Point to point does not imply that one EHR is communicating with one recipient via a specialized interface for that interaction. Requiring a custom interface for every connection between two stakeholders would not be scalable. Point to point simply implies that a transient message is sent from a data source such as a cloud computing EHR hosting center to a data recipient such as an e-prescribing gateway, a healthcare information exchange, or payer.
In Massachusetts we use interface engines, gateways such as NEHEN, and community-based health information exchanges such as EHX created by eClinicalWorks to connect thousands of users in dozens of organizations via transient messages.
There has been debate in the informatics community about using point to point messaging as a means of interoperability. Some suggest that all EHRs should have consistent data elements to foster the most complete interoperability. Although a common information model will be helpful in the future, we need to implement "good enough" standards now to improve quality and efficiency in the short term.
Sending packages of content between organizations using a common web-based transport mechanism enables such high value data exchanges such as e-prescribing, lab data sharing, and administrative workflow.
Point to point interoperability works very well for secure transmission of a content package between two stakeholders. To ensure that HITSP interoperability specifications using point to point approaches are sufficiently complete to test, we need to be very specific about the transport mechanism, as complete as possible listing the vocabularies/code sets, and as constrained as possible describing the package contents. ONC will soon release a Common Data Transport Extension/Gap document which illustrates the kinds of secure transport transactions we'll need to harmonize.
What are the disadvantages of the point to point approach?
a. It does not work for complex scenarios such as an Emergency Department requesting the lifetime clinical record of a person from all the places their data exists in the country. That requires a master patient index, a record locator service, or a national healthcare identifier. In the short term, there are enough high value provider to pharmacy, provider to provider, and provider to payer exchanges that waiting to solve the unique patient identifier problem is not necessary.
b. Auditing the transfer of clinical records between two organizations based on transient messages may be more challenging than exchanging persistent documents with a non-reputiable time/date stamp and signature.
c. Reconstructing a damaged clinical record by replaying transient messages from an interface engine may be harder than simply reassembling persistent documents.
While point to point interoperability uses a transient message from source to destination, a persistent document transfer uses the HL7 Clinical Document Architecture (CDA r2) to transfer an XML document between two stakeholders. That signed document is persisted by the recipient, providing a very clear audit trail about what information was transferred, by whom, and for what purpose. Examples of persistent document exchange include discharge summaries, quality data sets, and population health metrics being sent from one organization to another.
To support meaningful use such as medication workflow, laboratory exchange, clinical summaries for coordination of care, and quality reporting, it's clear to me that we need both point to point interoperability and exchange of persistent documents.
I hope this discussion clarifies the kind of short term exchanges that will accelerate interoperability. This morning I'll be in Washington at the NCVHS meeting testifying about meaningful use. I'll post my testimony on my blog as soon as it is delivered.
My personal opinion is that metrics for meaningful use of point to point interoperability and persistent document exchange may include
* Using a product that incorporates HITSP specifications and is certified by CCHIT using its Laika tool to validate conformance
* Passing an online test with a vendor recognized as compliant with HITSP interoperability specifications such as Surescripts
* Participating in a production health information exchange organization which incorporates HITSP standards such as NEHEN
Over the next few months, the entire healthcare IT community will engage in a very important dialog which will finalize these details.
John Halamka, MD, blogs regularly at Life As a Healthcare CIO.
