Matching patients with the correct—and complete—records of their care is a challenge for healthcare organizations. On average, 18 percent of patient records within organizations are duplicates, according to a 2018 survey conducted by Black Book Market Research.
At one point, Northwell Health, New York State’s largest healthcare provider with 23 hospitals and nearly 700 outpatient facilities, had a list of 220,000 possible duplicate records and was creating about 700 new duplicates a day, says Keely Aarnes, associate vice president of business operations.
An intensive effort—a combination of labor-intensive manual work and the use of new technology—eliminated Northwell’s backlog of potential duplicates and created a process to quickly address newly created duplicates so that another backlog does not build up.
“It took a lot of effort to get that worked down to where it is today,” says Richard Miller, Northwell Health’s executive vice president and chief business strategy officer. “There is an investment that the organization needs to make to address this, and we made that commitment a number of years ago.”
Three groups within Northwell Health—revenue cycle, information technology, and clinical operations—worked together to address the duplicate record problem. “As we brought more and more hospitals and ancillary departments onto our integrated medical record, this became more and more important to all of us,” says Frank Danza, senior vice president-revenue cycle management.
The next step: preventing potential duplicates from being created in the first place.
“Every workflow should begin with patient identity in mind,” Aarnes says. “Especially with the way payer models are changing to promote population health and the way we need to manage our patients, we have to get identity correct. If we don’t get the identity correct, we don’t have all the information.”
How Duplicates Proliferate
Patient matching—the capacity to successfully search each individual’s record in disparate locations and know that it refers to the correct person—is a major barrier to effective health information exchange.
A new report by the Pew Charitable Trusts says the problem of mismatches—patient records that have been incorrectly merged and unlinked records that actually pertain to the same patient—varies widely across institutions. As the healthcare industry consolidates, the problem increases exponentially.
“For an organization like ours that’s always growing, every time we bring a new hospital into the organization and start to integrate the records for that hospital, we immediately develop duplicate patient records,” Danza says.
Consolidation aside, the endless stream of patients—Northwell Health registers from 70,000 to 100,000 patients a month—creates its own opportunity for adding duplicate records. When patients are lined up, registrars often do not have time to carefully verify that they have found a patient’s correct record. The worst thing they can do is accidentally match a patient to the wrong record, so they act with caution.
“When you get a very common name like Smith, Gonzalez, Rodriguez, Miller, you can call that name up in the system and literally get 50 records,” Danza says. “Rather than try to figure out which of those 50 is the right one, sometimes the registrar will create a new case, knowing that there’s a way to reconcile it later on.”
For example, ss of early March, Northwell Health had 28,716 patients with the last name Rodriguez and 22,510 with the last name Smith. That explains why it is so hard for a registrar to find the right person quickly.
“We only surface 50 results at a time on a screen, so you would have to scroll through 100s of pages” to find the right patient,” Aarnes says. “That makes finding a common name almost impossible without other data elements added to the search.”
Beyond that, patient-matching challenges come from many directions, including typos, identity fraud, and lack of standard protocols for data entry.
Of course, patient-matching problems lead to payer claim denials. Black Book estimated that more than 30 percent of denied claims stem from inaccurate patient identification. On average, that cost each U.S. hospital $1.5 million in 2017.
While claims denials might have been a central concern in the past, patient matching has become more important as the healthcare industry has turned its focus to patient experience and quality improvement. For starters, correct patient identification is essential for patient-friendly communications.
“If you think you’ve got two different Millers, but it’s really the same person, you are contacting that same patient multiple times,” Danza says. “That can become very frustrating for the patient.”
Meanwhile, patient ID problems undermine the potential benefits of an integrated healthcare system. If a patient is treated by many providers, each physician should have access to the patient’s full medical record.
“But when you have two medical records for the same patient, the physician won’t have the full picture of what we’ve done for that patient,” Danza says. That can lead to unnecessary duplicate exams and, even worse, diagnostic and treatment errors based on incomplete information.
