Combining Clinically-Similar Bundled Payment Episodes to Reduce Risk and Improve Care

Jonathan Pearce, CPA, FHFMA and Coleen Kivlahan, MD, MSPH

The Medicare Bundled Payment for Care Improvement (BPCI) program allows participants to assume financial risk for all Medicare services occurring within 30 to 90 day period after hospital discharge. Model 2 participants give up 2% of the episode target amount as a discount to CMS in 90-day episodes, but are allowed to retain any savings from Medicare cost reductions below the target amounts.

Some BPCI participants have initially attempted to limit their financial exposure by selecting episodes such as cardiac valves or coronary artery bypass grafts (CABG) in which they can control inpatient costs, there is little post-acute utilization, and minimal variation in average episode costs. Alternatively, participants may choose episodes such as congestive heart failure (CHF) that have significant readmission and post-acute costs, but the large amount of clinical variation in these episodes may create an unacceptably high variation in quarterly reconciliation amounts.

To mitigate some of these issues, some participants are beginning to combine related episodes such as CABG, valve, CHF, and AMI, recognizing that the increased volume, clinical overlap and shared interventions allow more systemic opportunity, as well as reducing statistical risk.  Considerations include:

  • Post-acute utilization management – the ability (or lack thereof) to create reductions in post-acute spending that cover the CMS discount amount, and
  • Within-episode variation – the amount of variation in episode costs that will create instability in quarterly reconciliations.

Post-Acute Cost Reduction Opportunity

In the Cardiac Valve episode family, the index admission accounts for about 60% of the entire cost of an average episode. The episode composition graph below shows the composition of each episode and highlights the predominance of the index admission for a large majority of episodes.

This means that the entire 2-3% CMS discount must be recovered from less than 40% of the remaining episode cost. Many BPCI participants have found this to be challenging. By contrast, in the CHF episode family shown below, only about 20% of the cost is in the index admission, providing a much larger opportunity to both cover the CMS discount and create additional savings.

The amount of reduction in post-acute cost needed to cover the CMS discount is a function of the percentage of total episode cost occurring during the post-acute period, which varies significantly among episode families. The graph below shows an example of the percentage reduction that would be necessary to cover a 2% CMS discount in institutional post-acute cost of various episodes:

High Episode Cost Variations

Participating in episodes such as congestive heart failure creates a different situation. Because of the large variation across cost of episodes within the CHF episode family, the average episode cost in a reconciliation quarter could vary significantly from quarter to quarter. The episode average payment trend chart below shows swings of $10,000-15,000 between consecutive quarters. This lack of stability in quarterly reconciliation results causes some hospitals to avoid participating in these episodes. A larger episode volume could decrease the extreme cost variation.

Combining Clinically Related Episodes

One strategy to reduce the impact of these issues is to simultaneously participate in several clinically related episode families. Patients rarely present with a single illness, especially chronic conditions. Patients requiring a CABG or valve replacement frequently have underlying CHF, and are at risk for ongoing arrhythmias and chest pain.  The care management strategies applied across these several episode families are similar and support a comprehensive approach to patients with cardiac disease. These patients require focused attention on the risks for readmission, post-acute care needs, home care, and comorbidity management.  Care management resources can be leveraged across a wider group of patients, the population will exhibit lower levels of variation due to the higher volume and patients with CHF and other chronic cardiac conditions can be identified earlier. This larger population, composed of multiple episode families, will have less variation than the individual episode families – even though the average episode costs within the individual episode families are different - because extreme values in a larger population tend to offset each other.

An example of the effect of combining these episode families for a single participant is shown in the table below. This table shows the annual episode count, quarterly coefficient of variation (CV), annual Medicare cost and PAC% of cost for each episode family, and for a combination of five episode families.  The quarterly coefficient of variation in average episode cost shows the amount by which the average episode cost in a quarter might vary[1] from the overall average episode cost across all quarters. A lower CV indicates more stability in quarterly variation. Finally, the post-acute percentage of total episode cost indicates the potential for creating savings outside of the index stay.

 

CHF

Cardiac Valve

CABG

AMI

PCI

Combined

Annual Episode Count

184

74

36

28

97

419

Qtr CV

14%

11%

16%

39%

20%

9%

Annual Medicare Cost

$4.8m

$5.0m

$1.8m

$.8m

$2.8m

$15m

PAC % of Cost

77%

28%

32%

73%

51%

52%

The combined group of episodes has the lowest CV, offering the most stable quarterly reconciliation. The PAC percent of the combined group episode cost is considerably higher than those of Cardiac Valve and CABG, indicating greater opportunity to create savings, but lower than CHF and AMI. The combined annual episode count is more than twice that of the highest volume condition (CHF), adding to the quarterly stability and potential for patient care impact.  Below is the Episode Composition graph for the combined episode groups, showing that there is opportunity for readmission and post-acute care management across the included conditions.  

 

The combined Episode Average Payment Trend graph is below, showing a more stable trend (as compared with a single condition such as CHF) cost across time, in spite of the differences in episode costs of the included episode families.

Conclusion

Combining clinically related episode groups can help overcome several barriers to participation in cardiac episodes, and potentially other similar episode groups First, it creates greater opportunity to achieve savings in the post-acute period by combining episodes with a variety of post-acute utilization patterns.  It also increases the total population size, which stabilizes the average episode costs and quarterly reconciliations. And most importantly, BPCI participants gain experience in managing beneficiaries with multiple related conditions as a population, rather than by individual condition.

Jonathan Pearce is a Principal with Singletrack Analytics. Dr. Kivlahan is a family medicine physician and a Senior Director at the Association of American Medical Colleges, where she directs their BPCI convener program.


[1] The CV is based on one standard deviation, meaning that about 68% of quarterly episode costs would be within this difference from the overall average cost.

 

AttachmentSize
Combining Clinically Similar Episodes_Final.pdf561.94 KB