Our analyses examined the number of settings and types of providers that were involved in episodes related to the nine study conditions. Across all nine conditions, there was no standard pattern of types of providers and settings involved in the management of the episodes. The variations in care patterns and trajectories observed across and within episode types signals potential opportunities (i.e., to dampen down on unnecessary variation in care) and challenges (i.e., a one-size-fits-all approach may not be feasible) when considering performance measurement and/or payment applications.
The number and types of settings involved in episodes varied across the nine study conditions. Fifty-seven percent of hip fracture episodes included more than four settings, and only seven percent involved a single setting. At the other extreme, 52 percent of low back pain episodes involved only a single setting (typically ambulatory care) while only five percent involved more than 4 settings.
The number and types of settings also varied among episodes related to a single condition. There was no standard care pathway, or combination of settings, for episodes related to any of the conditions. For example, the most common combination of settings in AMI-related episodes was hospital inpatient, hospital outpatient, and ambulatory, but this combination occurred in only 41 percent of AMI-related episodes. Fourteen percent of AMI-related episodes involved only hospital inpatient and hospital outpatient care, while 12 percent involved hospital inpatient, hospital outpatient, ambulatory, and home health. Episodes related to other conditions had even more permutations of settings involved (e.g., the most common combination for hip fracture-related episodes was hospital inpatient, hospital outpatient, ambulatory, and skilled nursing facility – only 16 percent of episodes). The second most common combination for hip fracture included these settings plus home health (13 percent of episodes).
Applying the ETG and MEG episode definitions, episodes related to chronic conditions (e.g., diabetes, low back pain, CHF) may or not include a hospitalization related to exacerbation of the condition. Fifteen percent of diabetes-related episodes, 11 percent of low back pain episodes, and 55 percent of CHF-related episodes included inpatient hospital care. In an application that would hold providers accountable for measures of resource use during an episode, this expensive inpatient care would lead to penalties for the providers accountable for these episodes; this may or may not be desirable depending on whether the hospitalization is potentially avoidable through appropriate management of the condition in the ambulatory setting.
Care patterns showed regional variation across the three states. Some of the observed variation is likely related to differences in the supply of different types of health care providers in different geographic health care markets. For example, inpatient rehabilitation facility (IRF) care was more common for episodes in Texas, where these type of facilities are relatively numerous. In Oregon and Florida, use of IRFs was less common than in Texas, but use of SNFs was more common. The implications of these supply-related variations in care patterns are not clear and could be considered in future research exploring the potential applications of episodes of care constructs.
The lack of standard, or even predominant, patterns of care for a large fraction of any particular episode type could present challenges to approaches that include an element of standardization (note: some of the variability may be an artifact of the way in which the grouper tools assign claims). For AMI-related episodes, should the episode definition and, in turn, performance measurement, encompass home health care if home health is provided for only a small fraction of all patients having an AMI event? Is it fair to compare the quality of episodes including home health with episodes that do not include home health? Or should all settings be included in episodes irrespective of the variation in the extent to which settings appear in an episode and that each setting has its own set of accountabilities—since patients will follow different trajectories based on market structures, provider management preferences, and patient characteristics? In some cases, care in a particular setting may itself be an indicator of poor quality – e.g., hospitalization for exacerbation of CHF—which would suggest the importance of a more inclusive approach to defining an episode. Again, future research would help inform these questions.
The variability in the number of providers and settings encountered during a patient's trajectory for any of the nine conditions we examined highlights the potential challenges for providers to coordinate care, or in some approaches, form virtual groups to assume shared accountability when these configurations are not reoccurring. Given that we examined only one condition at a time, the picture could be even more complex when attempting to group together a broader array of episodes that a Medicare beneficiary has in a given year when there are an even greater number of providers involved who theoretically could be working to coordinate the care for the patient. The involvement of multiple providers located in different settings poses questions about how a bundled payment for an episode would be distributed. Possible approaches suggested in the literature and expert discussions include predetermined arrangements between the providers and/or a Medicare formula for allocating payments.