Across the healthcare ecosystem, many experts feel that the lack of a nationwide patient matching strategy remains one of the largest unresolved issues in the safe and secure electronic exchange of medical data.
Historically, patient matching has been important within organizations to help identify duplicate medical records. When master patient indexes (MPIs) moved from paper to electronic, organizations gave little thought to data exchange, data formatting, or how data is entered into a person management system, according to an AHIMA (American Health Information Management Association) "perspective" on the topic.
Indeed, as the AHIMA whitepaper continued, “Traditionally, patient matching has been done by health information management professionals who manually review possible duplicate patients and manually update paper and electronic systems as needed. Manual review will not be sustainable in the future because electronic health records (EHRs) have created a vast amount of data that puts an undue budgetary burden on the HIE (health information exchange) to employ additional staff responsible for ensuring data integrity. Currently, organizations are matching patient records within their own systems, but face challenges in incorporating patient matching techniques across care settings and different EHR systems.”
As such, stakeholders are looking for a solution. As John Halamka, M.D., CIO of the Boston-based Beth Israel Deaconess Medical Center put it in a recent Healthcare Informatics podcast, “What if the government helped us create a nationwide patient matching strategy? And whatever that is—it could be a thumb print, facial recognition, a retinal scan, a combination of demographics—the problem today is that we are asked to do all this care coordination but we have no idea who any of our patients are. If you are Maureen Kelly and you’re Irish and live in south Boston, you might as well be Jane Smith. So the government can help us with this.”
A Unique Challenge for HIEs
HIEs specifically face a significant challenge with matching and linking patient identities because of the diversity and independence of the institutions they serve. David Horrocks, president, Chesapeake Regional Information System (CRISP), a regional HIE serving Maryland and the District of Columbia, notes that at the very core of an HIE—as documents and records are pulled from organizations—is assembling them for a particular patient. “So [you want] the ability to say that this lab result that comes from Saint Agnes Hospital [in Baltimore] is for the same person that is currently admitted at the University of Maryland Medical System,” Horrocks explains.
“The matching is utterly foundational to what we are doing. And our challenge is that we don’t control the capturing of the demographic information and we don’t always get the cleanest information coming back from participating organizations,” he continues, noting that addresses are not always updated, misspellings occur, and the ability to match on the demographics that the HIE receives can be limited. “We do probabilistic matching, and we tune our algorithms so that a false positive is exceedingly rare, but in doing so we tolerate a fair number of false negatives,” Horrocks says.
For an end-user of the services—a clinician, for example—CRISP will present a possible match of duplicates. If CRISP is unsure that two records are for the same person, the HIE’s team will show those records to the clinician. “It could be two different people who might be the same and then we rely on the doctor to make a judgment as to whether they should or shouldn’t rely on the information,” Horrocks says.
And if the HIE receives too much poor data, the result of that will be multiple duplicates. “So you might have three or four identities, and for a doctor, this is a very classic case for us, trying to understand if the person right in front of him or her has already has received his or her opioid prescription, for example,” Horrocks says. “You have to go through each of the four [records], open the record, see if there is an opioid and then make a determination to see if it’s the same person. Clinicians hate that.”
As such, CRISP’s leadership team turned to the McLean, Va.-based software company Verato a little over a year ago with the idea to bring “richer” information to the table via the vendor’s cloud-based matching platform. The solution doesn’t provide a different algorithm; it “hydrates” the end-user’s records with additional information, Horrocks explains. So CRISP will send its identities to the platform, which then process and hydrates them with additional information, before sending them back to the HIE nearly immediately. And now that they’re hydrated, they run through the same algorithm as they did before in CRISP’s MPI, and they “match,” whereas they didn’t before. “In our proof of concept we have eliminated hundreds of thousands potential duplicates in our MPI,” Horrocks says.