As described in this Singletrack Analytics blog article, the three target adjustment factors (efficiency, patient case mix adjustment (PCMA), and peer adjusted trend (PAT) can vary widely among hospitals in episode types. This article will illustrate those variations and their effect on hospital targets in various episode types.
To perform this analysis we utilized the CMS target worksheets for 73 hospitals for which we are providing analytics services in the BPCI Advanced program. Since these factors are multiplied together to compute the target, we computed a "combined” factor as the product of three individual factors. We then assigned a key performance indicator (KPI) value for each combined factor; combined factor values between .90 and 1.10 were colored orange, higher values were colored red and lower values were colored green. These KPIs are shown in the image below, with each line representing a different hospital and the columns representing episode types.
Winners and losers among episode types
As can be seen, there is significant variation across hospitals in the combined factors for each episode type. For example, in the Cervical Spine episode column 18 hospitals have favorable target factors, 2 have unfavorable factors and the rest are indeterminate. The opposite situation is true for major joint replacement of the lower extremity, in which 15 hospitals have unfavorable targets and only two have favorable targets.
Hospital winners and losers
Similar variations occur across hospitals – the hospital indicated in the red-highlighted row has only one indeterminate episode with all of the other episodes being unfavorable. Similarly, the hospital indicated in the green-highlighted row has 12 episodes being favorable and none being unfavorable. Obviously, the targets provide different levels of opportunity for those two hospitals.
Reasons for variation
The major factor affecting individual hospitals targets appears to be the efficiency factor, which varies across hospitals. A high-efficiency factor, indicating low efficiency, can contribute significantly to a higher target. (Hospitals who have participated in the BPCI program may have lower efficiency factors because of their previous work in reducing costs, although we haven’t found anecdotal evidence to support this conclusion.) Larger efficiency factors will drive the targets higher, reflecting the implicit CMS policy that participants in BPCI Advanced can’t benefit from cost reductions implemented before the beginning of the BPCI Advanced program. The converse is also true – hospitals that were inefficient in the baseline period will have lower targets, allowing their cost reduction efforts during the participation period to create financial success.
On the other hand, the primary factor contributing to low targets across an entire episode type appears to be the PAT factor, in which episode cost trends throughout the baseline period have a significant effect. In major joint replacement, for example, PAT factors were lower than 1 for all hospitals, probably due to the significant reductions in episode cost created by BPCI and CJR participants in that episode type. This appears to reflect a similar implicit CMS policy that episode types that have had significant recent cost reduction in the baseline period by hospitals of the same type as the applicant must achieve a higher level of savings in the participation period to break even.
Implications for applicants
While the primary criterion for success in bundled payment programs remains creating clinical change, the introduction of this target methodology creates an additional criterion for participation for some hospitals in certain episodes. Having a target that’s 10-15% lower than the baseline cost creates an additional barrier to financial success, while having a target that’s higher than baseline creates additional opportunity. In rare cases an opportunistic target appears in a high-volume episode type, which can induce a hospital to participate in that episode even without the possibility of achieving significant clinical change. And while targets shouldn’t be the primary criterion for participation, it’s apparent that they must be evaluated and considered in episode selection.
