CMS Releases Initial Report on Risk-Based Bundled Payment Model

Hospitals and physician group practices (PGPs) participating in the first two cohorts of the Bundled Payments for Care Improvement (BPCI) Advanced have a good chance of reducing episode payments. Although the reason may not have much to do with altering practice patterns, according to the model’s first annual evaluation report.

“Because the hospitals chose clinical episodes for which they had higher episode payments, they stand a better chance of reducing episode payments than they would have with lower episode payments,” the report stated. “Similarly, PGPs may make strategic choices in the TINs to use for billing purposes to boost their chances of achieving NPRA [Net Payment Reconciliation Amounts].”

“All of these factors will make it challenging to generalize the results of BPCI Advanced to a larger or different group of participants,” authors of the report stressed.

BPCI Advanced is a new iteration of Medicare’s bundled payment model, Bundled Payments for Care Improvement. This version of the model was designed to encourage participation from providers with a range of historical episode costs and increase the likelihood that the Medicare program will achieve savings. It also qualifies as an Advanced Alternative Payment Model (APM) under the Quality Payment Program because of the level of financial risk associated with the episodes.

The risk-based bundled payment model is achieving its first goal, but it remains to be seen whether participants will achieve Medicare savings, according to the early assessment.

Data from the first six months of BPCI Advanced showed that 334 conveners and non-conveners participants as of March 1, 2010, which marked the end of the one-time retroactive withdrawal period. The participants represented 715 hospital episode initiators and 580 PGP episode initiators.

Approximately 22 percent of eligible hospitals participated in BPCI Advanced, the report found. Hospital episode initiators were larger and more likely to be in urban and more competitive markets that were eligible for participation, versus hospitals that did not participate. The episode initiators were primarily non-profit and part of a health system.

Participating hospitals also had a range of historical payments, indicating BPCI Advanced’s success at attracting more diverse participants based on episode costs, the report stated.

“Across all clinical episodes, the average historical payments for the hospitals that chose to participate in BPCI Advanced were higher than the average for hospitals that were eligible but did not participate. The distributions of historical episode payments overlapped for hospitals that participated and those that did not. This indicates that the target pricing method was successful in attracting hospitals with a range of historical payments to the model,” the authors wrote.

Prior to the launch of BPCI Advanced, CMS sent historical episode payment data and preliminary target prices to organizations that submitted applications to participate. The information was meant to help organizations make decisions about participation, downstream episodes, and clinical episode selection.

However, this may not translate to greater Medicare savings, at least beyond Model Years 1 and 2. Early data from the model showed that hospitals tended to select clinical episodes for which they already had higher episode payments, which would make it easier for them to reduce costs and receive reconciliation payments during the first performance periods.

The top clinical episodes selected by hospital participants included congestive heart failure, sepsis, cardiac arrhythmia, simple pneumonia and respiratory infections, and chronic obstructive pulmonary disease, bronchitis, and asthma.

For all 32 clinical episodes, including the top five selected by hospitals, the median historical payment for hospital episode initiators was higher than the historical payment for eligible hospitals that did not participate in the clinical episode.

On the PGP side, there were 580 episode initiators in Model Years 1 and 2, although 74 of these participants did not bill Medicare for any services during the first six months of the model. Overall, PGP episode initiators were smaller in terms of the number of unique clinicians, discharge volumes, and the number of hospitals where clinicians admitted patients. The PGPs also included more surgical specialties and fewer primary care specialties compared to the original BPCI model.

Researchers, however, could not determine how participating PGPs compare to all eligible PGPs because of the frequency and ease of clinicians changing PGPs and for PGPs to form or dissolve. This also made it difficult to say with certainty whether BPCI Advanced will bend the cost curve for Medicare, the report indicated.

“Creating new TINs may have been a BPCI Advanced participation strategy for some PGPs,” the report stated. “Creating a new TIN can potentially be financially advantageous under the model because of the target price calculation method.”

“Furthermore, a clinician can choose which TIN will submit the claim to Medicare,” researchers explained. “For a clinician with more than one TIN, it would be advantageous to submit claims for patients with lower expected episode payments under the TIN in BPCI Advanced and submit claims for patients with higher expected episode payments under another TIN. While a totally appropriate billing practice, this could limit the reductions in payments achieved under the model.”

The Medicare Payment Advisory Commission recently identified a similar practice in accountable care organizations (ACOs). The Commission advised policymakers to prevent such gaming of the model to generate greater savings from ACOs.

But that is not to say bundled payment participants do not implement key practice transformations that could lead to savings down the line.

A new study published in BMJ found that participation in the original BPCI’s bundles for four common medical conditions was associated with savings after three years. These savings were generated by practice changes that reduced the use of high-intensity care after hospital discharge without affecting quality, which also suggested that bundles for medical conditions could require multiple years before changes in savings and practice emerge.

Researchers stressed that “if bundled payment programs are to achieve their intended purposes, policymakers must design policy in ways that address potential bias and ensure fairness to participants.” For investigators and policymakers evaluating bundled payment programs, this means paying careful attention to how episodes are defined, including the methodology for assigning episodes since BPCI Advanced rules have been associated with increased bias.

“Failure to do so risks distorting the apparent effectiveness of bundled payments as a payment model and dampening enthusiasm from both policymakers and would-be participants,” they stated.