The analyses presented in this report demonstrate the effects of alternative episode definitions in terms of episode costs and composition. In summary, varying the episode definition affects
- The number of beneficiaries we see using different PAC services,
- The types of PAC services included in an episode, and
- The levels of utilization and payment for different PAC services during an episode.
Our results indicate that the PAC services that are most highly sensitive to episode definition are SNF and HHA because these services are characterized by longer average lengths of stay. In addition, although these services may be the first sites of care for some beneficiaries following discharge from the acute hospital, they may also be later sites of PAC for beneficiaries discharged to other settings (e.g., IRFs or LTCHs) following their index acute hospital stays. Differences in the utilization and payment amounts per episode are sensitive to the definition of fixed end points (i.e., allowing any claim initiating within a period to be part of an episode versus prorating the dollars associated with the claim based on the days within the fixed window). Allowing any claim to initiate within a fixed period of time will capture continued service use. In contrast, prorating only captures a portion of actual service use (and costs) within the bundle.
The results of this work also highlight differences in the cost and composition of PAC episodes depending on whether acute hospital readmissions and subsequent PAC are included in the definition or whether episodes end with an acute hospital readmission. Mean PAC episode payments including acute hospital readmissions and subsequent PAC are substantially higher than when readmissions are excluded because of the high costs of readmission and the relatively high rate of readmission across beneficiaries discharged from acute hospital stays.
Differences in PAC payments by geography are another important consideration emerging from this work. The results show that there is significant variation in the percentage of beneficiaries using PAC services in different parts of the country. For those who do use any type of PAC service, there is significant variation in the types and levels of care they receive. This has implications for potential bundled payment policy: at what level would a bundled payment be made? Would a bundled payment be made based on discharge from an acute hospital? Or would it be made based on beneficiaries discharged to PAC? As shown in the tables, the average payment per discharge is substantially lower than the average payment per PAC user or service user. Further, the average payment per LTCH is substantially higher than average payments for IRFs, SNFs, or HHAs. These differences have implications for access to appropriate services. These decisions are further complicated by the differences in services available in a local area. Average costs would vary tremendously depending on which types of services are available and used. The research shows that some of this variation can be predicted by type of index hospital condition. Still, within conditions, utilization varies depending on the services that are available in a local area. Bundling payments that are averaged on different size denominators may further influence the availability of these services in less populated areas.