3 of 3 people found this helpful
I've done a lot of work with readmissions, as I'm sure you know it's a massive topic. There are at least 6 different Medicare readmissions measures (PEPPER all cause, AMI, PN, HF, the all-cause risk adjusted measure that is part of Value Based Purchasing, plus the proposed COPD measure). Our system has chosen to use the PEPPER measure because the data is available to every hospital in a standard way. I did lot of work to build a data source, queries, and then calculated fields in Tableau to duplicate the PEPPER measure as best I could and produce a set of reports on that. We use a combination of our PEPPER estimate as well as looking at the larger population.
At the beginning of this stretch of work back in 2011 I did a lot of "deep dives" in the data to try to make meaning out of it, we're small enough that sample size is a big issue: For example, trying to look at readmissions per PCP doesn't work for us on an ongoing basis because they have too few admissions and readmissions in any given month or quarter to be statistically valid, and aggregating to a year doesn't help much for performance improvement, except to look for large outliers, but even then that doesn't necessarily tell us anything useful because PCPs can have such different patient panels.
Also, there's a fundamental issue that hospital quality of care doesn't necessarily have much impact on readmission rates, for example socioeconomic factors were shown to have quite a bit more impact for patients with chronic heart failure in one study last year. More than once we've implemented a change in some area of care and then watched the readmission rate spike for that area, there's a ton of variation in the process that we don't know about and don't have any control over, like when flu season hits.
Where that has led us is that through those deep dives we identified a few areas like the readmission rate from patients discharged to skilled nursing facilities, COPD, other high utilizers, and 0-3 day readmissions and we're working on those areas, plus transitions of care issues that we'd known about (med reconciliation, regular rounding, etc.). These issues mostly call for common-cause solutions instead of special cause solutions, and it's been really hard to tease any progress out in any short-term way. On an aggregate basis our readmission rate is declining, and we're hoping that it's a real change, but it's going to take a year or more to see if the change is systemic.
What we're doing for ongoing reporting is that we've got a line chart showing quarterly and annual readmission rates (with our estimates and the PEPPER actuals that are 6+ months behind), a bar chart showing our monthly variance from target, and some run charts. I also publish a list of system high utilizer patients for our primary care offices and transitions of care nurse, and there's a "kitchen sink" dashboard with a bunch of available dimensions (via a parameter) for our VP to slice and dice the data. We're also including floor/unit specific readmission rates on the floor/unit dashboards, as well as some scores for patient experience transitions of care questions that we capture along with the HCAHPS data.