Key to Analytics Success: Think Small

Mayo Clinic Health System is well-known for delivering quality care among its 1200 providers who work in Minnesota, Wisconsin and Iowa. The size of the clinic, however, is in marked contrast to its data analytics approach, which revolves around a measured, step-wise approach. Mayo’s Alan Krumholz, M.D., who serves as director of quality outcomes, described Mayo’s program at Health Data Management’s Healthcare Analytics Symposium in Chicago. “You need to focus on actionable opportunities,” he said. “You can’t cure hunger overnight.”

Mayo has deployed an analytics platform, from Humedica, which works in conjunction with its Cerner electronic health records system. Developed in conjunction with the American Medical Group Association, Humedica offers the ability to aggregate clinical data and parse it against various quality metrics. In addition, Mayo can compare its performance and patient demographics against other large medical groups in AMGA also in the Humedica database.

The analytics package is fairly easy to use, Krumholz said. Nevertheless, Mayo only grants access to the technology to a small number of staff who can configure reports. Mayo has settled on some 280 standard reports, which it pushes out to the medical staff in view-only mode.

In one project, Mayo used the analytics platform’s capacity to predict which CHF patients are at risk for readmission, a calculation based on a number of variables such as lab scores and other comorbidities. The data is drawn from the EHR as well as some claims data. It turns out the group of patients in question was relatively small, but it provided the springboard to a quality improvement effort among the physicians responsible for them. Mayo has one person who combs through the aggregate data to generate the list of at-risk patients, then disseminate the list to the local practice sites. That makes more sense—and is more likely to be done—than relying on the sites themselves to comb the data, despite the ease with which it can be done, Krumholz said.

Mayo also restricts how analytics are used. If physicians request a certain quality report, they must document why they need it and how they will apply it, Krumholz said. “We tell them that if we give them the report, we will measure their progress against it,” he said. “That caused report requests to drop 70%.”

Mayo is working with an all-payer claims database in Wisconsin, which combines data from providers, payers, and purchasers. Using the data, Krumholz can analyze how different Mayo care sites and physicians compare in various utilization and cost categories. The data has revealed wide variation in both cost and outcomes. It has also enabled Krumholz to respond to physicians who assert the reason their costs are higher is a sicker patient base. That’s not always the case, he said.  “There is zero correlation between quality and cost.”