BPCI Episode Selection – An Analytic Approach

Submitted by jonpearce on Tue, 2014-09-23 17:02

Achieving success in the Medicare Bundled Medicare Bundled Payment for Care Improvement initiative requires analyzing historical data to identify the opportunities and risks associated in each of the episode families, as well as the other decision points in BPCI participation. Careful evaluation of opportunities and risks before making participation decisions can pay off significantly when the at-risk period begins.

The three major decisions to be made in selecting BPCI episodes are the selection of the episode family itself, the length of the episode and the risk track that defines the episode outlier limits. Each of these decisions can be facilitated by answering two questions – “What’s there?” and “What can we do about it?” The answers to these questions define the opportunity and risk present in each of the alternatives, and facilitate categorizing the characteristics of the decision into desirable opportunities or undesirable risks.

This article will review these decisions for several different episode families and explore how risk and opportunity can be assessed by a variety of analytical techniques.

Episode family selection

The first decision for a potential BPCI participant is in which episode families to participate. An episode family consists of a group of clinically-related DRGs defined by CMS, and includes all DRGs in the episode family. Many non-financial factors influence this decision, including involvement by critical physician leaders; however this paper will focus on the financial and risk aspects of these decisions.

For this analysis an episode composition graph is helpful. These graphs show a single vertical bar for each episode (patient), with the length of the bar showing the cost of that episode and the segments of the bar indicating the type of cost in that segment of the bar. The colors of the segments indicate the primary cost drivers of the episode.  Below is such a graph for DRG 470 (Major joint replacement or reattachment of lower extremity w/o MCC).


Figure 1- Episode Composition for Major Joint Surgery

In this graph, the answer to “What’s there” is that there’s a relatively consistent mix of either inpatient rehabilitation or SNF post-acute utilization across almost all of the episodes, except for a few readmissions driving up the high-cost cases at the end of the episode cost continuum. The “What can we do about it” analysis will require additional research, but frequently SNF costs at the higher-cost end of the graph are driven by SNF payment policies that encourage longer stays.  This issue can often be addressed by cooperating (and possibly gainsharing) with the SNFs. Readmissions are infrequent, and an initiative to reduce them may not have significant effect in this DRG.

That analysis can be compared to the graph below for a Congestive Heart Failure DRG. Here we see a high incidence of readmissions that are a major cost driver in almost all of the higher-cost episodes. SNF utilization is also a factor in those high-cost cases, but not in lower-cost cases. The “What’s here” answer is primarily readmissions.

Figure 2- Episode Composition for CHF

To assess opportunities in reducing readmissions, additional information is necessary, some of which is presented in the table below. This graph shows the DRGs for which patients are readmitted when the initial (index) admission is for CHF. Most readmissions are also for CHF, although septicemia readmissions are also significant. The period within the episode in which these readmissions occur will also be a significant determinant, and we’ll explore that analysis later in this article. The answer to “What can we do about it” therefore depends on the participants’ expectation of being able to reduce these readmissions.

Figure 3- Readmissions for CHF

These episode composition graphs are significantly different from the graph for Cardiac Valve procedures shown below. Here virtually all of the cost is in the index admission (dark blue bar segment) with very little post-acute care. Since a common objective in BPCI is to reduce post-acute costs, this episode family offers little such cost reduction opportunity (although there may be internal hospital cost savings opportunities). The answer to the “What’s there?” question ends up being “Not much”, relative to the post-acute opportunities for the other episode families.

Figure 4- Episode Composition for Cardiac Valve Surgery

Episode length

The next area for evaluation is the length of the episode.  Most participants elect 90-day episode length because the CMS discount is lower (2% vs. 3%) for the longer episodes.  However, a closer inspection of the opportunities (or lack thereof) in the longer-length episodes may indicate that shorter episode lengths may occasionally be more appropriate. The key decision point is whether the costs in the later periods of the episodes are manageable and therefore represent opportunities, or whether they simply represent risk.

Below is a graph of the cost composition of the post-acute period for a major joint replacement DRG, showing the relative costs for post-acute services in the first, second and third 30-day periods of an episode. Most of the cost is in the first 30 days of the episode. Here the answer to “What’s there” in the later part of the episode is again “not much”, but the implication is significantly different. In this case there’s little cost past the first 30 days, which means that there’s also little risk. But given that the participant can select the lower CMS discount rate without incurrent significantly more risk, this episode family may be a good candidate for the 90-day episode length even though there’s very little opportunity for cost reduction in the last 60 days of the episode.


