It is now January 2013 and the “No-Risk” BPCI program has begun; how much more (or less) do we know about this program than we did nine months ago when CMS first issued its Request for Applications? In this paper, we will list what we know and highlight the remaining uncertainties about the Medicare BPCI program.
Data:
CMS has released some information about the data to be provided to BPCI applicant awardees.
We do know that the historical claims data, for price determination/validation, will be sent out in March. The data will include claims for beneficiaries who would have initiated episodes at the applicant’s hospital.
We do know that the data will include actual patient identifiers, enabling the applicant to track and perform root cause analyses on specific episodes. All applicant awardees will need to submit a new DUA to CMS for this research level data, which rises to the level of PHI under HIPAA.
We do know that, because that this data only includes claims for the specific applicant, it will not be usable for computation of benchmarks, validation of outlier limits, or anything else beyond validation of a specific applicant's payment rates.
We don't know whether CMS will filter the historical data to only include claims that would have actually been included in the episode, or whether they will simply include all claims for episode-initiating beneficiaries and require the applicant to exclude claims that are not part of episodes. We don't know which years will be included in the historical data, although it's likely to be 2009, 2010, and 2011, since 2011 is the most recent available year from CMS.
We don't know the format of the data, and specifically whether the claims will be "tagged" with episode identifiers, which would facilitate analyzing the episodes. Our best guess is that the data will be similar in format to the HRC data provided to BPCI applicants, which means that it will need to be processed using episode grouping software before it will be actionable.
We do know that in May, CMS will release six months of more current data to allow applicants to review/analyze their performance during the no-risk period.
We don't know what period will be covered by no-risk data, but it's likely to be for claims incurred beginning October 2012 and paid through March 2013. If this is the case, these claims will not sufficiently overlap the no-risk period nor provide for a meaningful analysis of the applicant’s performance during that period. This is because the episodes require between 30 and 90 days to complete, at least an additional 30 days for claims run-out, and probably another 30 days to produce the data file. This situation will be continuous throughout the BPCI implementation period, meaning that participating organizations will need to develop processes that work with incomplete episode data, such as IBNR calculations.
Episode Bundle Definitions:
CMS has provided detailed descriptions and details the MS-DRG-based bundles.
We do know that there will be a prescribed list of excluded diagnoses, procedures, and DRGs for each episode family. The notion of allowing providers to define their own episodes has been completely forsaken. The allowable episode lengths have also been prescribed at either 30, 60 or 90 days, nothing in-between. In effect, the BPCI episodes will resemble super-DRGs.
We don’t know whether CMS will issue its own software code for assigning claims to episodes, will tag each claim with an episode identifier, or leave the industry to build its own software.
Episode Price Determination:
CMS has provided a first look at its pricing methodology, but has indicated that there are several issues yet to be resolved.CMS will be sending data to applicants on its price calculation in March (see discussion on data above).
We do know that CMS wants to set a prospective episode price that provides a guaranteed savings to the Medicare program vis-à-vis the current fee-for-service program. Providers will then keep (or be reimbursed for) any savings beyond the CMS discount.
We do know that CMS is concerned about setting a price, based on historical data, for episodes with small volumes (e.g. a provider may not provide a significant amount of care for one or two MS-DRG(s) within a family). CMS would like to blend the provider’s data with claims from a larger population (e.g. region or nation), in order to smooth out variability caused by small sample sizes. CMS is exploring a few alternatives for accomplishing this blend (see our paper on options for price blending).
We don’t know which alternative CMS will choose, or what larger population(s) they will use for price blending.
We do know that CMS is considering allowing for the exclusion of episode costs that exceed defined outlier limits as part of the price-setting, payment and reconciliation process. CMS is considering allowing applicants the option of accepting or declining these outlier adjustments. This decision will depend upon the provider’s analysis of and attitude towards the potential risks and opportunities affected by the outlier limits (see our paper on the implications of outliers).
We don’t know how CMS will utilize the historical data for price setting. Will they average all three years equally or will the more recent year be given more weight? In addition, there are many changes in Medicare FFS payments from year to year, beyond the marketbasket update. How much of this will CMS incorporate into their methodology?
There are two reconciliation issues to consider under the BPCI demonstration program.First is the reconciliation between FFS payments and the episode price; second is the reconciliation of payments for the 30-day post episode period.
We don’t know how often the episode reconciliation process will take place, nor how lagged the claims data will be. Generally 2-3 months of claims run-out are necessary to get a reasonably complete set of claims.
We don’t know how the budgeted vs. actual payments/repayments between CMS and the awardee will be implemented.
We don’t know what CMS will use as the basis for the 30-day post episode reconciliation or what parameters will be applied.
The DataGen/Singletrack Team Approach
Singletrack Analytics has been working with our partners at DataGen for the past year to closely monitor the proposed BPCI demonstration program and its changes, develop models to evaluate the effects of these changes, and provide applicant awardees with the information they need to continue to assess their participation opportunities. For more information, contact Jonathan Pearce at jon@singletrackanalytics.com or (856) 762-0605 or Gloria Kupferman at gkupferm@hanys.org or (518) 431-7968.