Patient Identification Challenges

By Susan Chapman, MA, MFA, PGYT for For the Record

Patient misidentification is a costly issue plaguing health care today, affecting organizations and patients alike. The problems can be traced to several factors, among them variables in demographic data, the intake process, and patient portals. Fortunately, there are solutions to reduce the proportion of patient mismatching, with a focus on reducing costs and improving patient care.

The Process
Pew Charitable Trusts (Pew), in its 2018 report on improving patient-matching accuracy, explained how organizations typically match patients to their records, stating, “Patient matching—whether within an individual organization or between facilities—typically occurs through the use of algorithms, unique identifiers, manual review, or a combination of these methods, with a survey of health care chief information officers published in 2012 finding that 42% of respondents rely on two or more strategies.”1

The Impact
AHIMA, in its January 2023 letter to the United States Congress, noted that the “expense of repeated medical care due to duplicate records costs an average of $1,950 per patient inpatient stay, and over $1,700 per emergency department visit. Thirty-five percent of all denied claims result from inaccurate patient identification, costing the average hospital $2.5 million and the US health care system over $6.7 billion annually.”2 The letter went on to cite a survey by the Patient ID Now coalition, which brings together stakeholders to advance a national strategy for patient identification. The study revealed 72% of respondents concurred that inaccurate patient identification resulted in delays in billing and reimbursement. Furthermore, 70% of respondents stated that patients received duplicate or unnecessary tests or services because of the challenges in managing patient identification.

In its white paper, “People Matching in Healthcare: Challenges, Impact, and Solutions,” Harris Data Integrity Solutions noted that despite participating respondent organizations’ having unique patient identifiers, “[a] recent survey from HIMSS and Patient ID Now … found that health care organizations are spending an average of 109.6 hours per week resolving patient identity issues. Over half are spending 21 to 80 hours per week and have an average of 10 full time employees dedicated to patient identity resolution.”3

Megan Pruente, MPH, RHIA, director of professional services for Harris Data Integrity Solutions, explains, “Reworking claims is a timely process. Someone has to review the claim and then find out why the claim was denied. If it was because of an identification error, then the error has to be corrected in the system. There is also the average expense of repeated medical care.” In addition, she says, “If someone just had imaging or labs done at one facility, and they end up having the same test redone, the repeated claim could get denied, too. The patient would be left trying to figure out how to pay for the test or service out of pocket. Thirty-five percent of all denied claims being a result of patient misidentification is a huge amount and is why accurate patient identification continues to be a big concern.”

Pruente acknowledges that even with existing research, it’s challenging to quantify exactly how much errors in patient identification cost because there are so many factors involved in the process from start to finish. “You have the cost of correcting the error in the system, the cost of reworking the claim, the cost of the repeat test or service, the cost to research what happened, time clinical staff spend looking for the correct information—there are so many data points that go into determining the cost of duplicates. More studies and data quantifying the cost of duplicates and/or overlays need to be done so we can continue to advocate for resources to support more accurate patient matching,” she says.

Not only does patient mismatching have a fiscal impact but it also directly affects patient care. AHIMA reported in its 2022 white paper, “Recommended Data Elements for Capture in the Master Patient Index (MPI),” “It is widely recognized that lack of accurate patient identification can affect clinical decision making, treatment, patient outcomes, and patient privacy while resulting in duplicative testing and increased costs. Lack of a standard demographic data set can also lead to patient records not being linked to one another, resulting in incomplete health information being presented at the time the provider is treating the patient.4

The AHIMA report went on to say that inaccurate patient matching and duplication of records disproportionately affect individuals in underserved communities and can increase health disparities—the resulting dissimilarities in health outcomes among different groups of people. According to the publication, “duplicate records for underserved communities are double and triple compared to the average population percentage.”4

The Causes
One of the reasons patient identification has become a growing and increasingly costly issue is that data sets have become larger, which escalates the likelihood of patients having the same or similar names. “There aren’t specific fields in most electronic health records now to capture some of the variances among patients. Additionally, someone could arrive at a health care facility with a legal first name and sex that is different than their preferred name and gender identity. Many electronic health records don’t have preferred name or gender identity fields, so the legal name and sex fields are often changed to support patient care and changed back to get the bill to drop; there is a lack of standards around how to approach data collection of legal vs preferred names and SOGI [sexual orientation/gender identity] data,” Pruente says. “Some EHR vendors are exploring adding discrete fields for preferred name and updating the ‘sex/gender’ field to have values for ‘legal sex,’ ‘sex at birth,’ and ‘gender identity.’ All of these fields can be helpful in accurately identifying the patient. Once we have these fields, vendors can explore adding them to the algorithm logic used to flag possible duplicates. Without these discrete fields being captured, many organizations face bias and culturally insensitive care.”

Another area of concern is what transpires during the intake process. Individuals inputting the information can make mistakes and transpose numbers or incorrectly enter other data. Pruente notes that large health care facilities also have many points of admission and entry-level staff supporting critical data collection. Having registration staff responsible for patient intake without standardized policies and practices often results in inaccurate patient identification. “It’s the same with inputting data fields incorrectly,” Pruente says, adding, “If the patient is homeless, and the registration staff input ‘homeless’ into the address field and the patient has a name that is common, then that individual’s data could commingle with somebody else’s who is also homeless with the same name. Because of those instances, it can be difficult to know which patient is the correct one. It’s critical to have policies and practices to capture some of these unique-use cases.”

