In May, 2012, Bailit and Burns surveyed bundled payment programs nationally and found exactly 19 non-federal programs anywhere. Nine were Prometheus Payment implementations (Burns, 2012).
A year later they went back and found that of those programs, two were still conducting operational pilots, three were still in the process of developing a program, and five concluded or never started their pilots (Burns, 2013).
In the meantime, Medicare launched the Bundled Payment for Care Initiative (BPCI), involving more than 450 organizations, for which I provided a technical assistance Webinar on governance and contracting issues (CMS, 2012).
I also provided support to Mitre, the consulting firm that the Center for Medicare & Medicaid Innovation (CMMI) engaged to produce a monograph on “Contracting for Bundled Payment” (Mitre, 2012).
This means that most health plans have no experience with contracting for bundled payment. Yet the failure of most of the physician-hospital organization (PHO) ventures of the mid-1990s is likely to be replicated unless health plans approach today’s bundled payment differently from the take-it-or-leave-it approaches that have characterized most pay-for-performance programs, where the plan simply announces the amount that will be paid for the stated measures and providers either qualify or do not.
Bundled payment, properly constructed, gives the providers medical management risk — the risk of making good medical decisions within the payment boundaries, says author Alice G. Gosfield, JD.
In this article I will debunk some notions regarding bundled payment and address issues of fairness in episode construction and rules of engagement.
Because it has launched more bundles than other programs, reference is made to the Prometheus Payment model throughout. I am a member of the original Prometheus Payment Design Team and am the chairwoman of the board of the Health Care Incentives Improvement Institute (HCI3), but the opinions here are purely my own.
For the purposes of this article, bundled payment means a payment or budget which includes within it disparate providers who are not organizationally related. When a fully integrated organization that owns its hospitals and nursing facilities and employs its physicians accepts a single payment for all the services rendered, that is surely bundled payment, but the contracting issues are not those that arise when otherwise unrelated providers accept either a single payment made to one entity, which is then further disbursed, or a single budget for which all the participating providers share risk.
The Prometheus Payment model is a bundled budget approach that deploys a software package that allocates claims across the budget to the various providers. Payment can be made to the participants separately based on their performance.
Common risk pool
Although bundled payment can be made for a single day of care, the purpose of bundling payment is to put providers otherwise paid pursuant to disparate incentives into a common risk pool, which aligns their incentives. It is believed, although there is very limited data available (Thoumaian, 2010), that paying for services that extend across a time continuum is the best way to align incentives.
Therefore, unlike capitation, which generally covers all the conditions in the capitated population, most bundled payment models are oriented around episodes of care for a specific condition and typically, where a hospitalization is involved, include some pre-admission services when the condition at issue is a procedure or an acute medical condition. They also include the length of stay and some pre-determined post-discharge period.
While the Medicare bundled payment legislation calls for a 30-day post-discharge period, the Prometheus model uses 180 days post discharge, since most of the care a patient will receive after a hospital stay is likely to relate to the reason for the admission.
An entire year
In contrast, chronic care bundled payment episodes typically extend for a year to coincide with the patient’s premium payment, and because by definition the care is chronic and extended, the episode lasts an entire year.
The key difference between bundled payment properly constructed and capitation is that bundled payment does not give providers technical risk or insurance risk, otherwise known as “incidence risk,” that comes with paying the same amount regardless of the patient’s condition or needs.
Capitation is derived from insurance principles. Actuaries construct the premium price based on historical behavior of the providers in terms of utilization of resources. They project that utilization forward with some assumptions regarding the incidence of disease within the projected covered population.
These determinations are made completely devoid of any measurement whether the past performance was mediocre, stellar, or awful, or whether there was underuse or overuse.
As a result, in capitation, there is no way of knowing whether the amount to be paid will be adequate to provide the quality of care that any specific patient needs.
Bundled payment, properly constructed, gives the providers medical management risk — the risk of making good medical decisions within the payment boundaries.
Potentially the most significant issue of fairness in a bundled payment program is the foundation of the assumptions that were used in creating the bundled budget. If the providers are to respond to the intended incentives, they have to know what the bundle includes.
