HFMA recently published a short article about the data for the Comprehensive Care for Joint Replacement program that was released in August. However, we have some additional caveats that should be considered in using this data.
In looking at this data it’s important to understand how costs are distributed in major joint replacement episodes, and what data was included in the August CMS data release. The three major quantities that can be gleaned from type of bundled payment data are episode counts, episode costs and post-acute utilization. Episode counts are known at the beginning of the episode, so some might assume that this data would include all of the episodes that occurred during the 2016Q2 quarter. However, it appears that CMS cut off the extract of the claims data somewhere in the middle of June, so episodes starting after that date won’t be included. The episode counts for April and May are probably accurate, but June episode counts are significantly understated.
Most hospitals participating in major joint replacement episodes are strongly focused on post-acute utilization and follow metrics that show the first post-acute setting after discharge. Since most post-acute utilization occurs immediately after the hospital discharge, it might be assumed that this data is accurate for April and May episodes, but Kelly Price from DataGen (who’s quoted in this article) found that SNF claims after mid-May weren’t included in the data, even though other types of claims from later dates were included. Therefore, any analyses about post-acute institutional care for May and June would be misleadingly understated.
Finally, none of the episodes from this quarter will have sufficient runout to capture all of the claims. Major joint replacement episodes incur the majority (about 85-90%) of their costs in the first month of the episode, which means that their claims completion rates are faster than other types of episodes. However with SNF claims ending in mid-May, few episodes – even those starting in April - would have a complete set of these claims. Thus the costs of almost all episodes will be understated and aren’t reliable for target comparisons.
Unfortunately we’ll need to wait until the next quarterly update in October before we know what happened in this quarter, and this delay in data usefulness will continue to occur each quarter. Hopefully CMS will eventually see the light and begin to release monthly CJR data, which will greatly alleviate this problem.