The meeting with CCHIT worked. The FOSS community, to the degree that such a thing is possible, had authorized me to go nuclear on the issue before the meeting. I had been given assurance that the community has been so frustrated with dealing with CCHIT that if they did not work with us that if I started an alternative certification program that I would be backed up with the dollars and brains from the community needed to make an alternative certification go.
At this time it appears that such dramatic actions will be unnecessary. Mark Leavitt and Dennis Wilson were willing to consider the profound practical and cultural implications of the ‘rules’ of the FOSS. These implications are difficult enough for FOSS insiders like me to fully grasp that I realized during the meeting that there is still work for me to do make these problems accessible.
CCHIT has recorded the talk and published it here on their website. I have converted the file to an ogg, for those who care about patent issues in audio files. Contact me if you would like a copy. (its too big to host from this server)
So let’s take a 10,000 foot view of FOSS + Health IT + Certification of any kind.
The first thing to understand is that ‘ownership’ of FOSS projects is spread across all of the users and developers of a FOSS system. The true owner of the copyright involved is usually irrelevant and often impossible to calculate. ClearHealth for instance is a high level LAMP (Linux Apache MySQL PHP) application. Besides needing the considerable portions of LAMP, ClearHealth also makes use of tens or hundreds of sub-projects like smarty, phpgacl, scriptalicious, and adodb.
More importantly ClearHealth contains contributions from probably hundreds of people who have contributed bug fixes, clinical templates or modules. In the case of ClearHealth one company, which wisely has chosen the same name as the project, produces 99% of the core. While ClearHealth Inc. produces the vast majority of the code, there are several other companies, (including my own <- shameless plug) that support the same codebase.
It is not really possible to determine in any consistent way who is responsible for a codebase. Often ClearHealth Inc. employees will take code that I and others contribute on the forums and copy into the code repository in such a way that it appears that a ClearHealth developer wrote the code. The contributors do not care and ClearHealth Inc. does not care. My contributions are meaningless outside of what the ClearHealth Inc. team has given to me, and the license requires that my contribution falls under the GPL. There is no way to determine who truly responsible for a codebase, only to make good guesses.
Under the current certification model I could wait for ClearHealth Inc. to figure out how to pass the current CCHIT tests, and then republish the changes to the current ClearHealth codebase required to pass CCHIT. ThenI could apply for CCHIT certification with my friendly fork of ClearHealth. The real cost of doing the certification is the preparation, which is essentially an annual cost (You do not have to do it annually, but your are at a competitive disadvantage if you do not) of about 300k and which will probably be going up.
So I would be getting a certification for about 1/10th the price that ClearHealth pays.
The problem that is that while we collaborate extensively, ClearHealth Inc. and I still compete for customers. If I can offer support for my certified, re-branded version of ClearHealth without participating in the practical price of certification I would be able consistently undercut the support rates of ClearHealth Inc. This represents a disincentive for ClearHealth Inc. to pursue CCHIT certification.
Now consider the OpenEMR project. This project is made up of about 10 major contributors who all share the development duties. There is no single benevolent dictator and there are several companies with developer commit access. Like WorldVista there is a central non-profit that serves as a focal point for community issues for that project. Both of these non-profits will have trouble coming up with 200k a year for continued re-certification and no participating company is large enough to easily take that role.
The lesson here is that in the FOSS community everyone benefits from good code, not just the original developers. If the ‘Tax’ of certification falls to any one party in the community usually it becomes too great a burden for that party.
Practically, it is also impossible to allow a costless download of a CCHIT certified open source EHR. CCHIT requires CPT codes, (which it should not) and CPT codes are owned by the AMA. It is not possible to distribute CPT codes for no cost without violating AMA copyright.
Take away lessons:
- Under the current model it is difficult to have the cost and benefit of the certification evenly distributed.
- There is no way to easily ‘share’ the certification
- There is no maintainable benefit to being the organization that sacrifices to get a certification for a particular FOSS codebase.
- It is not possible to prevent other organizations to certify a system that has already been certified.
- proprietary ontologies, like CPT, are a problem for the distribution of FOSS EHR systems.
Most of these issues were brought up in the meeting, and CCHIT is listening to everyone. I just wanted to put down these issues all in one place for reference. Feel free to comment on this post with other issues that you feel are central to the problem with certifying FOSS EHR projects.