Home » Readers Write » Currently Reading:

Readers Write: Why a Unique Patient Identifier is Critical to Improve Patient Matching

March 12, 2014 Readers Write 4 Comments

Why a Unique Patient Identifier is Critical to Improve Patient Matching
By Barry Hieb, MD


In a recent HIStalk article entitled “National Patient Identifier: Why Patient Matching Technology May Be a Better Solution,” Vicki Wheatley argues that, “… healthcare organizations should instead focus on strengthening their existing enterprise matching strategies” rather than work to implement a national patient identifier (NPI). The article makes several valid points that contribute to the ongoing debate about an NPI:

  • No solution, including an NPI, can solve all patient matching problems.
  • Patient matching errors and healthcare fraud will continue to require special attention.
  • Accurate tracking of an individual’s information across healthcare silos is becoming increasingly important.
  • Any proposed patient matching solutions must not negatively influence privacy, security, or clinical outcomes.
  • Accurate patient matching is essential for activities ranging from clinical care to healthcare analytics to population health management.

In these and several other areas, Ms. Wheatley’s article makes a valid contribution to the ongoing debate concerning a national unique patient identifier.

There were a few areas, however, where we have a somewhat different viewpoint. The first of these is the implied assumption that healthcare organizations must make a choice between having an EMPI and having a national patient identifier. We believe that this is a false dichotomy.

Clearly, healthcare organizations must continue to improve their existing EMPI systems as much as possible. However, years of analysis and experience indicate that this will not allow them to achieve the levels of patient matching accuracy that are being required going forward. Those requirements include identification of individuals across disparate healthcare systems, the need for matching against ever-increasing patient populations, and the fact that patient demographic data has known variability and ambiguities.

These represent just three of the reasons why unassisted EMPI demographic matching cannot represent the sole patient matching strategy. Rather, the EMPI approach will need to be supplemented by techniques such as the use of an NPI, biometrics, digital certificates, and other technologies.

Virtually every EMPI system uses a patient’s Social Security number as a data element to improve the performance of their demographic matching algorithm. I was puzzled by the statement, “… even in theory, every single potential patient in the country would need to be assigned one…” as a condition for an NPI to work. Ms. Wheatley acknowledges that there are many people in the US who require healthcare but do not have an SSN. Despite this deficiency, the use of the SSN clearly adds value in those situations where it is accurately available. Similarly, an NPI would benefit each patient who chooses to use one.

An important point to keep in mind is that there is no mechanism to check for data entry errors in most of the data elements currently used for demographic matching. This includes the SSN, names, and addresses. For example, there is no reliable way to detect transposition of digits when a SSN is manually entered. Nor is there an easy way to automate the capture of a patient’s SSN.

Contrast that with a well-designed national patient identifier system. In most situations, the NPI would be read using automated technology such as a barcode reader or a smart chip that would virtually eliminate errors. Even when the NPI is manually entered, embedded check digits can ensure that any data entry errors are immediately detected and the operator is prompted to re-enter the NPI. When added to a person’s demographic profile, the NPI thus becomes the single demographic element that can lead to accurate patient identification on its own. These proposals represent a major advance from the current situation – i.e., an 8 percent or more error rate in EMPI matches.

It is very clear that healthcare organizations will continue their use of EMPI systems for the foreseeable future. That fact, however, should not blind us to the reality that these EMPI systems need to be augmented by additional capabilities going forward if they are going to meet the patient matching accuracy needs that are emerging in healthcare.

The use of a national patient identifier, even if it is initially only chosen by a subset of providers (or patients, on a voluntary basis), will enhance the patient matching accuracy for those patients and help avoid the medical errors that are associated with patient matching errors.

Barry Hieb, MD is chief scientist with Global Patient Identifiers, Inc. of Tucson, AZ.

HIStalk Featured Sponsors


Currently there are "4 comments" on this Article:

  1. As a reader, I appreciate the thought-out, respectful response, but would add that disparate identification solutions (even voluntary ones) seem antithetical to the goal of an interoperable health system.

  2. Interesting observations Dr. Heib. I will point out however, that establishing a NPI opens the door to that identifier being stolen or shared with other individuals – in other words, relying on “something you have” as a unique identifier has already proven to be easily compromised in other industries that have turned to this type of solution. It’s no different in healthcare where fraud and medical identity theft continue to run rampant, one of the primary causes being people sharing their credentials with others (usually family members) in order to receive medical services. I don’t believe a NPI will prevent this nor will it make a significant impact on increasing data integrity and although it is an interesting concept, it’s not truly the way to establish 100% patient identification accuracy in the industry.

    Instead, relying on “what patients are” – e.g. – the use of biometrics for patient identification – makes a much stronger case as a viable patient ID technology that can be used in combination with a NPI. Simply relying on a number that people can exploit, share, or even forget is dangerous and sets an unhealthy precedent as a viable solution. You can’t forget or lose what you are, you always carry it with you, and any identification strategy should definitely include this element.

  3. Having seen how inherently difficult patient matching can be across systems I have to agree with Dr. Heib. A Patient NPI will improve interoperability, medical errors, reduce waste/over-utilization and fraud.

    An NPI with added security measures (Photo ID/Biometrics/PIN) would be a great step forward.

  4. John Trader is correct about both the need for biometrics and the fact that some of the NPIs that are issued will be stolen or misused. There are, however, a couple of important points to be made about his observations. First, we agree that biometrics will be essential going forward in healthcare. But they will be used for authentication, not identity. Fingerprints, iris scans, facial recognition, etc. are all techniques that will help verify that a person is who they claim they are – the authentication process. But, once a person has been authenticated you want to use an NPI to identify them. For example, don’t try to send a person’s fingerprint as the identifying item for a lab test. It is too big, not standardized, etc. Instead send the person’s NPI which is compact, standardized and unambiguous.

    The second point to make is that the NPI can be easily invalidated and replaced if/when needed. This is unlike, a biometric which is yours for life and cannot be replaced. If fraud, identity theft or some sort of error invalidates an NPI you can simply replace it with a new, independent one and restore that person’s ability to carry out ‘normal’ functions. It thus is the combination of an NPI for identification plus appropriate authentication techniques (including biometrics) that permit us to hit the sweet spot of fully meeting healthcare’s patient matching requirements.

    One last observation is that an NPI can also serve healthcare privacy needs in ways that unaided biometrics cannot support.

Founding Sponsors


Platinum Sponsors



















































Gold Sponsors














Reader Comments

  • Your memory isn't right: The judge threw out the trade secret claim. The case was settled out of court, no one won anything. Especially IDX who...
  • HISJunkie: RE:"Epic’s lawsuit accused TCS employees who were contracting with Kaiser Permanente of misrepresenting themselves as ...
  • WHO fan: If the WHO is only feeling mildly petulant, they could simply charge the US for continued access to the ICD, an amount c...
  • Robert D. Lafsky M.D.: Yes all wrong. But some are useful....
  • El. Comandante: Re: Projections, trending and free projection tools. They’ve all been wrong and will continue to be wrong. The u...

Sponsor Quick Links