Choosing Fracture-Based Targets in BPCI

Submitted by jonpearce on Thu, 2015-08-27 18:24

CMS has made the option available to BPCI participants to have their targets for the Major Joint Replacement of the Lower Extremity episodes stratified by whether the patient had a hip fracture. This issue is of concern to some BPCI participants because episodes involving fractures are considerably more costly than non-fracture episodes. While non-fracture episodes may cost $20-22,000, fracture episodes are typically 45-50% higher at $35-40,000. Since the BPCI targets are based on baseline costs, an increase in the fracture rate from the baseline can cause significant financial disadvantages for the participant under the current target policy. The current targets have the fracture rate from the baseline built into the current targets, so a hospital whose average baseline fracture episode cost was $35,000, non-fracture cost was $20,000 and had a 5% fracture rate would have a combined target of $20,750 (ignoring the CMS discount and other adjustments). This would be the hospital’s current target for all fracture and non-fracture episodes.

Many participants, therefore, are currently attempting to determine which option (selecting the fracture-stratified targets or remaining with the non-fracture targets) is most advantageous. This analysis is typically performed by estimating the fracture-stratified targets (CMS did not provide provider-specific targets) and applying them retroactively to historical data, and then perhaps trending them forward using an assumed future split of fracture and non-fracture episodes. There are two problems with this approach – first that the targets must be estimated since CMS didn’t provide them, and also that the fracture rate for most hospitals varies significantly over time. In our opinion the accuracy of any analyses that must make such estimates has little usefulness because of the variability in the estimates.

Our preference is to review the trends in fracture percentage for any apparent trends, but to also validate the reasons that those trends may be occurring. For example, an apparent increase in the fracture rate may be because an orthopedist who specializes in fracture procedures has recently joined the hospital medical staff and is attracting patients who may otherwise be taken to other hospitals.  If these types of service changes can support the trends noted in the data, then those trends may help in the decision to accept the new targets (if the fracture rate in increasing) or decline it (if the fracture rate is decreasing).

Absent external information about these trends, our preference is to take the lower-risk option, which is to accept the new target pricing. While taking the new prices may cause an “opportunity loss” if fracture rates decrease, most hospitals have few fractures (generally fewer than 10/quarter) so the decrease wouldn’t be highly significant, especially because the difference between the proposed fracture-specific non-fracture target ($20,000 in the example above) and the current non-fracture specific target ($20,750 in the example above) is quite low. However, if the number of fractures increased significantly , perhaps as a result of a disaster, those episodes would receive the significantly higher target ($35,000 in the example above). The target would always adjust to match the risk profile of the patients.

So here’s a case where too much numerical analysis can result in a suboptimal choice, and where careful thought can create a better outcome. In retrospect hospitals may be able to determine if their choice was correct, but hindsight is always 20/20 and isn’t available before this decision needs to be made.