The current issue of Hospitals and Health Networks magazine contains an informative article about the broader issues involved in contracting with payers on a "bundled payment" basis. The most common definition of bundled payment refers to the provider organization accepting a fixed payment for defined episode of care, generally beginning with an inpatient admission and continuing for a period of time after discharge. Episodes generally include all services provided to the patient (other than prescription drugs) that have not been specifically excluded. The contracting provider organization is responsible for managing the care of the patients throughout the episode, and is at financial risk if the cost (to the payer) of the episode exceeds a target price.
Together, our two organizations[1] worked with approximately 2 dozen hospitals at the beginning of the Medicare Bundled Payment for Care Improvement (BPCI) project, providing data and analytic support for the financial, strategic, and clinical analyses necessary to evaluate the opportunities and risks associated with participation in this program. Throughout this process we encountered many of the same issues described in the H&HN article; however in other cases our experiences differed from those described.
The Medicare BPCI Program Characteristics
In the BPCI program, episodes are defined based on DRGs. This differs from other bundled payment programs that utilize ICD-9 diagnosis and procedure codes (which will presumably be recast in ICD-10 terminology) to identify the initiation of the episode. For BPCI Model 2 participants, the episode includes all post-discharge services unless specifically excluded, and terminates between 30 and 90 days after discharge.
A significant factor of the BPCI program is that CMS receives its share of "savings" as an upfront discount (generally 3%) of the bundled payment fee for all patients, but there is no "shared savings" component thereafter. Therefore, a contracting organization that is able to reduce unnecessary services such as readmissions retains all of the savings, and can use those savings to fund care management initiatives or to provide financial incentives to providers. This is a major advantage of the BPCI program, and one that makes it significantly more attractive than other bundled payment initiatives that require sharing savings with the payer.
Analyses
For applicants evaluating BPCI participation, the first task was to understand the services that would be contained in the episodes for each DRG under consideration. To provide data for this evaluation, CMS delivered historical claims data for large geographic areas (generally including about 1 million Medicare beneficiaries) to prospective applicants. From this data, the applicant was required to construct the episodes applicable to their prospective populations. The amount of data provided was quite large – generally 20 to 40 GB – and was contained in more than two dozen different files for different provider types. It would be impossible to process this data using spreadsheets, as was described in the H&HN article, and we utilized enterprise-level database management systems to build these episodes. Fortunately, CMS’s episode definitions were generally well-defined, and we were able to create transparency in the data so that prospective applicants could validate the calculation process and understand the effects of various factors such as outlier limits and elimination of unrelated services. Other bundled payment systems may be based on "black box" software packages that do not provide a similar amount of transparency for the users to understand the effects of various components of the episode payment methodology.
In addition, most prospective applicants were primarily concerned with the "cost" (to Medicare) of the services in the episode, rather than their internal costs, since the relevant comparison is between the payment that they would receive under bundled payment relative to the amount received under fee-for-service payment. CMS provided revenue center detail for hospital charges that could be utilized estimate hospital costs, but none of our clients were focused on that issue during this time period.
Data issues
The most significant data related issue in dealing with bundled payments is that few providers (unless they operate their own insurance companies) will have access to claims data for all providers whose services are included in the episode. Therefore, prospective bundled payment participants must obtain data from the payer for these services in order to compute a reasonable episode price.
In addition, a comprehensive data set was necessary to understand the network of providers involved in providing care to patients throughout the episode. Prospective applicants were often surprised at the diversity of medical specialties involved in providing care throughout the post-discharge period, especially for complex medical cases. In addition, variations in use of post-acute providers (generally SNF, HHA and inpatient rehabilitation), and in cost and utilization patterns among these providers, were often enlightening to the hospital staff and physicians who previously had anecdotal knowledge of post-discharge care, but never had hard data to analyze.
Potential lost revenue
The H&HN article references the possibility that bundled payment works against the hospital's best interests by reducing the number of "heads in beds". This situation would rarely occur to BPCI participating hospitals for several reasons. First, the episode is initiated by an inpatient admission; i.e., a head in a bed. Bundled payment systems differ from ACOs in that there are no financial incentives to avoid the initial admission. In addition, there is no shared savings component in BPCI, which means that an avoided readmission does not necessarily result in a loss of revenue to the hospital because reduction in hospital payment would generate a savings of an equal amount. This situation may not occur in other bundled payment initiatives that do involve shared savings component, though.
