If you've ever seen an extensive spreadsheet, and in healthcare the odds are you've seen a lot, it's easy to let your eyes glaze over. But, those seemingly endless rows of data are incredibly valuable, despite the format's lack of appeal. Data holds the promise of better understanding of healthcare operations, quality and outcomes.
Ari Robicsek, MD, vice president of clinical analytics and associate CMIO of NorthShore University HealthSystem in Chicago, describes how he and his clinical analytics team of 20 are creating visualizations to better utilize data and engage healthcare stakeholders through Tableau Software embedded in NorthShore's Epic EHR.
Editor's note: Responses have been lightly edited for clarity and concision.
Question: When did NorthShore first begin to explore data analytics?
Dr. Ari Robicsek: A big part of why I joined NorthShore in 2005 was because it was one of the first health systems to be fully paperless. The potential to work with large data sets would be high. Our data warehousing journey began in 2006. We pulled our data from Epic and other sources into an enterprise data warehouse. We were looking to use this architecture to improve care and how we ran our business.
Data visualization and allowing our end-users to explore data was one of the areas I felt we weren't as strong as we could be. We had found ways to integrate data into our EHR and send reports, but that didn't offer us a solution we needed to allow end-users the ability to explore data themselves.
Putting a data element into Epic is different than creating an interactive data set that can be integrated into the workflow. We needed a data analytics and visualization tool for someone other than a report writer. We looked for a platform and settled on Tableau. This has been transformational for us. Two years ago, the conversation would have been around "What kind of spreadsheets can we get?" Now, it's about creating visualizations.
Q: How long did it take to integrate Tableau into NorthShore's EHR?
AR: We have been using Tableau for about a year and a half. Integration into Epic was fairly straightforward. Creating a new program could be done in a matter of days. No one on our team knew how to use Tableau when we first started. We all went for training, and now everyone on the team builds things using the software. Additionally, some people on our team have emerged as Tableau experts.
Q: How is your health system using Tableau software to explore data visualizations?
AR: One of the first dashboards we created with Tableau was focused on length of stay. This dashboard helps track how we are doing on length of stay and drills into specific subsets of the patient population, [including] each specialty and the different hospital departments. We have also been layering a risk adjustment predictive model into the dashboard. This model will predict how long we expect each patient to stay, given his or her problem. We can also compare how physicians are performing compared to the risk-adjusted expectations. This first initiative demonstrated to leadership how useful this tool can be.
Q: How do end-users interact with Tableau?
AR: Tableau has allowed us to create an appealing visualization in Epic. End-users click on a button and are taken to a new frame. The skin is still Epic, but what is on the screen is Tableau. This came out of our quality efforts. Every primary care physician in the system has certain quality goals we expect them to meet. Tableau presents a scorecard of those metrics. It shows which metrics they are on target for and which ones they are falling behind on. Additionally, physicians can hover over a patient's name to see how many gaps there are in care. Does that patient need a colonoscopy or mammogram?
Right from there, physicians can get a specific recommendation about management. For example, we have models that live behind the visualization, which will recommend medication changes. The user can then go into the patient's Epic chart. The software will even tee up a message to the patient that can be sent through the MyChart portal. The integration of Tableau and Epic is seamless. Physicians don't even realize they are using something that isn't Epic.
In addition to the interactive scorecard, physicians can click on a visualization to show how they are trending on each metric over time. A new button will eventually show how physicians compare to the peers in their specialty. You will be able to hover over every point on this plot and see what percentile you are at relative to your peers. This visualization will also plot all of a physician's patients on a map and show where there are community resources for those patients.
Ninety-five percent of our physicians use these visualizations, and most of those who have used it are repeat users. Additionally, those who use it show quality metric improvements.
Q: What are the biggest challenges of working with data on such a large scale?
AR: The biggest challenge is getting the data organized correctly; 95 percent of the work is getting the field, granularity and structure right. From there it is fairly straightforward to create data visualizations.
Q: What future initiatives in this area would you like to see NorthShore implement?
AR: A lot of work we are doing is in the quality and population health space. Right now, we are building out dashboards that help us evaluate the care we are providing. The hope is we will soon have interactive score cards for our primary care physicians and analogous tools for every physician who works in our system. We will be working with orthopedics and general surgery first.