HFMA Comments on CMS Bundled Payments for Care Improvement (BPCI) Initiative

May 12, 2015

Patrick Conway, MD
Deputy Administrator for Innovation & Quality, CMS Chief Medical Officer
Centers for Medicare & Medicaid Services
Department of Health and Human Services
Hubert H. Humphrey Building
200 Independence Avenue, SW, Room 310G
Washington, DC 20201

Re: Center for Medicare & Medicaid Services Bundled Payments for Care Improvement (BPCI) Initiative

Dear Dr. Conway,

The Healthcare Financial Management Association (HFMA) would like to thank the Centers for Medicare & Medicaid Services for the opportunity to proactively comment on issues related to the Center for Medicare & Medicaid Innovation’s Bundled Payments for Care Improvement initiative.  

HFMA is a professional organization of more than 40,000 individuals involved in various aspects of healthcare financial management. HFMA is committed to helping its members improve the management of healthcare delivery systems, comply with the numerous rules and regulations that govern the industry, and further the principles of administrative simplification.

HFMA fully supports CMS’s goal of transitioning 50 percent of Medicare fee-for-service (FFS) payments to value-based arrangements (defined by CMS as models such as Medicare Shared Savings Program (MSSP), Pioneer ACOs, Next Generation ACOs, Bundled Payments for Care Improvement, and other similar models) by the end of 2018. Given the significant interest in BPCI to-date from physicians, hospitals, and post-acute care providers, episodic payments are an integral part of achieving CMS’s stated goal of shifting financial incentives to reward the value of care provided. However, we are deeply concerned about BPCI’s ability to reach its full potential. We continue to hear feedback from members that there are significant design issues with the episodes and participants in the program face many operational and administrative barriers.  

Below please find our members’ detailed concerns with the BPCI program along with recommendations for addressing these issues. HFMA believes strongly in the need for payment reform in order to realign incentives and would welcome any opportunity to work with CMMI and CMS to reshape the BPCI and other value-based initiatives into ones that are beneficial for patients, providers, and the Medicare Program. 

Risk Adjustment: HFMA believes the current mechanisms CMMI is using to risk adjust the BPCI program are insufficient.  

First, the current bundles are based on “clinical episode families” comprising groups of related MS-DRGs. Analysis has shown that there is significant cost variation across MS-DRGs within the same clinical episode family. This will have the most significant impact on Model 2 and 3 participants. Based on analysis by the Health Care Incentives Improvement Institute (HCI3), the average episode cost for major joint replacement is $20,522. However there is significant variation in the episode prices in the underlying MS-DRGs (469 – $32,345; 470 – $19,638) ¹ . Even a slight change in the mix of cases compared to the baseline data will have a significant impact on the participant’s actual performance relative to the target price. This impact will be the result of chance, not efforts (successful or otherwise) to re-engineer care delivery for major joint replacement procedures. 

There is also significant cost variation within a MS-DRG depending on the patient’s underlying diagnosis code. For example, based on analysis by HCI3, the average episode cost for percutaneous coronary intervention  is $15,693. However, for patients with acute myocardial infarction/cardiac dysrhythmias, it’s $17,293, for those with stable coronary artery disease, it’s $14,147, and for those with other principal diagnoses, it’s $18,292 ² . Again, any variance in the distribution of primary diagnoses during the performance period relative to the distribution during the historical period used to set the target will impact a participant’s performance. As a result, a participant’s financial results reflect both changes in care delivery and random variation.   

Recommendation: CMMI needs to incorporate a mechanism to eliminate the impact of random variation on the financial outcome of episodes. Without this modification to the program, participants are managing not only performance risk but insurance risk as well. Given that the price variation occurs in the post-acute portion of the episode, we believe that CMMI needs to develop target costs based on the patient’s principal diagnosis or develop clinical families of related principal diagnoses that have similar cost profiles. While CMMI could retrospectively adjust the target price based on the mix of MS-DRGs, HFMA does not believe this is a viable long-term solution. First, given that MS-DRGs were developed as a measure of resource utilization in acute settings, they may not be the best predictor of necessary post-acute resource utilization. Second, as discussed below, HFMA strongly believes participants need a benchmark that is prospectively set and not subject to subsequent adjustments.  