Duplicate records also make it difficult for health systems to accurately analyze readmission rates and many other operational issues.
“If a patient has two different registrations so that you actually believe they are two different patients, you may not recognize that you have a readmission for a patient,” Miller says.
Similarly, those duplicate records wreak havoc when health systems segment patients for population health management initiatives, which require accurate and complete information about patient demographics, medical history, health status, and other factors.
That’s why Aarnes believes patient matching is becoming more important than ever. “It used to be a cost factor for organizations, but now I think it’s the quality issue related to patient matching that is going to be more meaningful,” she says.
A Multi-Pronged Approach
In 2016, Northwell Health had more than 220,000 potential duplicates and a problem that was getting bigger by the day. Although some staff members were working on patient matching, they were unable to keep up.
“We were creating about 700 duplicates a day and were only matching about 200 a day,” Aarnes says. “We were taking on water, and we were sinking.”
Eliminating the backlog. Most of the duplicates in the backlog were what Aarnes calls “no-brainers” that could be addressed fairly easily. In many cases, the information needed to match two or more duplicates was available in each record.
“If you took a glance, you would be able to say, ‘absolutely these are the same patient,’” she says. “But someone still has to push the button and do the merging process.”
Northwell outsourced that task, an approach that Danza would recommend to others. “If you have a mountain you have to get out from under, there are vendors you can lean on to help you manage that inventory,” he says. “They are not inexpensive, but in the long run, they are invaluable.”
The vendor worked around-the-clock for two reasons. First, having a lot of queries for the patient-matching task was a drain on Northwell’s IT system, slowing response rates for the health system’s daytime staff who needed immediate access. Second, allowing the outsourced workers to work evening and night shifts from their homes made it easier to maintain a workforce for the analytical but tedious work of matching patient records.
Enterprise master patient index (EMPI) technology. While the backlog of duplicates was being addressed, Northwell began using EMPI technology to keep it from building back up. As new records are created, algorithms use demographic data to identify potential matches with existing electronic health records (EHRs).
Two types of algorithms are used: Deterministic algorithms look for data elements—address or birthdate, for example, that match exactly. Probabilistic algorithms look for data elements that indicate a likely match, even though there are differences. For example, a new record for a patient named Mary Ball who lives at 240 Crestwood St. probably matches to the existing record for Mary Ball who lives at 240 Crest Wood.
Northwell works with a vendor on this process, but the health system set its own parameters for determining when new records are automatically merged with existing records. “We did not want to be too aggressive,” Aarnes says. “So, we rolled it out very slowly and watched the results to make sure we weren’t linking records inappropriately.”
This process resolves about 60 percent of the new duplicates that are created each day. “We still create the same number of duplicates, but we went from 700 a day that need to be manually addressed to about 300 a day,” Aarnes says.
The next task is to reduce the number of duplicate records being created. Northwell began piloting a biometric approach—iris and facial recognition—at two outpatient practices this fall. The goal is to introduce the technology to all Northwell Health medical practices beginning in 2019.
The Importance of Patient Matching
Patient-identification-matching problems are proliferating as healthcare provider organizations grow through consolidation. And patient matching is becoming more important as providers need to analyze patient records to improve clinical operations and improve population health management initiatives.
A concerted effort—using manual record matching, a variety of available technologies, and vendor services—can reduce a backlog of duplicates and quickly resolve new duplicate record issues.
See related sidebar: 17 Seconds: Referential Matching Helps Reduce Duplicate Patient Records
Interviewed for this article:
Keely N. Aarnes is assistant vice president of business operations, Northwell Health, Lake Success, N.Y.
Frank Danza is senior vice president-revenue cycle management, Northwell Health, New Hyde Park, N.Y.
Richard Milleris executive vice president and chief business strategy officer, Northwell Health, New Hyde Park, N.Y, and is a member of HFMA’s Metropolitan New York Chapter.