Providers participating in Medicare's voluntary bundled payment models often drop out when forced to assume downside risk, according to a new Government Accountability Office report.
Groups did a risk-benefit analysis when deciding to participate in voluntary episode-based payment models in the first place, but when they entered the phase where they face financial penalties for failing to cut costs while providing high-quality care, they were more likely to leave the model altogether, GAO found. Just 39% of hospitals participating in the second iteration of the Bundled Payments for Care Improvement model remained in after crossing that line.
GAO compiled the report from CMS data and stakeholder interviews in response to a request from Sen. Ron Wyden, D-Ore., ranking member of the Senate Finance Committee, in December.
A major focus of the Trump administration's healthcare agenda has been cost-cutting, whether it be pharmaceutical spending, reducing user fees in the ACA exchanges or, in this case, alternatives to traditional Medicare payments.
HHS Secretary Alex Azar announced in November that the agency was taking a more serious look at mandatory payment models, especially for radiation oncology and cardiac care, though it hasn't issued anything concrete yet.
This was an about-face for an administration that in 2017 reduced the scope of the Comprehensive Care for Joint Replacement Model (CJR) and halted the launch of other mandatory bundled payment models before they could begin.
But CMS has either tested or is testing six bundled payment models: four versions of Bundled Payments for Care Improvement, the Oncology Care and the CJR models.
Rather than paying providers based on the amount or complexity of service, CMS establishes a target payment amount that encapsulates the total cost of the health "event," such as a surgery or hospitalization, for a Medicare beneficiary.
Providers that can supply that high-quality service without going over that target get to keep the extra dollars left over.
But provider participation in all but one of the six bundled payment models CMS is currently testing is completely voluntary, with 2016's ongoing CJR model the only one where 67 hospitals were forced to participate (reduced to 34 in 2017 and none forced in 2018).
In models with upside risk, providers are rewarded financially for keeping spending below the target. In models with downside risk, providers are penalized with reduced payments or other punishments for exceeding the target or not meeting care quality benchmarks.
Stakeholders interviewed by the GAO say the mandatory nature allows the agency to reap data from a more diverse panel of participants for evaluation. Another potential benefit is that participants can't exit if they get antsy, which is relatively common in voluntary models.
In 2018, when the CJR model became voluntary, all of the eligible safety-net hospitals left the program — save two.
Voluntary models allow CMS to test novel concepts among early adopters without a lot of pressure, GAO found. But such models largely favor participants more as CMS is incentivized to make the models attractive to entice participation, and they can bounce whenever they stop performing well.
GAO also pointed out that CMS must risk-adjust the target prices providers receive to account for differences in health among patient populations. This could cause participating hospitals to eschew sicker patients for healthier ones, a common fear among critics of mandatory APMs.
But two JAMA studies in 2018 cast doubt on that theory and showed bundled payment models resulted in better quality of patient care and lowered unsustainable costs.
GAO also found provider groups participating in the six bundled payment models had larger practices, more beds and higher episode volume on average, and were more likely to be located in urban areas.