Second, HFMA is also concerned that participants whose patient population includes a larger share of dual eligible patients could be disadvantaged in an episodic payment model. As research has shown, dual-eligible patients typically incur a higher cost, particularly across longer episodes. While this may not be as much of an issue with procedure-focused MS-DRGs, it causes concern for medical MS-DRGs (e.g., pneumonia, congestive heart failure).

Recommendation: CMMI should incorporate some level of adjustment for socioeconomic (SES) factors into each bundle’s target price. As a long-term solution, HFMA encourages CMMI to explore incorporating the National Quality Forum SES risk adjustment measure once it is adopted. However, in the interim we ask that CMMI explore basing SES risk adjustment on a hospital’s SSI ratio as a proxy. For post-acute providers, as an interim step toward SES adjustment, CMMI should calculate an SSI ratio for the facility and incorporate that into target price setting.

Third, while CMMI attempts to control for risk using winsorization, this still leaves organizations attempting to participate in episodes with relatively small volumes exposed to insurance risk. For example during Phase 1, an organization with approximately 300 joint replacement procedures incurred a significant loss relative to the target price on just two cases. One of the cases was a non-elective joint replacement for an individual with late stage chronic disease.  The individual required an extended stay in a post-acute facility not as a result of the joint replacement procedure but due to the underlying chronic condition. In this example, the winsorization and related risk track did not reduce the organization’s exposure to extreme outliers sufficiently to shield it from insurance risk. Had the participant been in Phase 2, the organization would have owed CMMI a significant payable and as a result has elected not to proceed to the risk-bearing phase. 

Recommendation: CMMI needs to take additional steps to reduce participants’ exposure to extreme outliers beyond winsorization for participants with relatively low volumes of episodes. If additional statistical trimming is not possible, CMMI should consider other program design changes. As an example of a possible solution for participants with volumes below a specific threshold, CMMI should consider offering an upside-only model similar to what is currently available to participants in the first contract period in the Medicare Shared Savings Program. While CMMI and CMS would like to see physicians, hospitals, and post-acute care providers engage in two-sided risk models, it may not be practical for organizations with relatively low episode volumes to do so as they will expose themselves to insurance risk, which they are not positioned to successfully manage.  

Further, for all participants CMMI should implement an additional stop-loss threshold beyond those currently available via winsorization for catastrophic cases where the actual cost (payments from CMS to the various providers involved) exceeds 25 percent of the target price.

Long-term, CMMI needs to work with state Medicaid programs and private health plans to implement a refined (as discussed in this letter) episodic payment methodology across payers. In doing so, CMMI needs to explore methods that would allow participants to aggregate volume across payers (providing a larger base to improve statistical stability of target and actual pricing) while maintaining separate target prices for each payer that is reflective of historically negotiated or administratively set rates for units of service. Beyond improving the statistical stability of calculations related to a given episode (which would likely encourage more organizations to participate) doing this would align incentives across payers for episodes which would have a multiplier effect on the inherent incentives to redesign care delivery for an episode. Further, to the extent that administrative processes related to episode management are aligned (similar to the episodes developed under Arkansas’s Health Care Payment Improvement Initiative), it would reduce administrative costs and reporting burden. 

Re-Pricing the Target Price: CMMI recalculates the target price in the initial and subsequent reconciliation periods. This was not communicated to participants prior to the start of the program as they were under the impression that the target price CMMI communicated to them at the outset of the program would be the final target price used for all reconciliations. Further, CMMI initially stated that the target price would be trended forward based on state experience.  

