Why Some DRGs May Never Be Appropriate for Bundling

 Initially some BPCI participants were intrigued with the idea of participating in limited, procedurally-based episodes such as percutaneous coronary interventions (PCI) episodes in the Medicare Bundled Payment for Care Improvement program. Many hospitals have sufficient episode volume to avoid large random variations in episode costs, average episode cost is relatively low, and participation allows involving several diverse areas such as emergency room, cardiologists and cath lab staff.  However, after reviewing several initial reconciliations with participating hospitals, we’ve come to the conclusion that episode families like this contain such significant risks that they may never be appropriate for bundling.

The first issue is the need to recover the initial 2% CMS discount before the participant breaks even. This is complicated by the fact that a large portion of PCI episode cost – generally more than 50% - is from the index admission from which there’s little opportunity to reduce costs. (This can only occur if there are significant IPPS outliers in the base period that can be eliminated in the performance period, which is unlikely in these DRGs.) Other episode families such as major joint replacements (MJR) or congestive heart failure (CHF) have a far lower percentage of episode cost from the index admission, more resource use in the post-acute period, and therefore have a larger cost pool from which savings can be achieved and the CMS discount can be recovered.

The second issue is the potential occurrence of unanticipated high-cost readmissions. As compared with episodes like MJR that have few high-cost readmissions, PCI episodes have experienced readmissions in the costly “PRE" major diagnostic category (generally transplants, ECMO, and tracheostomies) with costs in excess of $200,000. Clinicians generally agree that these readmissions are unavoidable. The episode composition graph below shows a graph of the costs of individual episode along with their cost components.

Because of the potential for these high cost episodes, risk tracks A and B generally leave too much of that cost exposed. The high outlier limits for DRG 247 (one of the highest-volume DRGs in this episode family) for tracks A and B are about $68,000 and $40,000 respectively, which leave a $200,000 readmission costing the episode about $95,000 and $72,000 respectively. The high outlier limit for risk track C is significantly lower at about $20,000, which reduces the cost of a $200,000 outlier to $38,000. These much lower cost limits may make track C selection the safest option for participation in PCI because it reduces the episode cost of these high-cost cases. Track C starts at the 75th percentile of the national average episode cost, and episodes costing more than that amount have their costs reduced to 20% of the difference between the limit and the actual episode cost.

The problem with selecting track C is that any cost reductions achieved in the episodes above the outlier limit are also reduced by 80%. Creating a $20,000 reduction in the cost of a $100,000 episode where the outlier limit is $80,000 yields a savings in episode cost of only $4,000 ($20,000 x 20%). Selecting track C limits risk, but it also limits opportunity to reduce costs in the most expensive 25% of episodes.

But look at the episode composition cost graph above and you’ll see that the highest cost 25% of episodes is virtually the only opportunity for cost reductions. The lower-cost 75% of episodes are low because they contain little more than the index admission, in which no Medicare cost reductions are possible.  

This creates a “perfect storm” of difficulties in episode participation:

  • High proportion of index admission cost creates little opportunity to recover CMS discount,
  • Need to select risk track C for protection against high-cost outliers, and therefore
  • Reduction in realization of cost savings in the most complex patients because of the outlier limits.

One additional factor threatens PCI episodes, although we haven’t yet been able to quantify its effect. Medicare Recovery Audit Contractors (RACs) have been active in disallowing one-day stays for PCI. Since those one-day stays existed in the base period, and presuming that they were less costly than the remaining two-day stays, their elimination from the performance period would drive up the cost of the performance period relative to the targets and create systematic losses. Supposedly this change would be mitigated by increases in the target rates as other hospitals performing PCI procedures experienced the same cost increases, but none of this has been validated yet. Several of our client hospital have declined to participate in PCI for this reason, though.

Several policy changes might enhance the attractiveness of PCI episodes. One major change that would fit within the current BPCI structure would be to eliminate readmissions in the costly PRE MDC from BPCI episodes. (Note – CMS has announced that some of these readmission DRGs will be eliminated from all episode families.)  This achieves two purposes for PCI; first, these readmissions are clinically related to PCI and are likely to occur, and they are also extremely costly. If they were eliminated from PCI episodes, participants could select risk tracks A or B and realize a greater proportion of their savings. Other objections would remain, but these changes might be sufficient to attract additional participants to PCI episodes.