An HIT interoperability hurdle: whose record is this, anyway?

One of the biggest challenges for linking together different health information systems is the inconsistency in referring to a given individual – be that a physician, a patient, or whomever. One system, for example a lab system, may refer to “Patient x” one way (using their own arbitrary internal patient identifier number), while a different system (for example, a hospital) may use a completely different identifier. Between ambulatory EHRs, each one will also likely refer to a given patient with different, internal methods. Cross-linking these systems so that a unified dashboard can be created that displays all the information from all these systems is made much more difficult as a result.

How can different systems “talk” to each other? It is a more difficult problem than simply creating a standard-format document (either a CCD or a CCR, or some other HL7 file) that systems can export or import. If a data file (in standardized format) is received, matching the patient to whom it belongs is mostly a “best guess” proposition, usually based on patient name, date of birth, and zip code. But human intervention is often needed for confirmation.

For example, when a refill request is sent from a pharmacy via Surescripts to a Surescripts-certified EHR, a “best guess” screen needs to be displayed in the EHR in order to confirm and link that refill request with an internal EHR-specific patient, so that a refill-request response (authorization of the refill, or denial) can be sent back to the pharmacy, and the data can be captured within the patient’s EHR record. Similar manual selection, even with “best guess” prompts, are also needed for laboratory data import, as well as document import into a system.

Web-based Health 2.0 companies could conceivably integrate with various EHR or PHR systems, so that a single sign-on would allow navigation back and forth across both the web application (for example, a diet tracker for diabetic patients) and the linked PHR. This, however, requires the vendors to create a back-end integration that allows single sign-on, and this would be local to only those vendors with a specific business relationship – it is not universal or exportable.

Even with more broadly-scoped health-platform style systems, such as Practice Fusion, where a given patient’s EHR chart can be shared between different, separate practices, and with a linked PHR coming soon, where the patient can see elements of the shared EHR as well as communicate back-and-forth between patients and clinicians – even in this setting, the patient identifiers are internal to the Practice Fusion platform. Granted, the barriers to entry into this system are very low (free EHR and PHR, and very rapid on-boarding of new practices and new users), but there remains the issue of linking to outside EHR systems – a shared Practice Fusion EHR/PHR would not easily, for example, link in a unified dashboard with a Kaiser Health EHR/patient-portal record, or with a Google Health or Microsoft HealthVault PHR record.

As the Office of the National Coordinator for Health IT (ONC) looks to develop standards for health IT interoperability – and an Implementation Workgroup of the HIT Standards Committee met today to review these issues – the way of tying disparate systems together remains a challenge. One proposal being offered has particular merit, referred to as “federated identity.” There are web-based, highly secure, identity-defining services already in place – OpenID is an example of this. This concept, of course, would require the development of processes which link a patient identity within a given system (a particular EHR) to the OpenID. Most of this linking will likely occur from the clinician-side, since the basic problem being solved by federated identity is to facilitate clinician-to-clinician health record sharing via an agreed-upon OpenID that is owned by the patient.

There remain a plethora of issues and ramifications to be worked out – an OpenID way of automatically linking health data systems (without having to go through the manual “best guess” step, described above) needs to be addressed separately from the question of privacy and security, and allowance of who has access to which records. Nevertheless, adoption of an OpenID approach to patient identification will go a long way to overcoming hurdles that have dogged health IT interoperability to date. We will watch the ONC closely, and hope that a workable standard emerges from the current proces.

Robert Rowley, MD – Chief Medical Officer, Practice Fusion, Inc.