 Figure 5-Cost Across Episode - Major Joint Surgery

By contrast, the same chart for congestive heart failure episodes shows significant costs in the second and third months, including a large amount of readmission cost in those periods.  Costs in the later periods, especially those of readmissions, should be carefully reviewed since they may not be manageable. In that case the costs in the later periods represent risk, not opportunity, and it may be advisable to choose the shorter episode length, perhaps 30 days.

Figure 6-Cost Across Episode-CHF

Risk Track Selection

The final BPCI selection is the choice of the risk track.  This determines the outlier limits to be applied to each episode’s cost, which limit the effects of high-cost episodes. Each risk track has a high and low outlier limit; episode costs that exceed the high limit are reduced by 80% of the excess over the limit, and costs below the low limit are reset to the low limit. The risk track selection affects many factors, including the average episode cost and the amount of variation in the episode. The risk tracks are based on percentiles of the national average costs of the episodes and are as follows:

Risk Track

High Outlier Limit

Low Outlier Limit


99th %tile

1st %tile


95th %tile

5th %tile


75th %tile

5th %tile


The risk track analyses below show the effect of each risk track on an episode family’s costs. The points on each line indicate the cost of each individual episode; the blue line shows the episode cost before application of the outlier limits, which the red line shows the effect of the outlier limit and indicates the cost of the episode that will end up being compared to the target amount and used in the reconciliation. The statistics show the average episode cost, number of outliers and other measures that result from the application of the risk track.

The importance of outlier selection is that cost reduction opportunities may exist in the higher cost cases, and those opportunities will be significantly limited by the application of outlier limits. If a high-cost readmission can be eliminated, the BPCI participant will receive the benefit of all of the cost savings, unless outlier limits take effect. Since the outlier limit may have significantly reduced the effective cost of the case, it also reduced the amount of savings that would be generated by eliminating that cost. Therefore, to the extent possible outlier limits should be set to limit risks, not opportunities. This issue is explored in more detail in Bundled Payment Analytics - the Effect of Risk Tracks and Outlier Limits on Payment and Incentives.

This requires examination of the Episode Composition graphs shown above, with particular attention to the high-cost cases that would fall under the outlier limits. The “What’s there?” and “What can we do about it?” analysis applies here – if there are significant high-cost cases they should be examined for manageability. Opportunities to manage controllable costs shouldn’t be limited by the risk tracks, but participants will want the risk tracks to protect them against uncontrollable costs.

Applying risk track A to DRG 470 shows that only a few higher-cost cases are reduced by the outlier limits. This may be an appropriate choice for this DRG since the few high-cost cases in this DRG are caused by unrelated readmissions that may be unavoidable. The remaining costs are primarily from post-acute providers, which is a major focus of care management initiatives in this DRG.


Figure 7-Episode Costs of Major Joint Surgery with Risk Track A

In Risk Track B for DRG 470, additional high-cost episodes are limited by the risk tracks. In this DRG those costs may have been manageable, so the application of the risk track may be reducing the participant’s share of those cost savings.



Figure 8-Episode Costs of Major Joint Surgery with Risk Track B

In Risk Track C below, almost half of the episodes have exceeded the outlier limit and their costs will be reduced. The effectively eliminates most opportunity to achieve reductions in the Medicare costs of these episodes. Participants who select Risk Track C generally do so because their strategy is to target internal hospital costs, not the costs to the Medicare program.


Figure 9-Episode Costs of Major Joint Surgery with Risk Track C


The three major BPCI participation decisions are the episodes in which to participate, the length of the episode and the risk track to select for each episode. These decisions can be facilitated by reviewing the analytics for the episodes under consideration and evaluating each by questioning “What’s there” and “What can we do about it”. Understanding the details of the cost drivers of episodes and coordinating that data with care management strategies provide an assessment of the opportunities associated with participation. Variations in cost that aren’t associated with opportunities are risks, and careful selection of episode parameters will allow participants to carve out risk while maximizing opportunity.