May 2014

Patient Identification in an HIE Environment — Where Everyone Doesn’t Necessarily Know Your Name
By Susan Chapman
For The Record
Vol. 26 No. 5 P. 10

If you think maintaining an accurate record of who’s who across one facility is difficult, try doing it across multiple organizations.

With the Affordable Care Act granting more Americans access to health care, the demand for better patient identity technologies continues to climb. The combination of additional patients and the growth of data-sharing health information exchanges (HIEs) may lead to serious patient identification problems, including duplicate records, overlays, and other inaccuracies. Hospitals with lax processes or that fail to employ strong patient-matching algorithms can potentially tarnish data flowing through HIEs. Bad data easily can spread as information travels across different facilities and organizations.

Identity Integrity and Correcting Inaccuracies
When two or more medical record numbers are assigned to one patient, duplicate records are the result, a situation that can lead to serious treatment issues. “The main issue with duplicate records is that it creates records that are incomplete,” says Beth Haenke Just, MBA, RHIA, FHIMA, founder, president, and CEO of Just Associates, which offers health information data integrity solutions. “Not everyone is going to have complete records all the time, but whoever is treating a patient believes he or she has complete records. Therefore, things like current prescriptions and allergies are of major concern to physicians. If that information is not on the record in front of them, then they wouldn’t know those important facts.”

Overlays occur when a patient has different medical record numbers at separate facilities within an organization. When facilities merge and create an enterprise master patient index (EMPI), overlaps are possible. In those events, HIM professionals must choose which medical record number to link with the patient. “We see overlays with merged hospitals or when outpatient facilities come on board,” says Nancy Kadish, MS, RHIA, FAHIMA, vice president and general manager of research services at Care Communications. “We have to make sure a patient is appropriately identified and matched, which can be a challenging process. For example, outpatient facilities may not require a photo ID, a last name could be entered as a first name, or the names may not be spelled correctly. There then may be multiple entries. When staff members try to match that incoming information with that of the existing inpatient facilities, there often can be discrepancies.”

According to Kadish, HIM staff can use several criteria to decide which patient record to choose. Although some facilities identify patients by Social Security number, the increase in identity theft has forced a move away from that process. “Dates of birth can help,” Kadish notes. “Addresses and phone numbers can also help, but they are not the best. Using next of kin, for instance, a mother’s maiden name is something that is easily matched. However, a government-issued photo ID is best.”

While overlaid records can occur during the scheduling or registration process (when the patient is inadvertently registered under another patient’s record), they also can be the result of identity theft or fraud. “In some situations, we face cases of identity theft, but often it tends to be a problem of ambiguous identities,” says Suzanne Layne, RHIT, HIM director at Main Line Health in suburban Philadelphia. “It can sometimes be a stolen insurance card, but it can also be that two people share one insurance card. Sometimes it’s deliberate, and sometimes it’s not. There may be fathers and sons with the same last name, and then at registration the wrong party is selected. Or someone has a name that he uses for one visit and then uses a variation of that name for the next. That type of thing makes keeping records clean very difficult.”

Michele D’Ambrosio, MBA, RHIA, director of health information for New Jersey-based Inspira Health Network, says once an inaccuracy is detected, it can be difficult to deconstruct the record to determine who is who. “One person may have been treated for one issue, then the other party comes in for something else. They are now comingled on one record. The problem then becomes very complex with many layers to it,” she says.

Because HIEs and data repositories access data electronically, if facilities are unaware that two records are comingled or duplicated, the error easily replicates throughout the system.

Preventing inaccuracies is difficult and time-consuming. Short-staffed facilities that must phone patients for verification consume valuable resources better spent elsewhere. Kadish believes registration is the best place to begin correcting duplicates and overlays as well as to prevent future errors. Vicki Wheatley, executive vice president of EMPI solutions at QuadraMed, concurs: “The problem starts with registration and/or scheduling. Hospitals schedule patients, and patients are now able to schedule themselves through patient portals. As we have more electronic records and biomedical devices to collect data, and we transmit that data via HIEs, whatever errors that exist are compounded at the speed of light.”

In the past, back offices manually corrected problems stemming from faulty registration and scheduling. However, today’s facilities rely more on technology to complete these tasks. While HIT makes the process more efficient, human error continues to be a large part of the problem. Consequently, HIM professionals are working to raise awareness among the caregiver community, payers, and patients themselves to ensure that accurate information is provided from the beginning of the health care process.

Judith Gash, RHIA, director of HIM and HIE and privacy officer for CentraState Healthcare System in Monmouth County, New Jersey, says viewing the problem from a national perspective is the best strategy. “Congress has been tabling the universal patient identifier for 15 years, but as HIM professionals, we’ve been trying to move it forward,” she says. “We know that Social Security is a breached system, and there is already an algorithm that would provide better patient matching.”

Barriers to Accurate Patient Matching
Many issues, including incorrect formatting within data fields, data entry errors, smaller organizations’ inability to afford patient matching capabilities, and patient engagement efforts that have not yet sufficiently evolved, create stumbling blocks to acceptable patient identification rates.

In addition, Just notes that differences among EMRs present major problems. “One of the issues widely discussed at ONC’s [Office of the National Coordinator for Health Information Technology] patient matching meeting in December 2013 was the incompatibility of the data’s formatting,” she says. “For example, are the first, middle, and last names all in a single field or are they in separate fields? If the entire name is in one field, the sequence in which the name components are entered will drive how that name field is parsed and electronically transmitted. Data formatting and the standardization as to how some of these data fields are captured is a significant obstacle for creating matching algorithms.”

