In my last blog post I shared some of the background on the December ONC Patient Identification and Matching Stakeholder meeting, as well as barriers explored during the meeting. Lively debate erupted around some findings, with a few participants feeling that the whole discussion was going down the wrong path because a unique national healthcare identifier and biometrics were “off the table.”
The majority of the December meeting was spent exploring the initial findings from the environmental scan, literature research and interviews:
- Require standardized patient identifying attributes. Specifically, data elements that are used in patient matching to underpin data exchange should be standardized. This is especially evident in the HL7 Patient Identification (PID) segments, as vendors and their provider clients don’t use the same data elements, nor do they capture them in a uniform fashion. Standard attributes might include first, last and middle names, suffixes, date of birth, current address, historical address, phone and gender. Using a standard format for addresses, such as USPS, was also explored.
- Introduce certification criteria that requires certified EHR technology to capture and use the standardized attributes. Certification is the obvious lever that the Federal government can use to implement standardization. However, this could take years to work through the Federal processes. Clearly, there will be no quick fix.
- Study the use of additional attributes to improve patient matching. Additional attributes like mother’s maiden name, driver’s license number and passport number might enhance matching, but statistical analysis must be done to define their true value.
- Develop or support an open source algorithm for patient matching. I have to admit, this one really had me scratching my head. There are multiple open source matching algorithms today and one of the best ones is provided by the CDC, so I’m not sure why the US Government would fund, or actually develop, more open source software.
- Introduce certification criteria that would require certified technology produce reports detailing potential duplicate records. While in theory this appears to make sense, I have some concerns about how this would actually be applied. The matching technology can be applied at the point of registration when a patient search is undertaken (active mode) or after transactions are executed in what is commonly known as a passive mode. The technology can also come within an EHR, standalone, or even applied in batch mode. Thus, applying certification to the various integration points, approaches and products that have nothing to do with the certified EHR would be very difficult.
- Consider a more formal structure to establish best practices for matching and data governance. I say a big "hurrah!" to this finding. Best practices that are based upon facts, knowledge, research and endorsed across the industry are sorely needed.
- Develop best practices and policies to encourage consumers to keep information accurate, and to actively participate in this practice. Consumer engagement was generally welcomed during the discussions, but, consumer engagement will be a journey that addresses more than the data attributes that underpin patient matching and data exchange.
- Work with industry groups and associations to develop and disseminate training material supporting best practices. I believe industry groups like NAHAM, HIMSS and AHIMA would welcome the opportunity to promote best practices, and they already do so in many areas.
I’m looking forward to reading the recommendations that come from the preliminary work, and I’m sure we will see lots of discussion around the topic at the HIMSS 2014 event in Orlando. I’ll be making visits to the various interoperability showcases and vendors to see patient matching in action at the event, so give me a shout and join me for a discussion on any of these fascinating topics at IBM's booth 1651.