Clarity is critical
Even for those bundled payment programs that go no further than to measure past performance and link gainsharing opportunities to lowered expenditures, clarity and openness regarding the financial assumptions are critical and should be clearly reflected in contract language.
The real problem with this approach, however, is that no one knows whether the amount paid before, now expected to be reduced, is enough to provide appropriate care, especially since the payments are made patient by patient.
By contrast, the Prometheus Payment model bases its budget on evidence-informed case rates (ECRs), which are created by working groups of practicing physicians who take into account good clinical practice guidelines and consensus judgments as to what services are necessary to deliver what science says is necessary to treat the patient for the condition for which the payment for care will be bundled.
In the very first iterations of the ECRs — for diabetes, for example — with reference to the Institute for Clinical Systems Improvement (ICSI) guidelines, the physician working group determined that in the claims database that was used to model the rate, the physicians treating diabetics historically had not rendered enough office visits to manage care effectively, so an additional $1,000 was put into the ECR budget to account for the additional visits over a year (Gosfield, 2008).
The Prometheus case rate budgets are completely transparent, as should all bundled payment budgets be. The incentives will not work if the basis for the budget is not known.
Then there is the further problem of risk adjustment. Again, a bundled budget consisting purely of the dollars spent in previous years, reduced by some arbitrary amount, can disadvantage specific providers considerably by giving them more incidence risk.
Patient comorbidities would be expected to increase the potential budget because additional legitimate costs will be incurred with additional medical need. In the Prometheus Payment model, a software package was developed which can, without delay, adjust the budget upward according to comorbidities reported on the claims.
Of course, all of these programs assume that the diagnosis is correct, although misdiagnosis is well known as a problem in managing care.
But in designing new payment models and testing them, one has to assume good faith between plan and providers or much energy will be spent in trying to avoid potential difficulties that may never arise.
Rules of engagement
The providers will need to understand and the contract should address what triggers a bundle (e.g., an ICD-9 code, a CPT code) and whether the episode reaches back in time to capture services that preceded the trigger, e.g., the diagnostic services that led to the establishment of the diagnosis or the admission. Then, there must be clear rules about when the episode ends, e.g., the number of days post discharge and what “breaks” the bundle.
If the providers are being paid for a congestive heart failure bundle and the patient is in a serious car accident, do the payments continue or is the bundle broken? If the providers treating the patient for CHF continue to do so during the hospitalization for the motor vehicle injuries, then there is no reason to stop the payments.
But if a patient treated on a bundled payment for stroke care is in the same accident, it may not make sense to continue the stroke bundle. These rules ought to be clearly addressed or the contract should provide a methodology to address them.
For otherwise independent providers to manage care under episode bundles, they will need to know how well they are doing both in controlling their costs and delivering quality care, so what reports are made by the payer to them, how often, and how timely will matter.
But more than that, what information will the physicians be given about the hospitals’ performance and vice versa? These are also important contract terms if, over time, the physicians are to manage care both effectively and in response to the intended incentives.
Critical contract issues for the providers will include when reconciliation of budgets occurs and when payment gets made. This, inevitably raises the issue of dispute resolution.
Because most bundled payment programs are established through amendments to the basic provider agreement, an initial issue is whether the established plan appeals mechanisms are appropriate for the types of disputes that can arise in bundled payment. In addition, the contract should make clear what is and is not subject to dispute resolution.
It is entirely fair for the plan to take the position that the budget itself; the rules for triggering, breaking, and ending an episode; and the rules for severity adjustment are not subject to dispute resolution since they should have been negotiated as part of the contract initiation.
On the other hand, there are a host of issues which may well become subject to dispute resolution, including:
- Whether an episode was triggered or broken
- Whether a provider qualified for upside payment or should pay on downside risk, if there is any
- The amount of payment if it varies based on quality scores
- Whether a provider met quality or efficiency thresholds
- Whether the data supporting the payments was accurate
There are other issues as well, but the point is that traditional appeals mechanisms for denied prior authorizations or lengths of stay do not lend themselves to these new judgments.