Currently, the target price changes based on precedence rules and adjustment of national trend factors. Given that this can occur through several reconciliation periods, the actual target price is unknown for up to a year after the conclusion of an episode. The quarter-over-quarter change is capped at plus or minus 3.5 percent. The maximum adjustment (3.5 percent) is greater than the discount rates offered in both models two and three. Beyond the sheer size of the potential target price adjustment, CMMI’s approach to rebasing is detrimental to participants for the following reasons:

• A shifting benchmark makes it difficult to focus on changes that will reduce episode spending when the baseline fluctuates without sufficient explanation as to why it changed. These random fluctuations can have a de-motivating impact on individuals involved in care process redesign.
• The reasons for the changes in the benchmark appear to be random within episodes, making it exceedingly difficult to communicate the reasons for the change to participating providers. This could strain the relationship between collaborating organizations. The risk of this occurring is exacerbated for participants who have gain-sharing agreements with physicians and other providers.
• A national trend factor, as opposed to a state one, will disadvantage organizations in high cost-growth areas while rewarding organizations in low-cost growth areas. Instead of being rewarded for making care delivery more efficient, providers are rewarded or penalized based on geography.

Recommendation:  CMMI should move to a stable prospectively set target price. Not only is this accepted practice in commercial bundling arrangements, but CMMI is implementing a similar conceptual model in the “Next Generation” ACO program in response to criticisms about the benchmark in the Pioneer program. Additionally, any changes in the composition of BPCI participants in a market that would impact assignment of cases based on precedence rules should be factored in at the beginning of a new contracting period. Implementing these changes to the program will give participating organizations a clear target to work toward.  

Additionally, for future programs, CMMI needs to communicate issues related to setting and updating target prices much more clearly so that providers have sufficient information to determine whether or not they want to participate.

Gainsharing Arrangements for Awardee Conveners:  HFMA is deeply concerned by CMMI’s decision to predicate gainsharing for awardee conveners undertaking multiple episodes on the results of the episodes in aggregate as opposed to the results of the individual episodes.  

As a simple example, assume an awardee convener agrees to move into phase two with two episodes—major joint replacement of the lower extremity with a group of community-based orthopedic surgeons and major bowel procedures with a group of community-based gastrointestinal surgeons. The awardee convener, working with an orthopedic surgery group, generates $100,000 of savings relative to the target price on the major joint replacement episode. However, the awardee convener and the gastrointestinal surgery group generate $100,000 in losses relative to the target price on the major bowel episode. 

Under CMMI’s current articulation of its gainsharing rules, the results would be netted leaving no dollars to share with the orthopedic surgery group.  This, despite the fact that the savings generated by the orthopedic surgery group on the major joint replacement episode have nothing to do clinically with the losses generated by the gastrointestinal surgery group on the major bowel episode.  

Stated another way, had the two groups elected to convene episodes themselves (as opposed to partnering with a health system to better align incentives across the care continuum), the orthopedic surgery group would be eligible to keep all of the savings it generated while the gastrointestinal surgery group would be responsible solely for the losses it generated.

Beyond the fact that this was not clearly communicated to participants, it is extremely problematic for the following reasons:

• By netting the results of gainsharing across multiple episodes for awardee conveners, CMMI is violating one of the basic tenets of incentive design by inappropriately holding providers at risk for performance results they cannot influence.
• It rewards awardee conveners and their physician partners for pursuing only episodes with a clearcut opportunity to reduce the overall cost of care. It discourages awardee conveners and their physician partners from accepting performance risk for episodes where the opportunity for improving the cost of care is either limited or uncertain. This is undoubtedly an unintended consequence given both the potential for the BPCI program to improve outcomes for Medicare beneficiaries and CMS’s stated goal of bringing 50 percent of Medicare fee-for-service payments under a value-based model by the end of 2018.

Recommendation: HFMA strongly recommends CMMI allow awardee conveners to develop and execute separate gainsharing arrangements that are tied to the individual episodes for which they assume performance-based risk. The ability to share gains with physicians and other participants should be predicated only on the episode-specific outcomes, not the results of all of the episodes aggregated. Moving to episode-specific gainsharing will hold the physician group(s) that is directly involved with re-engineering and managing beneficiary care for an episode responsible only for outcomes that they can control.

Data Issues: Beyond the timeliness of when participants receive performance data from CMMI, HFMA has identified three significant issues related to data-sharing in the BPCI program.  

First, the ongoing monthly data feeds from CMMI omit data elements that are used by the contractor to identify and reconcile episodes to target prices. Without these data elements, it is impossible to replicate the reconciliation results calculated by CMMI. Participants are left to hope the contractor did not make an error in reconciling the data. This practice would not be acceptable in a commercial episodic payment contractual arrangement.

