Wrap-up of the health care IT track at O'Reilly's Open Source convention

The first health care track to be included in an O'Reilly conference covered all three days of sessions at last week's Open Source convention and brought us 22 talks from programmers, doctors, researchers, corporate heads, and health care advocates. We grappled throughout these three days--which included two popular and highly vocal Birds of a Feather gatherings--with the task of opening up health care.

It's not surprising that, given this was an open source conference, the point we heard from speakers and participants over and over again was how critical it is to have open data in health care, and how open source makes open data possible. Like most commercial fields, health care is replete with managers and technologists who don't believe open source software can do the job of powering and empowering busy clinicians in high-risk situations. Some of the speakers spent time challenging that view.

I decided over the course of the week that the health care industry has two traits that make it more conservative than many fields. On the one hand, the level of regulation and certification is mind-boggling. Hardly any technical job can be taken without a particular course of training and a certificate. Privacy regulations--which are interpreted somewhat differently at every clinic--get in the way of almost anyone doing anything new. Software has to be certified too, not something that software firms in most domains are accustomed to. All these controls are in place for good reason, and help you feel safe proffering your arm for a needle or popping the pills each day your doctor told you to take.

Paradoxically, though, the health care field is also resistant to change because the actors in it are so independent. Health care is the most fragmented industry in the country, with 80% of medical practices consisting of one or two physicians.

Doctors don't like to be told what to do. A lot of them are not persuaded that they should supplement their expert opinion with the results of evidence-based medicine and clinical decision support, the big campaigns right now among health care researchers and leaders within the Administration, notably the recent appointee Donald Berwick at the Centers for Medicare and Medicaid Services.

And even medical researchers are hard to gather around one set of standards for data, because each one is looking for new ways to cut and crunch the results and believes his or her approach is special.

So these are the conditions that software developers and vendors have to deal with. Beckoning us forward are the Administration's "meaningful use" criteria, which list the things a health care record system should do to improve health care and cut costs.

Open source definitely needs more commercial champions to bridge the classic gap in packaging and support between the developer community and the not-so-computer-savvy health care teams. We heard from three such companies at the conference: Mirth, vxVistA, and Medsphere.

Of the major projects in electronic health records presented at the conference --VistA, Tolven, and openEMR--two were developed for purposes outside the mainstream U.S. health care industry (VistA for the Veterans Administration and openEMR for developing countries). Although all these projects can point to successful installations in mainstream organizations, they haven't hit the critical mass that makes inherently conservative health care practices feel comfortable adopting them.

But in this specific area of electronic records, I think the proprietary software vendors are equally challenged to show that they can meet the nation's needs. After some thirty years, they have become common only in large hospitals and penetrated only a small number of those small providers I mentioned before. The percentage of health care providers who use electronic health records is between 18 and the low 20's.

Licensing can easily be $15,000 per year per doctor, which small practices just don't have. I won't harp on this, because converting old records costs more than the licenses, and converting your whole workflow and staff behavior is harder still. More disturbing is that a large number of providers who go through the strain of installing electronic records find that they don't produce cost savings or other benefits.

Electronic records have been a success at huge providers like Partners in Massachusetts and Kaiser Permanente in California, but one speaker reported that Kaiser had to spend one billion (yes, that's a "b") dollars to implement the kinds of data exchange and quality control functions specified by the meaningful use criteria.

But we have to look pass the question of who would win the race to digitize the offices of doctors in the U.S.--and around the world--and envision a more open health care system where data can drive high-quality care. I covered the first two days of the health care track in the following posts:

I'll summarize the tracks from day 3 here.

Open source for the things that keep you alive

Karen Sandler, a lawyer from the Software Freedom Law Center, spoke about the hundreds of thousands of devices--pacemakers, insulin delivery devices, defibrillators, and others--that are implanted in people's bodies each year. These devices fail sometimes, and although reports do not classify which failures are caused by software problems, some of them pretty clearly are.

The FDA does not audit software as part of the approval process for devices, although it occasionally requires the manufacturer to show it the software when failures are reported. Devices are also controlled by unencrypted messages over ordinary wireless connections. (The manufacturers avoid encryption in order to spare the device's battery.) In short, software with control over life and death is being installed in millions of people with essentially no regulation.

Sandler's key policy call is to force the source code open for auditing purposes. She also would like to see open hardware and give the patients the right to alter both hardware and software, although these are more remote possibilities. Sandler's talk, based both on careful research and painful personal health experiences, drew a sizeable audience and excited fervent sympathy. The talk was aptly timed just as the SFLC released a report on this issue.

HealthVault and open data on the web

Two brief talks from Microsoft programmers, Vaibhav Bhandari and Teddy Bachour, did a nice job of introducing key standards in the health care field and showing how flexible, carefully designed tools could turn those standards into tools for better patient and doctor control over data.

I felt that standards were underrepresented in our health care track, and scheduled a BOF the night before where we discussed some of the general issues making standards hard to use. Bhandari showed a few of the libraries that Microsoft HealthVault uses to make standards useful ways to store and manipulate health data. Bachour showed the use of Microsoft toolkits, some open source in CodePlex.

As an example of what programmers can do with these libraries and toolkits, the Clinical Documentation Solution Accelerator enhances Microsoft Word enhanced so that, as a doctor enters a report of a patient visit, Word can prompt for certain fields and offer a selection of valid keywords for such fields as diagnoses and medications.

Data mining with open source tools

David Uhlman, who had spoken on Thursday about VistA and his company ClearHealth, ended the health care track with a dazzling tour applying neural network analysis, genetic algorithms, visualization, and other tools to basic questions such as "How many of my patients are likely to miss their visits today?" and common tasks such as viewing multiple lab results together over time.

Every conference has to have a final session, of course, and every final session suffers from decreased attendance. So did Uhlman's scintillating talk, and I left convinced that his presentation deserved more attention because he goes to the heart of our job in health care IT: to take the mounds of new data that electronic records and meaningful use will generate and find answers to everyday problems bedeviling practitioners.

Luckily, Uhlman's talk was videotaped--as were all the others that I reported in my three blogs--and will be put on the Web at some point. Stay tuned, and stay healthy.

Dave [26 July 2010 06:48 PM]

Awesome blog, Andy. One positive result of the health track: less organizational silo-ing. We from CHLA made some very good connections and are already planning meetings with some other presenters!

Faisal Qureshi [26 July 2010 08:09 PM]

Andy, the big takeaway seems to be a better demonstration of vistA. There is so much that newer healthcare applications can learn from. Of some concern, several health OS vendors have created derivatives from VistA but have yet to publish their source repositories. To be open source, means to take on that responsibility back to the community. Educating that responsibility is where O'Reilly shines.

Ed H [27 July 2010 05:14 PM]

There is no question that health care is indeed a very unique field. But you can't change it overnight which is exactly what Obama wants to do.

Out society is different that others and the notion of Universal health care provided and run by the government will not work.

Maybe he should look at things here in Ohio! Ohio Health Insurance rates are among the lowest in the country. I wonder why.

Matt [30 July 2010 09:33 AM]

I have skin in the game as the CTO of a start-up in this space. I agree - open source software can drive open data. Can open source save healthcare? No - not by itself. But it is a critical enabler. My thoughts here: http://bit.ly/b87BwQ