Aug 25 2013
Congratulations on your next step in public service as a member of Congress representing the 41st district, and your membership on the House Veterans Affairs Committee.
Your district was the site of the first VA/DoD health IT sharing, a system that I helped develop in 1983-5 when I was a Computer Specialist at Loma Linda VA. I worked closely with the committee and Chair Sonny Montgomery’s staff to demonstrate that the DoD could easily adopt the VA software, and we could communicate between Loma Linda and March Air Force Base.
This demonstration was studied by GAO, VA, DoD staff, the Veterans Affairs Committee, and other consultants. Except for the DoD-hired consultant (who later told me that he had been hired “to make the system look bad, but when I saw it, it looked pretty good to me”) Here is 2011 conversation I had with Beth Teeple, who helped make it happen from the Air Force’s side.
The committee noted that DoD had spent $250m (1980 dollars) to develop Initial Operating Capabilities (IOC’s) at a few sites as standalone demonstrations, while VA was spending $82m (1980 dollars) to deploy those capabilities in production across 172 hospitals. None of the IOCs were compatible with each other, whereas the VA system (later to be called VistA) was developed around a sophisticated “active metadata” system with which all systems were able to communicate by virtue of their shared metadata approach. It’s a bit like solving problems with algebra rather than arithmetic. A single algebraic formula can simplify problems that would generate an enormous array of arithmetic efforts. Algebra is a “meta” level way of looking at things.
This sharing effort, by the way, was made possible by the committee’s VA/DoD Sharing legislation championed by Sonny Montgomery. This allowed VA and DoD sites to share resources, and keep the cost savings at their local level, rather than returning the funds to headquarters.
Sonny Montgomery wrote this 1984 letter to Secretary of Defense Casper Weinberger:
Mr. Secretary, I cannot understand the DOD reluctance to try the VA system, which will provide on a timely basis the mandatory system compatibility between the two agencies.
The success of this demonstration (and a parallel one between Fitzsimmons AMC and Denver VA), lead Congress to require that one of the competitors for the DoD’s Composite Health Care System (CHCS) bid a adapted VA system. I left the VA in 1986 to work on the SAIC effort to propose the VA system. We won the CHCS “fly off” competition with a bid about 60% of the competition.
Unfortunately, the DoD dismantled the communications capabilities that would have allowed the graceful evolution of VA/DoD sharing (and the improved coordination of DoD facilities, as well). They also took many steps to make the system incompatible with VA. Whereas VA was thriving based on its “algebra” design ethos, the DoD continued its thrashing about, based on its “arithmetic” level of thinking.
When I first saw the AHLTA architecture, my initial reaction was that it was a “giant single point of failure.” A decade later, while Congress was holding a hearing called “AHLTA is Intolerable,” the system ironically went into a global failover mode; the central node had failed again. AHLTA is a rich source of counterexamples on how not to develop systems, but one of the most significant is its over-centralized single point of failure architecture. NASDAQ has a similar vulnerability: it suspended trading for three hours last week due to a failure at a single point. All European Blackberry’s were locked out of email service for a week a while back, again to a failure at a critical point. These systems were designed for efficiency, not resiliency. The brittleness that was “baked in” to their design also manifests itself in their ability to adapt to changes or surges of activity.
When I hear of a single, integrated electronic health record for the VA and DoD, I see brittleness, not efficiency. I see it devolving to the DoD’s lowest common denominator – based on DoD’s “arithmetic” approach rather than the VA’s “algebraic” model.
The President’s Council of Advisors on Science and Technology (PCAST) issued a report calling for greater use of metadata – the algebraic model, which is a positive step forward to what has been a root cause of success for VistA over the years. Unfortunately, I have seen these recommendations having much effect on the future health IT designs.
I hope that you thrive in you service in Congress, and I hope that you can bring fresh insights to the never-ending problem of VA/DoD sharing. I would be happy to provide any insights that may be helpful to you
Comments Off on Open Letter to Rep. Mark Takano (D-Ca)