The use of benchmark cost comparisons is common throughout healthcare. Benchmarks are used for comparisons of internal hospital costs across hospitals, for utilization rates in Accountable Care Organizations, and in many other places. The concept of the benchmark is that the hospitals whose costs or utilization are close to or below the benchmark will experience some level of financial success that will not be experienced by hospitals whose costs are above the benchmark. Participants in the Medicare Bundled Payment for Care Improvement (BPCI) and Comprehensive Care for Joint Replacement (CJR) programs often follow this concept. Our clients who participate in these programs frequently ask us for benchmark data, which will presumably be used to achieve greater financial success. These benchmarks are often derived by computing the average cost per episode for components such as skilled nursing facility (SNF), inpatient rehabilitation facility (IRF), readmissions, and similar categories.
Be Careful What You Ask For
The problem with this approach in these programs is benchmarks may have no relationship to the financial success of the underlying hospitals. Financial success in BPCI, and in the early periods of CJR, is related to the change in episode cost from the hospital’s own baseline to the hospital’s performance period, and not to the absolute level of the cost itself. For example, two hospitals may have an identical level of SNF utilization, yet one is achieving a financial surplus and the other is not. This is because one hospital had higher costs during the baseline period and was able to reduce them, while the other hospital was already cost-efficient in this area and was unable to reduce its costs and consequently achieve savings. In these cases, a benchmark has no relationship to financial performance and may be misleading in the conclusions that the hospital would derive from its use.
Below is an example of this issue, showing a comparison of average episode cost for each of the cost categories listed for three groups of CJR-participating hospitals based on their average per-episode savings when compared to their target. If the average cost per episode had a strong relationship to financial performance, the blue bars (highest performers) would generally be lower than the orange bars (mid performers), which would generally be lower than the gray bars (lowest performers). Clearly, this is not the case. A low performing hospital might note that its SNF cost per episode is lower than that of the high performers and could conclude that no further reduction in staff costs was necessary to achieve good financial results; however, that conclusion would be incorrect.
A Better Way
A better way to show the differences in episode cost components is to create graphics that show the disparity in episode costs among hospitals. The graph below shows the average SNF cost per episode for a group of 20 hospitals, with the columns colored according to their financial performance. Note that there is little consistency in coloring across the cost spectrum―some high performers have high costs and some low performers have low costs. However, the overall picture shows a plateau in SNF cost at about $4,000 per episode, and again at $2,200 per episode, suggesting many hospitals have achieved those cost levels. Those plateaus may be reasonable targets for CJR participants to attempt to meet.
A Comprehensive Look
Viewing the episode cost compositions of various hospitals can provide insight into their care management processes and their relative successes (or lack thereof) in managing post-acute costs. Compare the episode composition graphs of the two hospitals below. The first hospital sends almost all patients to home health agency (HHA) care―even those who also go to SNF care―and sends about half of patients to either SNF or inpatient rehabilitation. It also has a significant number of readmissions. This hospital has high costs in almost every category, with significant opportunities for care redesign that leads to cost reductions.
By contrast, the hospital below sends about 75% of patients home after surgery without any post-acute institutional services, and instead uses outpatient physical therapy for most patients. The hospital still has a number of high-cost patients because of SNF and IP rehab costs, as well as a few high-cost readmissions, but it has revised its post-acute care (PAC) protocols to minimize PAC costs. Note: even though this hospital’s utilization of PAC is minimal and could look great compared to benchmarks, the hospital’s financial performance would look poor if its baseline was just as efficient.
This type of analysis is more helpful than looking at the averages of cost components (SNF, IRF, HHA, etc.) individually because there is often significant interaction among these components; for example, HHA costs may increase as SNF costs decrease. Looking at your hospital’s episode cost compositions in comparison to those of other hospitals creates more complete context for evaluating costs and understanding opportunities.
bpci, bundled payment episodes, Bundled Payments for Care Improvement, CJR