For example, in some states, physician offices are not required to have an HIM professional on staff. Therefore, in-house decision makers are not always fully aware of processes such as version control or the need to leave both paper and electronic audit trails. “That is why it is so important that staff undergo a solid training process,” Kadish says. “Training generally happens in hospitals but not usually in a doctor’s office or smaller facilities. Consequently, the staff may not fully appreciate what it means when they make seemingly small errors.”

As a consequence, these errors can multiply throughout an HIE. Part of the problem is that smaller organizations cannot afford the best technology or simply lack the knowledge to take full advantage of what HIT can offer. “Still, though you may not have the best tool, if you make data accuracy a priority, you’ll be better off than if you don’t do anything,” Wheatley says. “If facilities don’t have good tools and are also not addressing accuracy issues, the potential is there for serious mistakes. Limitations on the technology are part of the problem but not the only thing.”

At times, one facility may be entering accurate information into the system while other organizations in the HIE are not. For example, a hospital can set and meet a certain standard for accuracy, but other facilities connected through the HIE have less strict requirements. “All organizations connected via the HIE have to have the same standards, but we can’t be sure that they do,” D’Ambrosio says. “Consequently, this can impact our system or their system if we are not on the same standard.”

Even organizations comfortable with their data accuracy must be cognizant that patient information is sometimes beyond their control. “Once [patient identities] are cleaned up, who stores the data? That’s hundreds of thousands of dollars that a hospital can be spending to ensure the information is accurate, but what happens to it after it flows through the HIE?” Gash notes.

Potential Solutions
The most effective way to combat data inaccuracies is to stop them before they happen, says Kadish, who, along with Gash, is a strong supporter of government-issued photo IDs. “I also feel that people need to appreciate the registration process from all sides and not become frustrated or annoyed when they have to produce identification,” she notes. “Facilities also need to have mechanisms in place to do quality checks, produce daily reports, and calculate costs. Money speaks so loudly. If you know that the costs are high due to errors, then that is significant. The cost of fixing just one duplicate can be anywhere from $15 to $60 per error, depending on the average hourly wage. What this can cost the facility is significant, and it’s important that everyone be aware and reminded of it on a regular basis.”

Standardizing key data fields that are used in patient matching can help ensure that patients are correctly identified and records remain accurate. “But for errors that occur in registration and scheduling, a standardized EMR will not be effective unless the software has scheduling and registration modules as part of its suite,” Just says.

D’Ambrosio advocates for some type of across-the-board standard but concedes that HIM professionals have only so much control over how the information is handled once it leaves their respective domains. “We work with data internally, in-house. The next struggle we deal with is the data within the exchange and being sure that they are aware of how that data impacts the exchange. We try to educate them on the importance and how at registration they can get the same patient information,” she says. “Doctors’ offices, rehab centers, nursing homes, those types of facilities generally don’t have HIM professionals and don’t realize the implications of bad data as that information flows downstream.”

Just adds that developing an open-source algorithm that providers and vendors could use to benchmark their current record-matching algorithms could be beneficial. “Even more advanced algorithms are still not created equal,” she says. “I’ve been working with record-matching algorithms for 20 years, and I know how long and expensive it can be. Organizations have to be prepared to invest a few million dollars and a fair amount of time in developing a record-matching algorithm. It’s not something they can turn out in six months. It’s a very iterative process.”

Just says making patients aware of the perils of duplicates and getting them more involved in the process can help matters. “I personally believe that the more we can put this key information at the patients’ fingertips, the better,” she says. “I think it’s beneficial for patient engagement to focus first on frequent fliers, patients with chronic diseases who will be coming in all the time. Trying to target this education to those types of patients to alert them to be careful, mindful when they are checking in and to carefully look at what data are being captured about them are all beneficial approaches. Patient portals will help with that effort if they are designed properly to allow patients to verify addresses and birthdates, for instance, and then allow them to message the institution to make corrections.”

HIM’s Crucial Role
In February, the ONC released the Patient Identification and Matching Final Report, part of an effort to better manage data governance and create policies for measuring data quality. The report centers on how best to maintain accurate patient identification across disparate health care systems whose goal is to share patient information.

The study suggests that standardized patient identification attributes become mandatory, data changes be coordinated across relevant organizations, and EMR technology be certified to uniformly capture patient data. Furthermore, the report calls for the HIM industry to develop best practices to encourage consumers to provide accurate information, disseminate those best practices and policies through detailed training materials, and continue to collaborate with federal agencies to ensure the accuracy of patient information.

“In HIM, we find ourselves cleaning up everybody else’s mess,” Wheatley says. “My challenge to HIM and IT people is to be more proactive for data quality. How do we get out of that mode of correcting information after errors happen? Can we engage our colleagues and vendors to help raise awareness within an organization and find out what can be done to get a handle on accurate patient identification? Not everyone can afford the large-scale data cleanup. But awareness and guidance on best practice can help facilities develop better processes and procedures. And once they have data quality in their own organizations, it’s easier to extend best practice to other entities with which they share data.”

“Obviously, data quality is paramount,” Just adds. “Otherwise, we are simply automating data sharing and moving ‘dirty data’ faster through the system. It can be very hard and expensive to clean up databases. It is far better to do it up front with policies and ongoing training, and that is HIM’s job. We are the information stewards, and we should be doing everything we can to focus on accurate data sharing, developing strong policies, and deploying them.”

— Susan Chapman is a Los Angeles-based writer.