In 2008, the Medicare program spent $49.9 billion on post-acute care (PAC) services including skilled nursing facilities (SNFs), home health agencies (HHAs), inpatient rehabilitation facilities (IRFs), and long-term care acute hospitals (LTCHs) (Medicare Payment Advisory Committee [MedPAC], 2009). This represents an increase of 8 percent over 2007 Medicare spending on PAC; much of the increase was driven by spending on HHA and SNF (MedPAC, 2009). Given the payment silos for each provider type in the current PAC payment systems, there is little incentive for providers to work across settings to improve coordination of care and achieve efficiencies over a beneficiary's course of illness.
One idea that has recently been raised by policymakers to improve incentives for coordination and efficiency is "bundled payment." Under a bundled payment, one payment would be made per episode of care. An accountable entity (i.e., a hospital, an insurer, a PAC provider, groups of providers or other organization) would then be at risk for the care that patients receive over the episode. Recently, several Medicare bundled payment proposals have been discussed (Office of Management and Budget, 2009; Senate Finance Committee, 2009; Congressional Budget Office, 2009; MedPAC, 2008). Each proposal recommends making a bundled Medicare payment for hospital and PAC services, although the proposals differ in which PAC services are included and how many days' services are included. These proposals also vary in how the end point of a fixed episode is defined (i.e., any claim initiating within 30 days versus 30 calendar days after discharge).
The goal of the work presented here is to inform this discussion of bundled payments by providing data on the implications of different episode definitions, including difference in average program costs when one includes or excludes hospital readmissions or uses different end points for the episode. The episode definitions that we explore in this work include both fixed episodes (30, 60, or 90 days) and variable length episodes defined as ending with a gap in service use (30 day, 45 day, or 60 day gap). Average Medicare payments per episode and per service are calculated for 18 different definitions of an episode (see Table 1 for list of episode definitions). These definitions are based on analysis of Medicare claims data reflecting actual utilization patterns of acute, HHA, IRF, SNF, LTCH, and hospital outpatient department therapy services. In addition, this work explores differences in PAC payments across geographic areas in order to give more context to discussions regarding PAC use and payments nationally.
In this work, we use a beneficiary-level data file of 2006 Medicare claims constructed in prior work for the Office of the Assistant Secretary for Planning and Evaluation (ASPE) examining patterns of PAC use for Medicare beneficiaries (Gage et al., 2009). This file is unique in its ability to track beneficiary service use across settings following an index acute hospital admission. Here, we expand our analyses to explore fixed PAC episode definitions and variable length episodes defined by a gap in service use. For each of the fixed definitions, we examine alternative methods of treating the last claim in an episode by including all days and dollars associated with a claim that initiates within the fixed time period rather than prorating days and dollars based on the portion of services that occur within the fixed time period. Another alternative end point examined for each of the fixed and variable length episode definitions is acute hospital readmission. In total, our analyses compare the use of PAC services under 18 different PAC episode definitions. Analyses of these episodes will demonstrate the effect of different definitions as we examine options for potential payment bundles based on episodes of care.
Each of the alternative episode definitions is compared on:
- Percentage of beneficiaries with a claim for specific settings of PAC,
- Length of stay associated with the use of specific settings of PAC, and
- Medicare payments associated with the use of specific settings of PAC.
In addition to analyzing the services included under different episode definitions, we examine variations in PAC episode payments across geographic areas. These analyses are presented at the national and state levels, as well as at the level of the core-based statistical area (CBSA), which includes both metropolitan and micropolitan areas. In order to allow for comparison of Medicare payments across these geographic areas, we standardized the PAC payments to remove payment variation related to geography including wage differences and urban/rural location, as well as other policy considerations that affect payments such as indirect medical education (IME) and disproportionate share hospital (DSH) payments. By comparing standardized payments, we can learn more about differences in patterns of use in the absence of payment adjustments.
The next sections of this report describe the data sources, methods, and key findings from this set of analyses. Section 2 discusses the alternative episode definitions and methods for standardizing payments. Section 3 presents the results of the analyses decomposing the use of PAC services under the alternative episode definitions including percentage of beneficiaries using each PAC service and their corresponding lengths of stay and Medicare payments under each of the definitions. This section also presents the results of the geographic benchmarking analysis. Section 4 includes a discussion of the implications of these results for bundled payment policy.