While working on my article for the September issue that covers what’s happening now in population health, I had the privilege to profile the many smart folks involved in Query Health (see my podcast with Rich Elmore, former Query Health coordinator), the project to develop standards for distributed population queries founded by the Office of the National Coordinator for Health Information Technology (ONC) in September 2011. The goals of these queries are to deliver insights for local and regional quality improvement and to facilitate performance measures and payment strategies for communities based on aggregated de-identified data. This summer three of the five pilots are kicking off to test specific data package and transfer standards.
In researching this topic, what I found most intellectually stimulating were the complexities involved with the dual languages being tested in many of the five Query Health pilots. For queries where the data responders are insurers the pilots will be testing the HL7-developed Health Quality Measure Format (HQMF) standard that expresses a health quality measure or "eMeasure" in a machine-readable electronic format. Through standardization of a measure's structure, metadata, definitions, and logic, the HQMF attempts to provide quality measure consistency and unambiguous interpretation to support meaningful use reporting.
On the hospital and ambulatory side, i2b2 (Informatics for Integrating Biology and the Bedside), a query language for organizing and transforming person-oriented clinical data to make actionable population health interventions, is being tested.
Jeffrey Brown, Ph.D., Department of Population Medicine at the Harvard Pilgrim Health Care Institute/Harvard Medical School, who is the principal investigator for the FDA Mini-Sentinel Query Health Pilot, said both query languages were instrumental to understanding the health of a population. “There is some inpatient questions, some medical product safety questions that the FDA is interested in, that the claims data can’t answer, but EHR data could,” he said. “So part of the FDA’s mission is to figure out how access to these different data sources can answer some of those questions.”
Some of these questions, like ascertaining the rate of anaphylaxis immediately following a blood transfusion in an inpatient setting, can only be answered by clinical data that has the time of transfusion, as well as other interventions.
Shawn Murphy, M.D., Ph.D., associate director of the Laboratory of Computer Science, Clinical and Research Informatics Division of the Department of Medicine at Massachusetts General Hospital (MGH), and an assistant professor of neurology at Harvard Medical School, (dubbed the godfather of i2b2) has been acting as a resource to integrate i2b2 into Query Health.
Murphy’s team is working on an intermediary query language or a generic way of talking to hospitals to perform population queries. “One of our very big projects at Partners Healthcare in the i2b2 group has been to formulate a transformation from the i2b2 query language, which is specific to i2b2, into HQMF; and then formulate a transformation from HQMF back into i2b2 query language,” he said. “What that would mean that through Query Health you wouldn’t have to have an i2b2 client to talk to your i2b2 hospital site. Your user facing tool could be any tool that knows how to talk in this HQMF language and that would be able to go against either i2b2 or HQMF sites.”
“The main challenge we’ve had so far is all the work so far has been on our own accord. There has not been a lot a funding to our groups to support this,” said Murphy. “We’ve only been able to grab cycles when we could to work on it.”
Later down the road, Murphy said the standards developed in Query Health could help inform the Shared Health Research Information Network (SHRINE) project, another query research project which has been ongoing for 11 years with a data pool of 6 million patients from five Boston hospitals.