Risk and opportunity
The H&HN article highlights the link between risk and opportunity, which is present in BPCI and most other bundled payment initiatives. We found this to be quite significant for BPCI applicants, especially for DRGs having relatively low volumes and higher inherent variation. (The inter-episode variation, as measured by the coefficient of variation, for medical DRGs such as CHF and COPD was generally twice the variation for surgical DRGs such as major joints or CABG.) For these DRGs, the random variation in cost among patients, even after eliminating "unrelated" services, was extremely high, sometimes creating the potential swing of 10 to 20% in the average annual cost of episodes in a particular DRG. This is one reason why most applicants elected to participate for Major Joint Procedures (DRGs 469 and 470), since DRG 470 has an extremely high volume and low amount of inter-episode variation. (See this blog article for a more detailed description of risk quantification.)
This type of variation is an important factor to consider in any bundled payment arrangement. One advantage of the BPCI project was that it was based on DRGs, which have a much broader scope than ICD-9 codes. While the use of ICD-9 codes to define episodes may allow a more granular definition, their use will also decrease the episode volume which will increase vulnerability to random variations. Thus, it will be important to model the risks associated with any proposed episode definition using data from the widest available population. Potential participants in bundled payment arrangement with payers should require those payers to quantify the episode cost variations (based on the payer’s proposed payment methodology) for all episodes occurring within the payer’s population – not just those episodes occurring at the applicant hospital.
Critical details in evaluating bundled payment programs
Like with any new payment system, potential participants must evaluate their organization's readiness to accept change, the level of physician participation required for success, and other critical organizational factors. However, two factors specific to bundled payment are critical in designing a structure with the greatest opportunity for success.
The first factor is the definition of the episode itself. As noted above, BPCI define episodes based on DRGs, which are relatively broad classifications of patients. These have the disadvantage of including patients that may have a variety of different health conditions, but also has the advantage of the relatively large patient episode population. In the article referenced above, we noted that population sizes of fewer than 50 patients generally have an unacceptable level of random variation unless the inter-episode variation was very low, as occurs with orthopedic and cardiac surgery patients. Unless that randomness can be eliminated through tighter episode definitions (which themselves will reduce the number of episodes), the broader episode definitions created by DRGs may ultimately create less random risk. Organizations such as Hoag Orthopedic Institute may, as described in the H&HN article, be able to dissect DRG definitions to a very detailed level, even to exclude patients having a BMI in excess of certain limits. However, other organizations with less market power may not be able to demand such restrictive definitions.
The other critical factor involves the distribution of shared savings. As noted above, CMS receives its "cut" through an “off the top” discount on all episodes. This allows the contracting organization to retain all of the savings that it creates through reductions in services for which providers are paid. Therefore, providers have no disincentive to working to reduce unnecessary services, since these reductions in payments will flow through as savings back to the contracting organization. By contrast, bundled payment arrangements in which savings are shared with the payer have conflicting financial incentives, since providers may not be able to recoup payments lost through utilization reductions.
The future
As the H&HN article notes, a lively debate continues between proponents of bundled payment and accountable care organizations as to which payment system holds the greatest future potential for hospitals. The report referenced in that article is available at this link. We continue to believe that well-structured bundled payment initiatives, such as the BPCI, offer the greatest near-term opportunity for hospitals to work with other types of providers to create more efficient patient care. While ACOs encompass a wider range of healthcare services, their breath and conflicting financial incentives may impede their growth and ultimate effectiveness. Participating in bundled payment initiatives such as the BPCI allows hospitals to play to their strengths in dealing with inpatients and recently discharged patients, rather than delving into the unfamiliar territory of population health management. Previous bundled payment initiatives such as the Acute Care Episodes demonstration project have shown some significant successes, while the results of population-based demonstrations such as the Physician Group Practice initiative have been uneven. The large participation in BPCI will create a unique opportunity to evaluate the effectiveness of these programs.