This study provides an opportunity to explore additional research questions as the Assistant Secretary for Planning and Evaluation (ASPE) and the Centers for Medicare & Medicaid Services (CMS) continue to consider alternatives to the prospective payment silos in Post-Acute care (PAC). Reports by MedPAC (June, 2008) and The Commonwealth Fund (Schoen et al., 2007) discuss the potential for moving to episode-based payments to better align incentives across providers, and the Patient Protection and Affordable Care Act includes a pilot program for a bundled payment. Episode-based payments may give providers a financial incentive to be more efficient and to coordinate patient services across settings, potentially helping to improve health outcomes and reduce Medicare payments. The work presented here provides more information on episodes of care and on the PAC services that are included and excluded based on different episode definitions. The results of this work can be used to inform discussions around bundled payments and for understanding the service use trajectories of beneficiaries using PAC across the country.
In work with ASPE over the last several years, RTI International has constructed episodes of PAC using 2005 and 2006 five percent Medicare claims data. These episodes were defined as starting with an index hospitalization and included all PAC services as well as Part B physician claims. While episode payments may be a way to improve care coordination across settings and reduce Medicare spending, there is no consensus on the definition of an episode. The 60-day gap variable-length episode that RTI has examined in past work with ASPE (Gage et al., 2009) has been used to explore trends in PAC use. Under this episode definition, all acute and PAC services prior to a 60-day gap in services are included in an episode. However, there are many alternative episode definitions, including fixed-length episode definitions, some of which were examined by RTI and ASPE in 2009 work (Morley et al., 2009). Fixed-length episodes—for example 30 days following discharge from an acute hospitalization—may provide administrative ease, but there is debate on how long episodes should be given that a beneficiary may have several unrelated services during a potential episode. Fixed-length episodes may also exclude services that are related clinically but initiate beyond a fixed period. This issue is of particular relevance for beneficiaries with longer PAC use trajectories. It is important to examine the impact of different definitions as policy makers consider alternatives.
In the current work, RTI has expanded the data file used in the episode analysis in terms of both sample size and the number of years of data used in order to provide more detailed information on the characteristics of PAC episodes under different definitions. The data used in this work include 30 percent of episodes initiating with an acute hospitalization, 30 percent of episodes initiating with home health (HHA), and 100 percent of episodes initiating in a long-term-care hospital (LTCH) or inpatient rehabilitation facility (IRF) in 2006, 2007, and 2008 Medicare claims. Expanding the analytic file provides information on changes in PAC use over the period 2006-2008, allows for a more detailed understanding of the patterns of PAC use by geography, and provides the opportunity to follow patients over time. This work also differs from past work looking at episodes of care in that it includes analysis of PAC use for beneficiaries without an acute hospitalization at the start of an episode. Although much of the discussion surrounding episodes of care focuses on PAC use after a hospitalization, many beneficiaries are referred to HHA from physician offices, and a smaller number of beneficiaries enter IRF or LTCH without prior acute hospitalizations. Understanding use patterns for these types of beneficiaries is important in establishing context for discussions on how episodes of care are defined.
The next sections of this report describe the data sources, methods, and key findings from this set of analyses. Section 2 describes the analytic samples, the episode definitions explored, and the methods for standardizing payments. Section 3 presents the results of analyses of acute hospital-initiated episodes, including a discussion of the types of cases, episode patterns, and service use under alternative episode definitions. Section 4 presents the results of analyses of community-initiated episodes for beneficiaries entering IRF, LTCH, or HHA without a prior acute hospitalization. For each type of community-initiated episode, we examine the types of cases treated, patterns of care, and use of the range of PAC services under different episode definitions, and we compare these beneficiaries to those with acute hospital-initiated episodes discharged to IRF, LTCH, or HHA as their first PAC setting. Section 5 stratifies episode descriptives based on whether beneficiaries die during their episode in order to compare use patterns for those who die during the episode to against patterns for those who survive to the end of an episode. Section 6 provides detail on how PAC use and PAC episode payments differ by geographic area to demonstrate the implications of setting episode payment policy in different areas of the country. Section 7 presents the results of a cohort analysis where use of acute and PAC services was examined over a 2-year period for beneficiaries with initiating events in 2006. This includes analysis of the percentage of beneficiaries using PAC services in 30-day windows following discharge from their initiating event as well as mean payments per PAC user overall, by type of service, and by Medicare Severity Diagnosis Related Group (MS-DRG). Section 8 includes a discussion of the implications of these results for payment policy.