How to improve patient outcomes with accurate identity matching

The U.S. House of Representatives recently voted to lift a decades-old ban on developing a national patient identifier, a unique identification number that could be used across healthcare systems nationwide.

It’s a fairly simple concept, but very complex in practice, with deep implications for both privacy and well-being. It is critical to patient safety, public health and program integrity that we all get it right.

While a national system is still a long way off, there are things we should be doing to improve our patient matching and unique identifier capabilities. We must work toward standardizing processes to accurately tie dissimilar sets of data together while maintaining the highest level of privacy safeguards. This will potentially provide a more complete picture of each patient’s health and an improved understanding of public health, while reducing the burden of the cost and time on the healthcare industry to create a patient matching system.


Today, a majority of healthcare systems rely on various combinations of name, date of birth and Social Security number to first confirm the identity of patients across different data sets and then assign identifiers to those combinations that seem to be ‘unique’ or belonging to a single individual. This reliance on personally identifiable information (PII) raises both privacy and accuracy concerns.

While this information may be closely tied to identity, it is in no way unique to an individual. Take, for example, the number of John Smiths that could have a July 31 birthday. Add to that issues with data standardization (even as simple as using a four digit year vs. a two digit year), default or null fields for key identifiers and the frequency with which patients change addresses and use different healthcare facilities, and it becomes clear how complex the challenge really is.

Even so, the industry is expanding the access to this very same data. More patients are using patient portals linked to electronic health records—these records are shared between health providers and research organizations. It’s vital to wrap them into a standardized patient identifier system that allows access without exposure, so the records follow that patient—and only that patient—wherever they go.

The critical component is the ability to accurately link the data together and assign a unique identifier regardless of the source, structure and completeness of the data. The most effective way to do this is not through deterministic matching, but referential and corroborative matching. This means rather than a one-to-one comparison of records, organizations would corroborate identity data by consulting a central database built from multiple contributing sources.

This database would serve as a safe, accurate and continually updated repository. The corroborated data would give patients, hospitals, insurers, physician practices and pharmacies the confidence of seeing the whole picture of the patient—the correct patient.

For example, let’s apply this approach to disease research: an agency, such as National Institutes of Health (NIH), compiles data sets that provide a snapshot of a patient at a given place in time. What if NIH could tie their data to the same patient’s data held by another agency, such as the Centers for Disease Control and Prevention? They would have a better, longitudinal view of the patient’s health to help with the research. This also eliminates duplication across data universes, ensuring that each patient is only counted once in the data sets examined.

In a more down-to-earth example, when a patient goes into an office for care, they frequently re-enter the same information over and over again on forms. One time, they may enter their middle name, another time they may spell out “Road” in their address. These inconsistencies make it more difficult to match data sets across providers, which can result in a less than full picture of patient history and health, impacting the care they receive at any facility. With a patient identifier, all of their critical demographic information and health history is already available without the need to re-capture or re-enter any static information.

An effective approach to implementing improved patient matching is starting small. Pilot programs can help organizations and government health care agencies understand what data they have today and what’s needed to translate it into unique identifiers.

In such a pilot, it’s important to work as a consortium with highly codependent systems to identify the right data to start with as well as the stakeholders, the mission, and the appropriate industry partner who can provide the best referential data sets. Then you can determine the most effective patient matching processes and start to create uniformity between systems.

Better patient matching and unique identifiers can also lead to huge cost and time savings. Supporting today’s complex matching algorithms requires a huge investment in time and resources, yet, often, these systems don’t provide the confidence we should demand from them. Referential and corroborative patient matching processes reduce the complexity of the technology tremendously. Additionally, a lot of time is spent today on manual validation of the ‘fall out records’. What if that expertise could instead be applied to research or patient care, using the comprehensive data provided via the identifiers?

The implementation of effective patient matching and unique identifiers is a next logical step in the evolution of electronic health records, delivering enhanced privacy and security along with improved provider experiences and efficiencies. The data needed to do this already exists and, with the right application of technology, holds great promise to control costs, reduce effort and improve outcomes.