As healthcare becomes more digitalized and networked, issues around data exchange make patient identification an increasing challenge. Without a national patient identifier (NPI), doctors rely on patient’s names and birth dates for identification and something as simple as a middle initial can cast doubt on a patient’s identity, not to mention the myriad identical names stored in EHR systems. Mismatching patients and records can lead to missed diagnoses or treatments.
Patient-matching is also time-consuming. The American Health Information Management Association (AHIMA) conducted a member survey on the topic in 2015 and found that most organizations were spending time weekly doing manual patient matching cleanup. “It’s a significant problem,” says Lesley Kadlec, a director of HIM practice excellence at AHIMA.
Significant indeed. ECRI’s Patient Safety Organization found 7,613 wrong-patient events voluntarily reported by 181 healthcare organizations between January 2013 and July 2015. Those reports likely represent just the tip of the iceberg and show that anyone on the healthcare team can make an identification error.
But barriers to developing an NPI persist. Congress continues to ban federal funding to develop an NPI and tech vendors are reluctant to share their proprietary information for fear of losing market share.
Proponents of an NPI say it would increase patient safety and help with interoperability. In April, 25 healthcare organizations, associations and academies sent a letter to Congress urging lawmakers to overturn a nearly two decades old ban on HHS funding and implementation of an NPI. “As data exchange increases among providers, patient identification and data matching errors will become exponentially more problematic and dangerous,” they wrote. “Precision medicine and disease research will continue to be hindered if records are incomplete or duplicative.”
An NPI could help solve these problems by creating a means to identify individuals from birth throughout their lives, much like a Social Security number.
The trouble with algorithms
Current patient matching relies on algorithms, but these are only as good as the data they use. Discrepancies in how a person’s name is spelled, whether a middle initial is used, birth dates or zip codes can cast doubt on a patient’s identify. “I think we’re at the point where algorithmic matching can only get us so far,” says Rob Tennant, director of healthcare IT policy at the Medical Group Management Association (MGMA). “It can probably be improved, but I think we’re at the stage where without a national patient identifier you’re always going to run the risk of mismatched records."
An NPI would not only increase patient safety but could also help with population health, a key focus of the move to value-based care. “It should help referrals and care coordination, and ultimately one would think that would also save money because if you’re able to move records more efficiently and accurately, you’re probably going to be able to coordinate that care better [and] keep folks out of the emergency room,” Tennant says.
Kadlec says having an NPI is key to having an interoperable healthcare system. “To ensure that patients can get the care they need wherever they are, we have to have a unique patient identifier that allows you to be able to speak across systems,” she tells Healthcare Dive.
ID management a ‘priority’ at ONC
There are signs that Congress’ ban on an NPI may be softening. The 21st Century Cures Act, enacted last year, calls for HHS to study the issue of patient ID management. And the Office of the National Coordinator for Health IT (ONC) has launched a $75,000 challenge to improve patient matching algorithms.
“If we’re going to articulate the two ONC priorities, interoperability and usability, both are affected by the ability to accurately, effectively and efficiently match patient information,” says Andrew Gettinger, CMIO and executive director of the Office of Clinical Quality and Safety for ONC.
Congress barred HHS from working on an NPI amid concerns it would result to a huge national patient database with significant privacy and security issues. That was never the idea, experts say, and today, most privacy advocates believe an NPI would actually enhance privacy because patients’ personal information would not constantly be in circulation.
Gettinger says identification management is a priority at ONC. “We have a number of offices within ONC working on the patient matching problem and have made some progress in trying to standardize what EHR developers do in this space,” he tells Healthcare Dive. ONC’s interoperability roadmap also addresses patient identification and suggests mileposts for accurate individual data matching. For example, it calls for all organizations that match electronic health information to have an internal duplicate record rate of no more than 2% by the end of 2017. The goal is to achieve a 0.01% duplicate record rate by the end of 2024. The office is also highly supportive of private sector efforts in this space.
Getting competitive
Last year, CHIME launched a $1 million global competition—the National Patient ID Challenge — to incentivize innovators to design a private, accurate and secure way to identify patients. Four finalists have emerged from a field of 370 competitors, and testing of their NPI prototypes will begin in July.
“While many organizations have workarounds and algorithms that they use, there’s really no way to completely and with 100% accuracy identify a patient,” says Barbara Sivek, interim vice president of CHIME. Even the best organizations still fail misidentify between 5% and 10% of their patients, she adds.
If a patient is inaccurately identified by an organization, they’re going to carry with them all the wrong records, which can lead to medical mistakes and incorrect info going to payers' systems, Sivek adds.
In search of a better mousetrap
In the meantime, many organizations are working on their own to improve patient matching. “We’re seeing more biometric adoption, more adoption of other types of patient matching technology that can be used within the individual healthcare organization to help them be able to do a better job with identity," says Kadlec. But biometric technology is still evolving and it is costly, so there hasn’t been widespread adoption that could enable an almost unfailing match, she adds.
At Nemours Children’s Health System, clinicians do patient matching to reconcile medications and verify information from outside sources. The process takes longer when the algorithms they use don’t find a hit, says David West, chief of medical informatics at the integrated health system. In the case of medication, that can mean prescribing with less than optimum information. In the case of health information exchange, it means having to call other providers and go through the process of getting records released and faxed to the attending physician.
Nemours tells its vendors when they don’t hit and when it thinks they can improve their results, but the health system typically doesn’t tweak the algorithms itself. “We believe the algorithms they are providing are probably as good as they can be,” West says.
But an NPI would be better. “A national patient identifier would shortcut that need for an algorithm to establish confidence that have linked two identical patients,” he adds. “You would have a much higher frequency of hitting the patients that are the same person at different organizations and not be subject to the vagaries of somebody who stores the name with a middle initial at one organization and without he middle initial at the other organization, creating doubt.”
Getting there won’t be easy, though. Even with a viable technical solution, questions remain regarding who should get an NPI (e.g., only U.S. citizens or anyone who enters a medical facility?) and how to implement it. It could also be a huge undertaking and a huge cost, both for the government to create the database and issue the numbers and for vendors, providers and health plans to incorporate the numbers into their systems.
“One of the reasons why other countries have done it … is because they have a single payer system, so they essentially can derive all decision-making from a central position,” Tennant says. “Ours is a more diverse system and it could take years and years to get to that point. So maybe the solution is just to get a better algorithm and spend our efforts there.”