If the plan is contracting with one provider entity (e.g., a large physician group, a PHO, a hospital) that will have the obligation to pay others, it is legitimate to seek an indemnification from the entity receiving the money that the plan will be held harmless for disputes among the providers.
But without imposing its will on the providers themselves, it is also legitimate for the plan to require them to have a program to manage internal disputes among themselves.
One of the essential issues will be the basis for allocating dollars among the disparate providers. When the bundled budget is based on clearly stated assumptions, then the parties can refer to those assumptions to mediate disputes.
Rules of attribution may be important if two providers seek the same portion of the gainsharing monies. If the plan is paying providers separately, this issue may also be a plan issue, and there are some choices to be made. One is to have attribution based on majority of visits, if, for example, the dispute is between a primary care physician and an endocrinologist.
Another choice is to require the providers to work it out themselves. A third choice might be a tribunal to hear the issue. Another choice might be, if there is a dispute over attribution, no one gets the money. The point, however, is not to dictate a choice, but to have the rules agreed to and stated up front.
Keys to success
Since there is so little experience with bundled payment, there are few long-term results to cite. Still, as of June 30, 2013 (HCI3, undated), the Prometheus Payment program had launched, with its plan partners, more than 7,100 bundles.
Three Blues plans (South Carolina, North Carolina, and Horizon in New Jersey) have reported their approach to relatively easy implementation, and Burns and Bailit (Burns, 2013) focused on Horizon and North Carolina as case studies.
Several themes emerge: Real collaboration with providers around issues including quality metrics and transparency of all data are key. Selecting provider partners carefully is also important since openness to new ideas is critical to managing the multiple moving parts of bundled payment. Being certain that participating hospitals and physicians work well together is important. Identifying senior leaders in the participating organizations and maintaining their engagement has been cited as well, and that means a bundled payment program needs high level plan executive support, if it is to be sustainable.
The promise of bundled payment is beginning to be seen in a few pilots around the country. Lessons have been learned already. To fulfill the potential of these innovations, the contracts between plans and providers should be fair, open, detailed, and clear. True collaboration will be an essential feature of successful bundled payment programs.
Burns and Bailit, “Bundled Payment Across the U.S. Today: Status of Implementations and Operational Findings,” (May 2012) http://www.bailit-health.com/articles/052912_bhp_hci_issuebrief_4L7.pdf.
Burns, Bailit and Huoy, “Bundled Payments One Year Later,” Health Care Incentives Improvement Institute, Inc. (May 30, 2013) http://www.hci3.org/content/bundled-payments-one-year-later
CMS (March 22, 2012) http://www.innovations.cms.gov/resources/Bundled-Payments-ADLS-4-5.html. The ideas there were further expanded in Gosfield, “Avoiding Food Fights: The Value of Good Drafting to ACO Physician Participants,” American Health Lawyers Association, Physician Organizations Practice Group newsletter, Vol 15, No1, (June 2012) pp. 10–12, http://www.gosfield.com/PDF/PhysOrg.Avoiding%20Food%20Fights.June%2020
Thoumaian, Mango and Mason, “Medicare Bundled Payments” in Institute of Medicine (US) Roundtable on Evidence-Based Medicine; Yong PL, Saunders RS, Olsen LA, editors. The Healthcare Imperative: Lowering Costs and Improving Outcomes: Workshop Series Summary. Washington (DC): National Academies Press (US); 2010. 15, Payments for Value Over Volume. Available from: https://www.ncbi.nlm.nih.gov/books/NBK53930/; and Vesely, “An ACE in the deck?, Bundled-payment demo shows returns for hospitals, physicians, patients” Modern Healthcare (Feb 7, 2011) http://www.modernhealthcare.com/article/20110207/MAGAZINE/110209990#ixzz2BvzXwzg1?trk=tynt.
Gosfield, “Making Prometheus Payment® Rates Real: Ya’ Gotta’ Start Somewhere” (June 2008) 15pp. www.hci3.org