sided risk. This new model has had very low participation, with only 28 practices currently participating, which will limit what we can learn from it. Oncologists seem reluctant to be in two‑sided risk models. Understandably, they worry that a bad year could threaten their practice’s long-term viability.
Overall, I’m encouraged and think that there is still hope for episode‑payment models in oncology. But we’re still trying to figure out how to best implement value‑based care models in the United States. It’s really complicated, and we need to keep working to get it right.
This interview was edited for length and clarity.
Authorship, funding, disclosures
Keating is senior author on the first study and first author on the second. Additional authors on the JAMA study include Gabriel Brooks, Matthew Trombley, Sean McClellan, Qing Zheng, Mary Beth Landrum, Van Doren Hsu, Colleen Kummet, Pang-Hsiang Liu, Derek Hoodin, Colin Doyle, and Carol Simon. Additional authors on the JAMA Internal Medicine study include Miranda Lam, J. Michael McWilliams, Alexi Wright, Jose Zubizarreta, Benjamin Buzzee, and Bruce Landon, as well as Landrum and Brooks.
The authors of the first study were members of an evaluation team contracted by the Centers for Medicare & Medicaid Services (CMS) to evaluate the Oncology Care Model. The analyses on which the JAMA publication is based were performed under contract HHSM-500-2014-000261, sponsored by CMS. The JAMA Internal Medicine publication was supported by a grant from the U.S. Agency for Healthcare Research and Quality (R01HS026498), with no funding or involvement from CMS.
A full list of disclosures can be found in the studies.
