CMS’ first mandatory bundled payment model, the Comprehensive Care for Joint Replacement (CJR) program, launched April 1, and the first data feed from the program was released in August. Now, a second data feed has been released, with critical information on episodes of care that completes the picture from the first data feed.
For the more than 750 hospitals participating in the program, these data feeds serve as benchmarks to track their CJR strategy. While the first CJR data feed served chiefly to indicate how well systems were able to identify attributed patients, an important first step, the second data feed will be more helpful in tracking the post-acute paths for patients.
Reading the data
In analyzing the second data feed, we looked at episodes that began during the first quarter of the program—from April, May, and June of 2016. Those were compared from the first data feed that was cut in July and released in August, to the second data feed that released in October.
It’s important to think about these episodes of care in terms of the months during which they started, because the timing creates significant difference. In terms of episode counts, for episodes that began in April or May, there was no change, or very minimal change. However, for April the total episode spend for episodes that began during that month, went up by 13% from the first data feed to the second data feed.
What the second data feed reveals
There were two major changes that contributed to that 13% increase in average spend per episode:
1. In episodes that went to home health, in the first data feed hospitals may have only had the initial claim, which is called a “request for anticipated payment,” for which CMS makes a payment for half of what’s expected to be the episode spend. So by the second data feed, CMS has made the full home health payment.
2. There was also a significant increase in readmissions for episodes that began in April. In the first data feed, these episodes had a readmission rate of 8%. By the second data feed that rate was 13%.
The nature of CJR is that there is a claims lag, so each data feed reveals a little more of the big picture. Another example is that in the first data feed, we knew we would not see skilled nursing facility (SNF) claims for episodes that began in June. With the second data feed, participants can see percent discharge to SNFs across April, May, and June, so if their strategy was to send fewer patients to SNFs, they wouldn’t have been able to tell from the first data feed, but they can with the second.