A hidden menace lurking in the technical world of information exchange is correctly identifying patients and matching them properly to their records. When matching goes wrong, two things can happen, neither of them good: Either there are two or more records for the same person, which can make it harder to access vital information, such as a drug allergy; or the information for one person gets attributed to the record of someone else with a similar name, raising the potential for medical errors.
While it might seem simple enough to figure out who is who, there are several reasons why it’s not:
- Many people with the same name and birthdate live in the same geographic area. In Houston, the Harris County Hospital District found 2,488 patients named Maria Garcia, and 231 of them share the same birthdate.
- Errors in typing in names and numbers are fairly common—think about how many different people in a hospital add information to patient records, and under what chaotic conditions.
- There’s a lack of standardized data fields in information systems. For example, birthdates can be listed by month, day, and year or by day, month, and year.
Health care systems have whole teams of administrative staff who track down and fix duplicate records and other confusion. Still, individual systems are likely to have error rates from 7% up to 40%. When you combine a number of hospitals’ databases, all of them with some patient ID errors, and mix them up in an HIE, the potential for chaos is multiplied (an estimated 50% to 60% error rate), and a little frightening.
It’s certainly scary to the people who run data systems for hospitals, and their professional societies have been writing white papers about the problem for years with little notice. This year, the issue started to get more traction and has been talked about in Congress as part of the interoperability discussion.
A million-dollar prize
What they’d really like to see is every American receiving a single, preassigned identification number to be used in all their medical and health insurance encounters, something that was actually called for in the original 1996 HIPAA law. But federal agencies have been prohibited from even studying this possibility since 1999, when Congress agreed to language inserted by Ron Paul, then a Republican congressman from Texas, who has strong libertarian views like his son, Rand Paul, the Kentucky senator and Republican presidential candidate.
Because of the political sensitivity of the issue, it’s unlikely that a mandatory system of patient identifiers is going to be part of legislation, according to the latest talk in the capital. Many insiders say they hope that “someone” will put together a voluntary personal identifier that patients could adopt themselves. “Even the best patient matching we have has errors,” says Aaron Seib, CEO of the National Association for Trusted Exchange. “It’s expensive to manage it this way and it involves patient safety risks. You should have the right to have a voluntary identifier so there is no fuzzy matching of your medical information.”
High-tech solutions for reliably identifying patients depend on biometric technologies such as retina or fingerprint scans, which a few hospitals are already using to foil medical identity theft. The College of Health Information Management Executives is offering a million dollars to whoever can come up with a “universal solution” to the patient identification problem.
Part of the problem is being tackled by a group of provider and health plan representatives brought together by a health IT organization called WEDI. The Virtual Clipboard project is developing some kind of digitized data concept, possibly using a QR code, that includes the patient’s demographic and insurance information and maybe current medications, and could easily be swiped to get a quick eligibility verification. “It would remove a level of frustration from the registration process,” says Rob Tennant, health information technology policy director for the Medical Group Management Association. “There’s really no downside as long as it’s voluntary and we’re all in agreement.”
Health plans have stayed out of the patient ID mess for the most part, as they have managed claims by using their own assigned membership numbers. As they join with providers to swap data, though, they’ll have to ensure they are applying the right clinical records to their members. “It is exponentially important that we don’t guess; because of privacy laws, being wrong about something like that is potentially very expensive for a health plan,” says Availity’s Mark Martin. This is accomplished in part by specialized software that uses algorithms to predict whether a match is correct based on whatever demographic fields are available. Humans have to manually review those that don’t match.
Cal INDEX, the HIE set up by Anthem Blue Cross and Blue Shield of California, says its software has managed to make successful matches about 85% of the time. For the rest of the questionable records, staff members have to manually reconcile them, which is expensive. “It’s one the more difficult parts of HIE and will be until we can uniquely identify people,” says Cal INDEX CEO Dave Watson.