Care Transformation Center Blog

Tracking readmissions is good—but not for the reason you might think

by Eric Fontana

I recently met with a hospital executive team that was shocked to learn their hospital received a readmission penalty for FY 2015. They had diligently tracked performance and seen reduced readmission volumes compared to the previous year. Not only that, when they ran their own numbers and compared their organization's data to national averages, the results looked pretty favorable. So what went wrong?

Turns out it was multiple problems, and they’re not alone in their experience. We’ve been having similar conversations with other teams who are also trying to analyze their performance based on their internal data. In short, calculating your own readmission penalties ahead of time is a pretty tall ask for almost everyone except CMS, thanks to several analytical limitations.

Read on to learn what these common mistakes are—and how our tools can give you a quick penalty estimate so you can spend more of your time focusing on your largest opportunities for improvement.

1. Lack of visibility into all sites patients get readmitted to

A significant challenge with capturing a comprehensive set of readmission data is a lack of visibility into all the sites your index admissions are being readmitted to. Our examination of raw CMS claims data from 2010-2013 indicates that other-site readmissions can make up a large chunk of the total readmission pie.

Tool: Find out where your readmitted patients are seeking care

Over 30% of all readmissions occurred at other hospitals—not the same facility the patient first visited. For some conditions, like heart attacks (AMI), the proportion of discharges readmitted to another hospital was as high as 47%. It’s difficult enough trying to forecast readmissions penalties, but almost impossible when you’re playing with an incomplete set of data.

2011-2013 Raw Readmissions Data for Same vs Other Site
Advisory Board analysis of CMS claims level data (standard analytical files)

2011-2013 readmissions data for same site vs other site

2. Not considering risk adjustment

Another common mistake is attempting to extrapolate raw readmission rate performance into a penalty projection without accounting for risk adjustment. Risk adjustment allows CMS to compare readmission rate performance in a more “apples- to-apples” fashion and factor in elements that make readmissions more or less likely. This includes patient age, certain types of diagnoses, and the presence of comorbidities.

Moreover, even though the full risk adjustment methodology is available, you can’t accurately project a penalty without a full national dataset. This shouldn't come as a surprise, as CMS actually told us about this in the FY 2012 Inpatient Final Rule: "...because of the comparative nature inherent to calculating the measures, we note that the statistical models…require data from all applicable hospitals and cannot be replicated using only a single hospital’s data." Further complicating matters, that data typically isn't available until a few months after the readmission penalties are announced.

3. Limited insight into payments, especially for other-site readmissions

Limited financial detail is also a barrier to calculating a readmission penalty. A key mathematical component of the adjustment factor is payments must be associated with readmissions by condition. Again, this data is difficult to get, especially without detail on cases that started as an index admission at your hospital but are readmitted at other sites.

4. Measuring the wrong timeframes

Lastly, many organizations use data from a timeframe that is different from what CMS uses. Common mistakes we’ve seen include only counting readmissions in the most recent year, or using calendar year data instead of matching with the mid-year start and end points that CMS employ.

A step-by-step guide to lowering your readmission rates

Conditions with very low readmission rates, like total hip and knee arthroplasty (THA/TKA), are particularly susceptible to inaccuracy when the wrong timeframe is considered, as a small change in the number of readmissions can have a big impact on readmission-related penalties.

FY 2016 Timeframe Not Finalized but Likely Already Completed

FY 2016 Timeframe Not Finalized but Likely Already Completed

So, is collecting readmissions data a waste of time?

Definitely not. Tracking same-site readmissions is still a useful exercise—just not for projecting readmission penalties ahead of time. Individual cases can provide insight into factors that contribute to unplanned readmissions, such as suboptimal care transitions, gaps in patient education, and post-discharge needs.

Get Advisory Board best practices for each measure

We've mapped our publications and resources to each pay-for-performance program, domain, and measure, such as our preventable readmissions awareness workshop and our dedicated discharge nurse job description.

Browse our readmissions research