The proliferation of patient portals is another factor that can influence the accuracy of patient identification. When patients are able to sign into a portal, create a portal account, or schedule an appointment with a physician by inputting their own data, particularly when inputting information from a mobile device, there’s an increased chance of error. While it’s not common that patients go into their portals and change their names, they can often change other information, like their address, phone, preferred name, gender identity, ethnicity, marital status, and other demographic information. The question is whether any of those changes would affect patient identification and if those fields would be used in the EHR algorithm logic for possible duplicate identification. “Algorithms look for duplicates using demographic data—first, last, and middle name; date of birth; and legal sex—those are the main ones that an algorithm uses to identify possible duplicates,” Pruente explains. “However, many EHRs also use a patient’s address and phone. Phone numbers change a lot less these days, but it still is something that could change from time to time and even more so with an address. A common occurrence is when someone gets married and changes their name or they move after getting married. A lot of demographic data could change in those accounts. There could be a record with a person’s old surname and former address and a new one with that person’s new last name and new address. All of that can complicate the patient-matching process.”

Pruente notes that there’s often no historical record of a patient’s address and phone number in the record, and if there is, it’s not stored in the MPI to be used by algorithms to identify possible duplicates. “A system may not keep historical accounts of addresses and phone numbers, which are critical for patient matching, especially if you have a duplicate/overlay scenario. There is a joint project between AHIMA and ONC [Office of the National Coordinator for Health Information Technology] focused on how to capture address information from a technical standpoint. There’s also an operational-use manual that’s really good. One of the suggestions or things that they were asking for from vendors was to have the ability to have historical address and historical phone number fields because they are valuable in patient matching. When we do clean-up projects, we’re typically using vendors that have that historical information on these data points to help prove more matches,” Pruente says.

Pew pinpoints identity fraud as another factor that contributes to patient misidentification, with a patient using another person’s information to receive care. Pew cited a 2016 study of 555 errors across a five-year interval in a single health care organization that attributed more than 2% of matching errors to fraud.1

In addition, Pew found that current data-capturing processes are not effective for all populations, reporting, “[s]ome patient populations—particularly children or individuals with lower socioeconomic statuses who don’t have certain identifiers or who often move—may have unknown or nonstatic demographic information. Similarly, some patient populations—such as undocumented immigrants—may be reluctant to provide accurate information out of fear of deportation. As a result, the use of demographic data elements for matching
may be less effective in these populations.”1

Potential Solutions
Experts have proposed a range of solutions to help improve patient matching. One possibility is a multistep approach. “There is no one-size-fits-all solution. Patient matching is usually a combination of people, processes, and technology, so using technology in a smart way is the one thing that will help eliminate costs the most,” Pruente says. “Many health care organizations are using auto-merge functionalities, which are great as long as they’re monitored regularly and appropriately to identify possible errors and work to remove scenarios where errors might happen.”

Pruente recommends involving the HIM team, registration, and information technology teams, to regularly review and audit the auto-merge process when it’s in use and attempt to identify where errors might arise. “Algorithms usually identify possible duplicates in many electronic health records. They’re then presented to HIM to review and resolve. Some of the functionality can auto-merge at a set threshold, but there are still patients in the data set that are not the same and are being merged. This can be especially likely with twins or family members who have the same names. Utilizing technology is great, but we have to make sure that we use it with caution and are auditing the work of machines to refine the process,” she says.

Health care organizations also must create policies and procedures for standardization at the beginning of the patient matching process that governs how data are being searched in the system when a patient presents. “We need to research how we’re inputting data when it’s a new record,” Pruente says. “Luckily, we have tools now like the AHIMA naming-convention policy that has really great suggestions on some tricky cases, such as when a patient only has a first name and no middle or last name or baby records, multiple births, safe-haven baby, etc. How do you input information on a patient with a foreign name or one with special characters? That’s where the AHIMA naming-convention policy can be especially helpful in capturing that data correctly.”

Pruente believes it’s important to optimize facilities’ back-end technology tools to resolve duplicates. Additionally, organizations must monitor issues in the MPI to make sure they are keeping pace with technology, culture shifts, and any projected cost savings or goals that are being set. “Based on what the research says about how much money is being lost, organizations need to keep accurate patient identification an ongoing discussion within their facilities. AHIMA authored a white paper that offers advice on how to achieve a one percent or less duplicate rate, and every organization should be working to make this a reality.”

Pew noted that as part of the goals set forth in ONC’s 2015 interoperability roadmap, organizations should have aimed for a less-than-2% duplicate rate within each facility by the end of 2017. The goal is even smaller, one-hundredth of a percent, by 2024. The report continued, “Few data exist, on a national scale, to indicate progress toward that goal, though recent analyses indicate that the industry has fallen short of ONC’s goals. A 2018 survey from Black Book Market Research revealed that an average of 18% of patient records within organizations are duplicates, which is consistent with what some technology vendors have reported.”1

Pruente takes a pragmatic view of organizations’ abilities to reduce their patient-matching errors. “For most organizations, even in countries that have a national strategy for patient matching with a unique identifier for health care, it’s impossible to keep the rate at zero because the data collection is really challenging, and human data-entry errors will always be a challenge. However, the lower we can keep our duplicate rate percentages, the better because the more duplicates you have, the more confusing the data will look. There is a saying that errors are the gift that keeps on giving—the more errors you have, the more you’ll continue to get.”

— Susan Chapman, MA, MFA, PGYT, is a Los Angeles–based freelance writer and editor.