What data can reveal about the drivers of hospital readmissions [podcast]

Short-term readmissions are a frequent and costly contributor to healthcare utilization. Readmissions among Medicare patients cost billions in healthcare dollars, and those costs are expected to increase in the next decade.

But not all readmissions are created equal, and not all patients experience readmission in the same quantity. Among young people, readmissions are largely driven by psychiatric and substance abuse problems; for the elderly, it’s pneumonia, heart failure, and heart attacks. This means that reducing readmissions requires a targeted approach for high-risk populations, rather than a one-size-fits-all solution for all patients.

Data is king when it comes to developing this targeted approach. In this podcast, Nordic Manager of Data & Analytics Tim Grilley and Senior Consultant Lisa Maenner discuss how they have used Qlik data visualization to identify the causes and commonalities of readmissions – and helped healthcare organizations improve around this critical metric.


If you’d prefer to read rather than listen, the podcast transcript is below.


Show Notes
[00:00] Intros
[01:05] Who’s a readmission, and who’s just an admission?
[2:50] Identifying the key drivers of readmissions
[4:18] Short-term and long-term solutions
[5:45] The coolest features in Qlik
[7:17] What a successful data visualization project looks like



Tim Grilley: Hi, I’m Tim Grilley, I’m the Manager of Data & Analytics, and today I’m talking to Lisa Maenner, a consultant who’s been working on a visualization project for a while. Lisa, would you like to introduce yourself?

Lisa Maenner: I’m Lisa Maenner. I’ve been a consultant at Nordic for about four and a half years, in healthcare IT for about eight, and recently I started working more in data and analytics, switching over from implementations and optimization.

Tim: Awesome. Specifically, you’ve been working on a readmission application with Aspirus Hospital, and readmissions nationally are something that everyone is trying to lower. It has to do with the fact that you get lower reimbursement from a lot of health plans and particularly the government, and it also represents care you want to improve. You are sending people home that you could potentially give more care in the hospital, and not have them have to come back and go through the admission again.

But the problem is that readmissions are pretty nebulous. It’s hard to pin a cause and effect to why people are being readmitted, so it’s a good fit for analysis and visualization when there are a lot of different causes. So what types of things were you doing in the analysis to find a solution to that problem?

Lisa: With readmissions, it’s hard to see them in Epic because the patient doesn’t come in as a readmission. They come in as an admission. So first you have to identify which patients are readmissions, and from there you can start looking at what they have in common.

Tim: That’s really cool. You were able to assemble your detailed list of patients, make sure it’s accurate, and then do an analysis or plot things onto visualizations so you could see correlations and data. And if I understand, you were using this in Qlik. How did that work? What types of correlations were you able to find?

Lisa: Once we identified the patient population that were readmissions in Qlik, from there we could start looking at what the patients had in common, and what contributed to or was correlated with their readmissions. First, we looked at the care coordinator, and whether the care coordinator was doing everything they could to prevent the readmission. We also looked at other clinical factors and which factors might relate to this: the patient age, their service, and even what day of the week they came in on. The last thing was looking at things around the patient – their zip code, etc.

Tim: That’s cool stuff. Right now, we’re looking at those analyses and starting to identify the correlations. How does that tool augment the decision-making process once it’s fully implemented?

Lisa: This gives managers all the information they need to go in and start doing their own analysis. It helps them find trends in the data and implement decisions based on that, so they can create new policies and figure out what, clinically, they need to do to prevent these patients from coming back.

Tim: That’s really powerful. Where could this application be integrated into the workflow and decision-making?

Lisa: It’s split into two portions. One is the long-term challenge of reducing readmission rates. That’s things like documenting assessments, making sure the patient has a follow-up visit scheduled within a week of leaving, and trying to prevent these patients from coming back. The other immediate cause is – we gave them the ability to look at patients who are already readmitted, and they can concentrate on those patients because they’re already higher-risk. There are immediate actions they can take without even doing a lot of analysis, to go in and prevent that specific group.

Tim: In that workflow, are they accessing this through Epic to get the patients? Could they access it from a web-based portal? What are they using?

Lisa: Qlik is web-based, and right now they’re accessing it through the web. But it does integrate through Epic as a component of Radar, so it can show up on their dashboard as the first thing they’re looking at when they sign into Epic. It gives them the ability to take immediate action, and it’s there with them throughout the day.

Tim: Inside this app, what were your favorite features, and if you had more time, what would you improve?

Lisa: I think my favorite part of the app is how we structured the data. We tried a few different ways of flagging readmissions, and it gives a lot of flexibility. The standard definition of a readmission is a patient coming back within 30 days, but there are other variations that we might be interested in looking at in the future. And that way, we left it open so we can change it out to look at shorter periods, longer periods, different patient classes – if we want to include observation and emergency patients as well as inpatients. It really opens up the future to look at a greater variety of situations and improve more than just the typical inpatient readmission.

Tim: In that sense, broadening to the quality of care perspective than just the financial repercussion perspective. That’s really cool. What would you do to improve if you had more time?

Lisa: As you use it, it’s about getting [provider] feedback on different factors they want to see, looking at data in different ways, and just expanding. For instance, looking at both volumes and rates and giving them the ability to switch between them, so they have all the data they need.

Tim: So the structure is there, and it’s about fleshing it out a little more. In this project or others you’ve worked on, what do you think the keys to success are?

Lisa: The biggest thing is understanding your client’s need – what they want to see and how they’re going to use the data. That way, it’s formatted in a way that it’s useful. With the visualizations themselves, the most important thing is that they’re meaningful, so everything within the graph means something and tells a story, so someone looking at it knows exactly what the data is trying to communicate.

Tim: It sounds like, for you, having a good operational understanding of the question, so you can tailor it to that – that’s key for you because you’re a strong implementer. You’ve been in healthcare informatics for a while, so your key to success makes a lot of sense.

What does the ideal team look like? What players are key, and what skillsets inside of that team?

Lisa: I think you need a group of subject matter experts that you can get information from on what they want. It’s also about working with them the entire development process, so that they can provide feedback along the way. On the technical side, make sure you have clean data in a format that can be read by Qlik and the Qlik developer to put it all together, work with the different groups, and work with the subject matter experts to make it fit for them.

Tim: How long, inside a visualization project, is it until you can get the first agile review—the first look into the Qlik app that’s meaningful to the SMEs? And then, how long until you’re at training and then until full roll-out for these visualization projects?

Lisa: That varies a lot based on the requirements and the status of the data. Making sure you have clean data is going to make a big difference on how quickly you can get this rolled out. If you do have data that’s there and ready, you can probably pretty quickly start to work with your subject matter experts – maybe within a week. If you need to do a lot of data visualization, that could be pushed out to a month before you could even begin to have something to show them.

Tim: It’s been really great talking to you about these visualization projects. They’ve been really interesting to watch and see, so I’m really glad you shared and talked about some of your successes there.

Lisa: Thanks Tim.


Topics: data and analytics