Recommendation: CMMI needs to provide participants with all data, methods, and underlying calculations necessary to replicate the reconciliation results. This check is necessary to ensure that inadvertent errors did not occur (as will happen from time to time) when the contractor reconciled actual episode prices to targets. The ability to replicate the reconciliation results also helps maintain a transparent and open relationship among the BPCI participant, CMMI, and CMMI’s contractor.  

Second, HFMA continues to hear from both current and potential BPCI participants that the process of identifying new episodes to participate in is challenging due to a lack of flexibility on the part of the program. If a participant or potential participant has requested data for an episode of a specific duration (e.g., PCI 90 days) for initial analysis, the program will not accommodate subsequent data requests for the same episode of a shorter duration (e.g., PCI 30 days). Instead, the participant (or potential participant) will need to manipulate the data themselves to understand the potential opportunity created by participating in the episode. While this is possible for the example above, if a participant (or potential participant) had initially asked for an episode of shorter duration (e.g., PCI -30 days) and wanted to evaluate the potential of a longer episode, they would have insufficient data to complete the analysis.  

In some instances, additional data requests were accommodated. However, there were still issues with the satisfaction of the request as data was not provided in a timely manner to complete the analysis prior to CMMI’s deadline. Given the issues CMMI is having satisfying data requests from both participants (and potential participants), this calls into question the organization’s ability to manage the program beyond a relatively limited pilot.    

Recommendation: CMMI and CMS need to develop data capabilities that are more responsive to provider needs. CMMI also needs to impose more strenuous service-level agreements on the contractor it uses for data extraction.

Third, CMMI initially specified a file layout for the data it would send participants to support their efforts to appropriately manage patient care. Based on this file format, participants developed analytic tools to turn the data into actionable information for care improvement. However, CMMI, without warning, changed the file format, making it impossible to upload files for analysis. This significantly delayed participants’ ability to address variances with CMMI and provide feedback reports to participating physicians.

Recommendation: CMMI needs to develop a consistent file layout/format. Once the format is established, CMMI should not make additional changes.  If a change is unavoidable, the change should be communicated with sufficient lead time to allow participants and their consultants to change their analytic systems.

Administrative Burden: HFMA continues to hear from participants that the BPCI program entails a significant volume of administrative work, which drains time and resources from patient care activity. Examples cited include:

• Overlapping reporting requirements related to updating program implementation plans and data requested by Lewin in an attempt to identify best practice/shareable care protocols.
• Meeting monthly with program monitors.

Recommendation: CMMI needs to look for opportunities to eliminate redundant administrative activities. As an example, Lewin should be required to use the updated program implementation plans to search for best practices. Additionally, HFMA wonders if it’s premature for CMMI to seek out best practices. The program, due to the time in-field, has produced limited data from which to extrapolate and analyze results. Given the relative lack of quantitative results from the program, CMMI should allow sufficient time for the program to generate results before it attempts to define best practices or make other significant changes to the program.     

Physician and Post-Acute Education on MS-DRGs: CMMI has chosen to organize the clinical episode families that comprise BPCI’s episodes around MS-DRGs. While the MS-DRG construct is one that is familiar to acute-care facilities, it is not a native concept for physician practices or post-acute care providers. For physician practice and post-acute care settings, this poses an additional barrier to participation for these organizations must learn the nuances of the MS-DRG system.

Recommendation: At a minimum CMS and CMMI should develop baseline education materials to help physician practices and post-acute care providers develop a functional understanding of the MS-DRG system.

We are at your service to help CMS gain a balanced perspective on this complex issue. If you have additional questions, you may reach me or Richard Gundling, Vice President of HFMA’s Washington, D.C., office, at (202) 296-2920. The Association and I look forward to working with you.

Sincerely,

 
Joseph J. Fifer, FHFMA, CPA
President and Chief Executive Officer
Healthcare Financial Management Association

Footnotes

^1,2 Health Care Incentives  Improvement Institute Issue Brief, “Designing the BPCI for Success,” 2012.