In 2016, it's not uncommon for individuals to juggle dozens of social media accounts and provide information ranging from email to home address and phone number with many transactions — some even unlock their smartphones with thumbprints. In a climate where individuals so readily link themselves to digital identities in so many ways, it's surprising that hospitals still have a such a difficult time properly identifying patients and matching them to medical records.
But for all of the tech-based innovations that have helped advertisers, banks, schools and other institutions solidly identify their patrons, a single, catch-all method for doing the same in hospitals has remained elusive. In fact, patient matching has only gotten more complicated with improved technology, and it's a wonder medical errors due to hospitals mistakenly identifying patients don't happen more often.
"Patient data is very disparate just by the nature of how it's collected by different organizations," says Beth Haenke Just, CEO of Just Associates and author of Why Patient Matching Is a Challenge, published in AHIMA's Perspectives in Health Information Management journal. "That's what's causing a lot of the dirtiness we're seeing in the data."
When hospitals collect information from patients, they input that data into a master patient index. On a small scale, before the advent of electronic health records, these systems could work with relative efficiency, using name, date of birth, address and Social Security number as primary identifiers for patients within the community. But through the expansion of EHRs, the rapid growth of many health systems, and a lack of organizational and industrial standardization for how to collect patient data, MPIs quickly become muddied and full of overlapping records.
Something like a hyphenated name may be collected in different ways by different individuals whose job it is to intake data within the same organization. If that problem is multiplied across a metropolitan area with many health systems, each of which has networks of clinics and locations, a patient who receives a good deal of medical care could quickly rack up a number of various digital identities across hospitals.
These identities might each contain important medical information that is pertinent for treating clinicians to have in hand, but with no standardization and an incomplete or fractured master patient index, hospitals are left asking whether they have all of the information they need, or in rare instances, if the record staff members are looking at is from an altogether different individual.
It's not uncommon for patients to have similar or identical names or birth dates, so hospitals used to rely on Social Security numbers as a failsafe because they're the only "truly unique" identifier most U.S. citizens have, Ms. Just says.
"But we've seen a radical drop in the collection of Social Security numbers during patient registration," Ms. Just says. "A lot of times nowadays patients don't want to give them, and there's obviously a huge amount of resistance in the healthcare industry to use Social Security numbers as unique patient identifiers because there's too much liability."
Adding telephone numbers to the mix — along with address, name, date of birth, and Social Security number — to create a core of five primary identification tools helps, but doesn't solve the problem, according to Mark LaRow, CEO of McLean, Va.-based Verato, a technology health startup striving to fix the patient identification quandry by combining advanced matching algorithms with a database of reference data.
In addition to "low-level" difficulties like separating proper names from nicknames, first names from last names and sometimes working around the absence of a Social Security number altogether, Mr. LaRow says just getting accurate home addresses and phone numbers is another huge hurdle.
"About 12 percent of the U.S. population moves every year, so within a year, more than 10 percent of the population's identity package is out of date," Mr. LaRow says. "On top of that, the error rate of transcribing a nine-digit Social Security number or a 10-digit phone number is about two percent. That means about one time out of 50, even the person who is providing the number gets it wrong."
Mr. LaRow says the problem doesn't just arise during patient registration or when a hospital has difficulty matching a record to a patient, but can impact reimbursement when a hospital submits a claim to Medicaid or Medicare and has the claim denied to due out of date or incorrect information.
A number of potential solutions to the patient matching issue have come in the form of biometrics technology. Whether palm-vein scanning, iris scanning or another type of physical identification, biometrics require hospitals to sign on with vendors who supply them with the necessary technology at physical entrance points or patient registration points.
"It's not the silver bullet, but the technology does keep getting better," Ms. Just says.
Biometrics aren't infallible, she says, but the risk of a mistaken match using the technology is very, very low. Where biometric technology falls short today is primarily cost and logistics . Some hospitals would have to outfit their physical campuses with hundreds and hundreds of facial recognition or palm scanners to account for all patient intake points. Additionally, there are patients who may be comfortable providing a Social Security number, but very uncomfortable with registering biometric data about themselves.
What's more, Mr. LaRow says, without a unified rollout of such a system across the whole country, let alone large regions or cities, biometric databases quickly break down. A hospital may be able to outfit its network with a vendor's technology, but that doesn't mean other nearby organizations will use the same vendor, or even the same kind of biometric, making the databases incompatible and fracturing the medical records of any patients who do not exclusively seek out care within one organization.
"Unfortunately," Mr. LaRow says, "Those limitations drop you back into the name-address-date of birth-social security number identifiers. And the question is whether there is some way we can make the systems more accurate and reliable in tracking those."
He says even the best state-of-the-art matching systems can only see through about 70 percent of these types of errors and overlaps.
"In a medical setting that's not good enough, that 30 percent could mean life and death," Mr. LaRow says. "It could mean right leg versus left leg in a medical record, or Mark LaRow Sr. versus Mark LaRow Jr."
Verato's solution to the problem is to fight fire with fire in a sense — the company is untangling mistakes in patient data by bringing in even more data. Mr. LaRow says in order to get the most accurate information possible about patients, Verato has created an enormous digital database of addresses, birthdates, names, phone numbers and Social Security numbers with the aim to create patient profiles that aren't bound by health system or geography.
"We're trying to take it one step further by having each patient's entire history of addresses, not just their current residence," he says. "We collect every version of a patient's name to account for common errors or misspellings."
By trying to collect every potential variation on an individual's information, Mr. LaRow says Verato can establish the correct profile that will also account for incorrect data, overlapping medical records or gaping holes in medical records.
While he thinks such a database could be the baseline for making a big dent in the patient matching problem hospitals face, Mr. LaRow says standardization of intake practices is also paramount.
"If we could get the entire medical community to standardize how we store identifying data and capture it, we could get matching up from about 70 percent into the 90th percentile," Mr. LaRow says. "But it's an uphill battle right now because no one is setting those standards or working to get them widely adopted."
The paper on patient matching challenges co-authored by Ms. Just and a number of other health IT experts came to a similar conclusion. According to the authors, while more sophisticated technology like enormous information-crunching databases or biometrics will likely be a part of the solution, standardization is key. With such measures in place, staff could be effectively trained to reduce the amount of duplicate records created.
"Thus, the establishment of policies and procedures (such as standard naming conventions or search routines) for front-end and back-end staff to follow is foundational for the overall data integrity process," the authors wrote. "No amount of advanced technology or increased data capture will completely eliminate human errors."