Using DevOps to Save Healthcare

The following is an abstract for a presentation I hope to give in October to the DevOps Enterprise Summit conference in Las Vegas.

I went to Caltech to become a world class mathematician but decided for family reasons to train as a doctor instead. I spent nights picking the lock and programming the old Burroughs 220 vacuum tube computer, learning years later that its daytime job was to do the math on dark matter in our galaxy. I graduated BS Ch ‘65.

I went to UTMB Galveston for med school and was immediately recruited to do a PhD in physiology. Between drinking too much and spending too much time writing Fortran code to calculate my data, I got an ABD instead – All But Dissertation. By age 40 I was a sober Family Practice doctor with a rapidly growing practice, but I suffered from severe depression and burnout. I went through two expensive and lengthy hospitalizations, sober and suicidal. All this therapy and meetings taught me to understand my patients and their sicknesses better, and I became amazingly good at getting them well. One of my patients was a world-class teamwork trainer. He took on my practice as a pet project, teaching my staff of 10 to talk to each other with respect. I wrote a productivity incentive spreadsheet that effectively made them all financial shareholders in the practice. I made money billing insurance and Medicare with a commercial electronic billing product written in MUMPS code. At the end of 2003 I finally burned out and sold the practice for a nice profit. For years the employees would hold a reunion dinner to thank me for changing their lives.

American doctors die of suicide at the rate of 300-400 a year, or about one a day. Their Electronic Medical Record is a top burnout complaint.  The two hours they spend each evening checking boxes on the EMR and NOT spending time with spouse or children, they refer to as“Pajama time”.  The American Foundation for Suicide Prevention concluded that “drivers of burnout include work load, work inefficiency, lack of autonomy and meaning in work, and work-home conflict.” These are all symptoms of low performing IT operations.

In 2007 I discovered WorldVistA, a nonprofit formed by a core group of VA developers in 2003 to take VistA to the world. VistA is the server side of the VA electronic medical record. Written in MUMPS, it launched at 20 VA hospitals in 1978 and has been up and running since then. It is documented to save lives, and it doesn’t slow productivity. But politics forked the VA version from the Department of Defense version, so that veterans could not transfer their health data from the military to the VA. This is a huge technical debt, whose resolution gets high priority but no significant progress, and now the whole VA-DoD medical record system is being outsourced to Cerner Corp. The new 10 billion dollar effort went live at four test locations and promptly collapsed.

Meanwhile, VistA code is public domain, open source and available for free. At semi-annual VistA Community Meetings I would meet developers and operations people from all around the world. I gave a couple of presentations, did some support work, and started my own two-man company to sell and support VistA to small primary care offices. I ran out of money after two years. I applied for a federal Small Business Investigative Research grant to fund the idea, didn’t get it but learned a lot in the process. Three years ago I learned of a small group of family doctors in West Virginia trying to bring up VistA to replace their commercial EMR. I drove over to say hello and was hired the first day; in a week I was promoted to “Project Director”. I had never in my life made a dollar off of IT. The CEO kept firing people who couldn’t deliver a usable product in 2 weeks. I kept using the skills from my old team-building mentor to get everyone to laugh and work together. In 6 months we had launched VistA in beta, and it was working. Winter came to West Virginia and I was unprepared for 30” snowfalls, so I asked permission to go home and return in the spring. The project fell apart within a month after I left.

It was a good team, and there were many reasons for the sudden demise, but I kept thinking about a WorldVistA talk I gave on the “Clinical Champion”. This is an individual with clinical medical expertise who seems to be necessary for the success of any EMR installation. Perhaps, even though I didn’t really think of myself as an IT leader, I had actually been essential to the project.

Then my former IT employee told me about “The Phoenix Project”. I listened to the Audiobook with my wife, and we both loved it. She would beg me to turn it on in the car – “We have 15 minutes, can’t we listen to the book?” She had been the office manager of my medical practices and had been through all this with me. We both realized we had been doing DevOps all along, and we knew what it could do – but we could not see a way to help with something as big as healthcare in America, which costs us nearly 18% of GNP and is clearly underperforming.

My idea is to be a Project Manager, and not to try to do everything myself. I visualize a three-part structure modelled on what Kaiser Foundation uses. They enroll their patients through a non-profit organization, which in turn negotiates for healthcare from their two for-profit corporations, one for the doctors and one for the hospitals.

My first corporation would be a nonprofit “Foundation for Healthcare Transparency”. Founders of this foundation would be owners of a human body, preferably eligible to vote. We claim to have government of the people, by the people, for the people, but we have outsourced healthcare to our representatives because it seems so complicated. Turns out it is the other way around. American healthcare is made complicated so that voters will turn their rights and duties over to elected officials who will in turn sell themselves to healthcare lobbyists. My motto is “We must be anger-free. We will not win this debate. Instead, we the people will show them how it can be done.”

The second corporation will be the DevOps team. Their corporation will be employee-owned so that they share in their own success. And the third will be the small primary care doctors’ offices, who will be subsidized to own and use VistA, and will be required to maintain one or more IT employees on site.

I plan to start with a citizens’ meeting at my local YMCA, a focus group to assess interest in the process of planning, organizing and monetizing a Foundation for Healthcare Transparency. I will bring those results to the DevOps Enterprise Summit in Las Vegas where I expect attendees to recognize American healthcare as a typical picture of a low-performing business model. I want to assess suggestions for starting an initial DevOps team to install and maintain VistA in one or two small primary care doctors’ offices.