By Howard J. Anderson, Executive Editor
Data mining can be the foundation for meaningful changes in the practice of medicine. Inova Health System has evidence that proves this is far more than just a hypothesis. The Falls Church, Va.-based system, which owns five hospitals, is using the information pinpointed by data mining to help devise new clinical processes. Then it's using its electronic health records system to guide clinicians on how to follow those processes, providing rules and alerts to steer them on the right path.
The result? Serious safety events-those that cause serious harm or even death-declined by 60% from May 2005 to February 2009 at Inova's hospitals. Hospital-acquired infections declined 60% during the same period. And the mortality rate has substantially declined.
Inova is using Web-based data mining software called Quality Manager from Premier Inc., a Charlotte, N.C.-based purchasing alliance. It's a participant in Premier's Quest, a quality improvement benchmarking project. The alliance recently announced that it will expand the project beyond the original 166 hospitals.
"Quality Manager allows us to analyze the opportunities we have to put new processes in place," says Stephen Moore, M.D., Inova's senior vice president, clinical quality and patient safety. "Then we use the EHR to ensure in real time that key processes occur."
The clinical system, from GE Healthcare, Waukesha, Wis., includes treatment protocols that alert nurses to appropriate protocols while they complete documentation of cases, Moore notes.
Premier Inc. acquired Quality Manager from the now defunct Care Science Corp. David Brailer, M.D., former national coordinator for health information technology, led that company when it developed the data mining application in the 1990s.
Inova is using the data mining software to benchmark performance on three severity-adjusted indicators: mortality rates, morbid complications for certain types of patients and average length of stay. It's also measuring quality performance based on costs per case.
Each month, Inova uses a self-developed data warehouse to gather data from multiple information systems, including registration, charge capture, laboratory, claims coding and several others. The delivery system then submits that data in a batch to Premier via a virtual private network. A month later, Innova can perform detailed retrospective analyses of the cleansed data. It uses the Web-based Quality Manager to mine data across 30 characteristics, Moore explains. Some of the data is benchmarked against information from hundreds of other hospitals.
Moore runs reports ranging from big-picture mortality rate trends to such narrowly focused analysis as the units of blood needed for patients treated by doctors who completed certain orthopedic procedures.
Inova spent about nine months preparing to use the data mining software before going live in 2005, carefully determining how to retrieve the necessary data elements and testing the system, Moore says.
Data in Action
Using Quality Manager, Inova determined that its patients with sepsis, pneumonia and kidney failure had higher mortality rates than those at other organizations. "We challenged our nurse leaders to come up with methods to either prevent a patient from getting one of these conditions or detect it faster so we could make a more rapid intervention," Moore says.
Eventually, the hospitals implemented new treatment protocols for these patients that were built into the nursing documentation program. The protocols, for example, remind nurses that when treating elderly patients at high risk for pneumonia, they should make sure to elevate the head of the bed and order a special diet, among other steps. Because congestive heart failure patients are at high risk for developing post-surgical complications, protocols now call for assigning them to a hospitalist to co-manage their care with the surgeon, Moore adds.
The research also led to several new standard order sets that physicians use when prescribing medication and tests. For now, these order sets are paper-based; the hospitals will implement computerized physician order entry in the months ahead.
Making the most of data mining to change the delivery of medicine requires three essential steps, Moore contends. First the data mining application must accommodate "a very broad and deep analysis of your patient outcomes" that goes far beyond measuring mortality rates, he argues. Second, that data must be used to support improvements in clinical processes. "Third, we must take those process changes and be sure that they are occurring more than 90% of the time," he adds.
This requires building the protocols into electronic records that selectively fire off rules and alerts. "You don't want alert fatigue and you don't want to alert clinicians to issues that lack a major impact on quality or cost," he says.
And not all data mining is done retrospectively. For example, in the months ahead, Inova will be rolling out Safety Surveillor from Premier. The software mines laboratory and pharmacy data in real time to identify patients with emerging infections so they can be treated in a timely way.
This article can also be found at HealthDataManagement.com.
Register or login for access to this item and much more
All Information Management content is archived after seven days.
Community members receive:
- All recent and archived articles
- Conference offers and updates
- A full menu of enewsletter options
- Web seminars, white papers, ebooks
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access