Patient Identification: A Missing Link in Quality & Safety Goals
David Avrin, MD, PhD, Vice Chairman of Informatics and Professor of Clinical Radiology at University of California, San Francisco met with Health Imaging & IT to discuss the implications of the failure to adopt a unique patient identifier.
Fifteen years ago, the Health Insurance Portability and Accountability Act (HIPAA) included a provision for a Unique National Identifier to streamline sharing of patient records across enterprises. Although experts described the identifier as a critical building block of electric health records and essential to achieving quality and safety goals, the privacy lobby persuaded Congress to repeal the provision one year later. Today, as the U.S. pushes toward a National Health Information Network (NHIN) and universal EHRs, lawmakers are eerily quiet on the issue.

How does not having a unique identifier impact patient care, quality, safety and efficiency?  

It’s not possible to achieve quality or safety across institutional boundaries without a unique identifier. This presents a major problem especially for referral centers. According to David Bates, MD, professor of health policy and management at Harvard Medical School in Boston, Partners Healthcare employs a staff of 30 to disambiguate medical information. The problems are identifying and authenticating the patient and granting permission to move information from A to B.

How does the issue impact radiology?

Consider the idea of a national radiation dose registry. It’s impossible without a unique identifier because it is so hard to identify patients across sites of care. Alternatives like proxy identification servers, which cross-reference and identify patients across boundaries, have not developed to the point that they allow robust cross-enterprise data exchange.

The other issue is malpractice. The second most common cause of malpractice errors in radiology relates to failure to have access to relevant priors. Yet it’s cumbersome to move radiology data around partly because we don’t have a universal identifier.

Are there models such as algorithms or statistical matching under consideration that might serve the same purpose as a unique identifier? Are there any working models at the state level?

In a recent seminar, Bates predicted that the National Medical Record Identifier will not happen soon, but he claimed that new technology for proxy identification servers may eliminate the need for a unique identification number. However, I am not aware of any site in the U.S. solely relying on proxy identification servers for patient identification.

I have not seen a good model at the regional level. States have avoided the issue, yet the NHIN has to be developed from multiple regional health information exchanges (HIEs). The Veterans’ Administration uses an identification system that combines the last four digits of the social security number with the first initial of the last name and yields 10 duplicates per number. Ultimately, the VA relies on the social security number.

There are good working models in Europe. England assigns patients a national health identification number, which differs from the government pension number, at birth.

Where do we stand today?

Technically, we’re in the same place as 1997—except electronic banking, which allows account access from around the world is widely employed. Technology for proxy identification servers to identify patients across multiple boundaries has improved, but willing, meaningful sharing of information is lacking; and referral centers still have to employ large staffs to disambiguate medical records.

We need a legislative solution. If a healthcare provider receives funds from the Centers for Medicare and Medicaid Services (CMS), the provider should have to make records accessible to those who are authorized to see them. But there is a lot of political opposition. The November election ushered in an uncooperative Congress that will slow the healthcare bill certainly by financial means and secondarily by privacy concerns.