The program reduced per-episode spending as intended, the researchers found. But those savings were offset by bonuses paid to hospitals for beating financial benchmarks, leading to a net loss.
One of CMS’s signature value-based care programs wound up costing rather saving CMS money, according to results of a study published this week in JAMA.
Although the results can be interpreted in different ways, the study may put a dent in the reputation of value-based care and its promise as a way to rein in U.S. healthcare spending while also keeping the quality of care high.
The study by Sukruth A. Shashikumar of the Washington University School of Medicine in St. Louis and his colleagues showed that Bundled Payments for Care Improvement Advanced (BPCI-A) was associated with a $279 million increase in Medicare spending.
The program did achieve reduce per-episode spending as intended to the tune of $175 per episode, which tallied up to $75 million across nearly 430,000 episodes of care during the first two years of the program, according to the researchers’ calculations. But that savings in Medicare spending on clinical care was more than offset by the $762 million in the bonuses that CMS paid out to the hospitals participating in the program for beating the financial targets for those episodes.
CMS also assessed $408 million in financial penalties on hospitals that exceeded financial benchmarks, according to the figures reported in the JAMA. The bonuses exceeded the penalties by approximately $354 million, so the net financial result of the program was a cost of $279 million (the net payments of $354 million minus the $75 million in savings).
Shashikumar and his colleagues highlighted another finding: BPCI-A payments were skewed in favor of hospitals with a greater proportion of patients in poverty and from racial and minority groups, and those receiving Disproportionate Share Hospital payments.
An accompanying editorial by Joshua Liao, M.D., M.Sc., of University of Washington School of Medicine in Seattle and Amol Navathe, M.D., Ph.D., of the University of Pennsylvania in Philadelphia, said that “acknowledging net losses may reflect investments in more equitable participation” in the bundled payment program. “Ultimately,” wrote Liao and Navathe, “the priority of reducing spending through payment models must be balanced by the priority of combating structural inequities.”
Liao and Navathe also argue against emphasizing net losses when the BPCI-A was associated with changes in healthcare practices that reduced spending on healthcare. Benchmarks for episodes of care are difficult to set, they said, because the costs of episodes of care are volatile and influenced by factors such as the price of infused drugs.
Shashikumar and his colleagues say their results show that the target prices used by CMS are “not true counterfactuals.” They also mention the difficulty of prospectively determining target prices and recommending using Bayesian statistical methods to set them.
Another explanation for the net losses from the researchers is that the hospitals that saw higher price targets (which they believed they could beat) were more likely to participate in the program.
The findings in JAMA were based on calculations of payments made to 694 hospitals who volunteered to participate BPCI-A. The first two years of the program covered the period from Oct. 1, 2018, through Dec. 31, 2019.
BPCI-A sets up payments for 29 different kinds of episodes, ranging from myocardial infarction (heart attacks) to back and neck spinal fusions to urinary tract infections. According to Shashikumar and his colleagues, the most selected episode among the participating hospitals was congestive health failure, followed by sepsis and simple pneumonia and respiratory tract infection. The intent is to encourage cost-effective care by setting a target price for these episodes of care and then paying hospitals bonuses if they spend less than the target. Other value-based programs are based on patients rather than episodes of care.
The researchers calculated BPCI-A’s effect on per-episode spending by comparing spending differences between 2013 and 2019.
After the first two model years, the BPCI-A program was put on hold by CMS in 2020 because of the COVID-19 pandemic. The target prices were modified for the fourth year of the program, which covers 2021, but results from that year are not available yet for study, according to Shashikumar and his co-authors.