The first article of this series tried to convince you that community is important, and perhaps even the secret weapon behind open source software. Some open source project leaders understand this better than others, so a range of approaches to community has been developed.
In this article I’ll jump right in on the most critical open source project in health care–the famous VistA electronic health record–while saving some other impressive, although less well known, projects for the final article in the series.
Many open source projects in health IT don’t try to build communities. They feel that they put out useful software and they hope people use it–but they don’t do the work that, for instance, attendees at Community Leadership Summits have put in to make sure they make community members full-fledged partners in their work.
Brady Mathis, a health IT developer, discovered this problem when he became an enthusiastic adopter of the Tolven EHR. He told me that the project leaders seemed to lack a focus on community–a lapse all too easy to observe across many health IT projects. Specifically, he observed little responsiveness on forums, and when his firm offered back code improvements, he found no plan for developer contributions and guarded interest from the project team. However, he remains an enthusiastic support of Tolven, as one can see in a recent article he wrote, and he hopes to help it develop more involvement by its community.
The most famous open source EHR is VistA, and it has been widely adopted around the world (notably in Norway and Jordan) but has not enjoyed the penetration one would expect from such a mature product in the United States. As we saw in my previous article, the state of community around VistA may be implicated.
VistA has one of the most unusual histories of any open source project. As documented in Phillip Longman’s book, Best Care Anywhere, its primeval development was a famously grass-roots efforts by doctors and IT experts in the Veterans Administrations (now the Department of Veterans Affairs). VistA ultimately was accepted by VA management and recognized as a public resource that should be shared. Citing its code as public domain, the VA “threw it over the wall” (a phrase I have heard from VistA supporters) and continued to maintain it internally while having minimal contact with people outside.
A number of projects grew up around VistA, hoping to turn its illustrious success within the VA into an open source miracle in the rest of the globe. And indeed, the true community effort was the WorldVistA project. Several companies also grew up around VistA, two of whom I interviewed for a previous article about open source EHR projects.
All of these projects have survived, but none have broken through to the kind of success that VistA would seem to deserve in the swelling EHR market created by Meaningful Use. There could be many reasons for this inherent in VistA software. But I can’t find a technical reason. A basis in MUMPS, which makes VistA harder to understand, has not stopped companies such as Epic and InterSystems from reaching big adoption. Furthermore, the functions that the VA didn’t see as necessary (such as support for pediatricians) could be added by others.
Roger Maduro of Open Health News told me that licensing was a hurdle to pulling together a VistA community. As mentioned already, VistA itself is in the public domain. The WorldVistA team put their version under the GNU Public License (GPL), which has worked well for Linux and many other free software projects. But other GPL projects use programming languages that allow commercial projects to be built on top of a free software base, but the MUMPS language underlying VistA does not allow that.
The ungainly relationship between the VA and the putative community thus becomes an obvious candidate for improvement. And in 2011, the VA took decisive action in that area.
The VA had observed the success of many open source communities, notably the Apache web server, a project created totally by a committed community. Web servers are some of the most important software in the world (being the means by which people read this article and millions of other sites), and Apache has been the leader in this area for many years.
It so happens that one of the Apache leaders, Brian Behlendorf, also led one of the key open source projects promoted by the US government in health care, the CONNECT project for health information exchange. The VA consulted with Brian and others to develop an audacious plan for creating a healthy open source community out of the disparate stakeholders in VistA. The result in 2011 was the Open Source Electronic Health Record Alliance (OSEHRA).
OSEHRA has learned the lessons of successful community-building from other open source projects and has pursued them doggedly. They solicit input from users as far afield as Jordan and India, major users of VistA software. So far, these foreign collaborators have not returned changes. Culture change is hard, especially across cultures!
In an interview with Seong K. Mun, President and CEO of OSEHRA, I learned that it uses regular summits to develop “two-way conversations.” One success is contributions to a fundamental module called Fileman. The current version (20.2) was developed by a community over two-year period, with up to 20 people participating in discussions. The WorldVistA team reportedly feels sidelined by OSEHRA, but a fresh approach was needed.
In particular, OSHERA knew they had to get rid of the proprietary variants created over time by the companies that market VistA software. They needed one, consummately unified version of VistA across the VA and all outside users. As suggested by my earlier article, they are inspiring vendors to contribute code back to this harmonizing project.
However, when VistA felt it needed to do a major refactoring of VistA, it did not ask the community to step up, but hired a consulting firm. The sense I got from VistA supporters was that this job was too big for the current community community to take on. I suspect that, in particular, it required MUMPS skills the community didn’t have.
It’s hard to decide whether technical upgrades or community upgrades are harder. OSEHRA is dealing with both, and with notable success. My next article will cover some other open source projects dealing with communities.