One of my collegues was recently at the CONNECT Code-a-thon held last week. CONNECT is an open-source initiative being developed through the Federal Health Architecture to connect communities through the NHIN using HITSP developed specifications. What follows is a brief summary of his observations. Note that I personally was not present (due to HITSP meetings the same week in Chicago).
He observed that the tools are good, and include a complete open source HIE package, including the Mirth Interface Tool, the Sun/Mural EMPI Solution, NIST's XDS Registry, the GlassFish ESB and more. One of his concerns was that implementation still requires sophisticated domain knowledge and persistence to make them work for an implementation. While the tools are open source and platform independant, they have so many parts (an Enterprise Service Bus, a Database, an application server, a JDK and more) than many developers at the conference spent the entire day trying to figure out how to host it.
He complements the archicture as being very modern. However, much of the implementation still passes through the GlassFish ESB. Much of the documentation speaks of two modes of operation, with and without the ESB. There is also mention of avoiding being tied to any particular component (ESB, application server, database, et cetera). Some redesign efforts in release 2.0 and 2.1 and the upcomming 2.2 around the messaging seem to be targeted around addressing performance issues. Experiences of my own and reports from others with architectures that rely heavily on an ESB leads me to believe that they are encountering similar issues, but that is only a guess. So, it appears that the architecture is still in flux, but that they are also addressing the right issues.
The FHA approach with CONNECT appears to be gathering steam, but significantly sized pilots are still needed. As best we can tell, the pilots with the Social Security Administration, Beth Israel Deaconess and MedVirginia have processed transactions on the order of 1000's. This isn't surprising given the SSA use case is for reporting on disability claims, but I know also from prior experience that a medium sized community hospital can product millions of documents a year. A pilot with the Veterans Administration and Kaiser Permanente may result in more data being exchanged. However, it too only addresses about 1500 patients.
Another concern heard over data exchange with the Federal Government are the federal security and privacy requirements specified in FISMA. FIPS publication 800-53 from NIST describes what is required, and a quick reading of this 188 page document should make it clear that these requirements are much stronger than those for compliance to the HIPAA privacy and security rule, or regulations that are required under ARRA.
FHA, NHIN and the CONNECT open source project are certainly aligned with initiatives in HITSP and IHE. They have developed what seems to be a fairly complex layer addressing Security and Patient Preferences that is much more involved than support for the HITSP TP20 and TP30 specifications or the base standards recommended by the HIT Standards committee. The requirements for exchange of patient preferences are still being developed by ANSI/HITSP in conjunction with NHIN and ONC, and so the work here may not be fully aligned with the outputs expected from HITSP later this year. I expect due to the fact that these groups are all working together now (see I need a new Joke) that future releases of CONNECT will be better aligned.
CONNECT has engaged some very bright people in this effort, including several who have been involved in previous projects such as AHALTA (The DOD's effort similar to the VA VISTA effort), and members of the Federal Health Architecture. Having worked with a few of them, I can tell you that this is a very bright group. Unfortunately, only two EMR vendors were represented at the Code-a-thon, my collegue from GE and another that I know well from Epic. Due to the timing, many people involved in HITSP activities were in Chicago at the HITSP face to face, which certainly could have had an impact on their attendance.
It will certainly be interesting to see how CONNECT evolves. They seem to have made a good start, and there is certainly more to be done.