Another Look at BPCI Risk Track Selection

As the deadline for submission of April 2015 applications for the Bundled Payment for Care Improvement (BPCI) initiative looms, many applicants are trying to figure out the best approach to selecting risk tracks for various episodes. Some participants are looking for "algorithms" that can assist them in selecting the most appropriate risk track, while others are noting that risk track can be changed throughout the participation period, and may wish to switch risk tracks based on their performance during a previous quarter.

Our approach is a little different. Risk tracks are designed to protect participants from unusually high-cost episodes, which in most cases are randomly-occurring events. This means that the events of a preceding quarter may not be representative of future periods. In addition, we believe that the selection of risk track depends on an evaluation of risk and opportunity among the high-cost cases, and that this evaluation is primarily a clinical one that involves examining those cases that exceed the risk track limits and evaluating whether they represent risks or opportunities. These evaluations should be made over the long term; short-term quarterly results may vary because of randomness, so strategy should be based on longer-term data.

Risk tracks reduce risk and opportunity

The effect of outlier limits on high-cost episodes is to reduce the cost of the episode when compared to the target price. For an episode whose cost exceeds the respective outlier limit based on the risk track, the episode cost is reduced to the amount of the outlier limit plus 20 percent of the difference between that limit and the actual episode cost. For example, if an episode has an actual cost of $100,000 and the outlier limit is $80,000, the amount of that episode applied against the target amount is $84,000 ($80,000 + ($100,000 - $80,000) * 20%). This process is referred to as “Winsorization”. It reduces the risk of uncontrollable high-cost case, but also reduces the opportunity to benefit from the reduction of costs in controllable high-cost cases, since only 20% of the cost reduction would be counted in comparison to target amounts. This issue is discussed in more detail in this and other Singletrack Analytics blog articles.

Calibrating your costs with the national averages

Risk tracks are derived from the national average episode costs for each DRG, adjusted for geographic differences. Nationally by definition, 1% of cases exceed the outlier limit for risk track A, 5% of cases exceed that limit for track B, and 25% of cases exceed the limit for track C. The national averages provide some level of understanding of the number of episodes that may be expected to exceed a particular outlier limit.

But individual providers may have average episode costs that are different from those in the national sample, so those results need to be "calibrated" against the national risk tracks. While 1% of the national cases exceed the outlier limit for risk track A, a particular provider may have a different number of cases that exceed that limit. This can be determined from longer-term historical data, as shown below. In this example of the risk track selector worksheet from the BPCI 360 analytical model using all episodes from the baseline to the current period, 9% of the episodes for the selected provider exceed the risk track B outlier limit, which was exceeded in the national data by only 5% of episodes. (The blue line indicates the actual episode cost of individual episodes, while red line indicates the "Winsorized" cost after the outlier limits applied.)

Also, 34% of that provider’s episodes exceeded the risk track C (75^th percentile) outlier limit (not shown). This provider appears to have higher average episode costs than the national averages, and therefore may expect slightly more episodes to exceed the outlier limits than the national percentiles.

Separating Opportunity from Risk

Once the percentage of cases that are expected to exceed the outlier limits in the participant's population has been established, the next step is to evaluate whether those cases represent risk or opportunity. This decision is based on whether clinical intervention that will impact cost of those cases is possible. In some episodes, high-cost cases may be driven primarily by costly post-acute providers, or by readmissions, both of which the care management teams believe may be avoidable. In the Episode Composition graph below for the Major Joint Replacements of Lower Extremity episode family in which each vertical bar represents an individual episode, the high-cost cases are generally driven by high post-acute provider utilization, which is a target of the care management strategy at this particular hospital. Selecting risk track B or C may limit the ability to create savings from reducing utilization in those episodes. Therefore, risk track A is probably the most appropriate choice.

By contrast, for the Congestive Heart Failure episode below, the cost of the highest 5% of cases may be uncontrollable; these patients are extremely ill and require significant amounts of acute care which may be delivered in the latter months of the episode. If that's the case, then the cost of those episodes represents risk, which means risk track B may be more appropriate.

Care management strategies and capabilities vary among institutions, and what’s characterized as risk in one facility may represent an opportunity in another facility. Therefore, risk track selection must be aligned with each facility’s plans for episode cost management.  

Risk Track C Is Different

While the differences between the effect of risk tracks A and B are relatively small, risk track C carves out risk and opportunity from about one quarter of all episodes. Participants should evaluate this risk track carefully, since in many cases those high-cost episodes may be the only opportunity to achieve savings. Note in both graphs above the small amount of post-acute cost (which is the only opportunity to achieve Medicare cost savings) that occurs in the lower-cost half of the episodes (the half of episodes to the left of the graph). In most episodes, the only opportunity for significant cost savings occurs in the highest cost of the percent of episodes (those to the right of the graph). By significantly reducing the potential savings achievable in half of the episodes in that group (the top 25% of all episodes), the participant may be eliminating the opportunity to recover the CMS discount, which applies to all episodes, and to create additional savings. Therefore, this risk track is only appropriate in those situations in which the opportunity to achieve Medicare cost savings is believed to be limited, and the participant’s focus may instead be on creating internal cost savings.

Summary

By definition, risk tracks affect the highest cost episodes, most of which are highly influenced by random variations in patient condition. Therefore, selection of risk tracks should be made utilizing large amounts of historical data, and a deep understanding the types of cases that fall above each risk track. These cases should be evaluated by clinicians to categorize those costs into controllable opportunities or uncontrollable risks. Only then can a knowledgeable decision about risk track selection be made.