This was March’s Journal Club article, selected by the members for review. The whole hospital’s administration got excited to review this article. The reason being (and I’m assuming the reason it was investigated) is that CMS is already penalizing acute hospitals for readmissions within 30 days, and intends to bring that into the IRF as well within a year or two. In the mean time, we’ve got to get ourselves together and 1) figure out a way we can track the trends, 2) how to predict the trends and 3) put processes in place to curb the penalties as much as possible.
According to the article, CMS has already laid out #1 for us; however they are using more social/demographic info. The aim of the investigators was to look at if and how physical functioning plays a role in that within the “high-risk for readmission” group of patients. This group of patients is a often over looked group and we have discussed them here before. They’re the ones with non-descript diagnoses, like “general weakness” or “deconditioning.” The term they are using for these people is “hospital acquired deconditioning” HAD– the people that have been stuck in acute for a good while for medical reasons and/or have a lot of comorbidities. These are the red flagged folk.
So the investigators further subdivided this group into 4 groups deemed “Risk Quartiles” from mild risk to ultra high risk for readmission. They created a classification and regression tree (CART– see below) with the 3 highest predictors– Functional Independence Measure (FIM) change rate of 12.5 or less, length of stay (LOS) at rehab of less than 9.5 days, and discharge FIM motor section of 49.5 or less. Using this method, they were able to accurately predict if a patient would be re-admitted to acute with 30 days of discharge with 75% accuracy.
Obviously that is far from perfect. That’s an academic C. But that’s better than going in blind! I do have a couple things to point out. The investigators have a special relationship with CMS, so their N=62,426… which is obviously ginormous, giving the findings a lot of weight. However, as a clinician, I don’t really like that they included patients that actually discharged to acute directly from the IRF. To me, those patient’s FIM scores don’t accurately represent the time, effort and gains that may have been gained and lost or just not gained period due to poor medical status. I’ve always kind of thought of those as “half baked” FIM scores.
So take it, ponder it, do with it what you will. Such as keeping a closer eye on your LOS for these people.. Those non-descript diagnoses don’t tend to be given much time from CMS to begin with, so use it wisely! Track your FIM change rate, so maybe you can better predict what kind of LOS you will need to make that patient not readmit. And really hone in on those lagging motor FIM items!