In a June 15 slide presentation that is available online, Baker noted that after reviewing more than 280 IT programs at the VA, many "exhibited signs of trouble," including some that were more than 13 months behind schedule or 50% over an initial cost estimate. Baker also noted a diminution in software quality between releases and "inadequate skills" available to complete some projects, adding that "substantial change is required."
In the slide show, Baker outlined his new “program-management accountability system,” or “PMAS,” under which software development "will be managed rigorously to schedule" with frequent, documented delivery milestones occurring "at most every six months." Under the system, any program or project will be halted on its third missed customer delivery milestone, and once halted, a number of prescribed steps must occur before the project can be restarted. Those include reassigning project staff, a review of service contracts and program design and a reassessment of outsourcing or in-house development decisions.
"I wouldn't want to say I created any of this," Baker said. "There are a number of organizations that go by schedule-driven development. Microsoft is notorious, in a good way, around that. They can tell you exactly what the dates are for release. They can't tell you what the functionality is, but they can tell you when they will release. Schedule becomes paramount because it is the easiest thing to measure to."
"In the private sector, if I want to fire a vendor and hire somebody else, I fire the vendor and hire somebody else," Baker said. "The state of Virginia has a right-to-work law and you can let anybody go with two weeks' notice for any reason, so discipline is easy. In government, I've found because those things are so onerous, everyone shies away from making hard decisions."
The VA runs the largest integrated delivery network in the U.S., with 153 hospitals and more than 700 outpatient clinics, all connected by one of the most advanced and interoperable electronic health-record systems in the world, the Veterans Health Information Systems and Technology Architecture, known as VistA.
Achieving interoperability between the medical records systems of the VA and defense healthcare organizations has been a goal dating back to the Clinton administration. According to testimony at a recent hearing before a congressional committee, however, there is some doubt whether the two departments will make the Sept. 30 deadline for interoperability set in a 2008 defense department authorization law.
Baker said he's confident the goal will be met, however, success will hinge on the definition of "interoperability," which he says is "a word big enough to drive three trucks through."
Several development projects on the list of 45 up for review are components of HealtheVet, the overarching VistA replacement project that according to the Government Accountability Office last summer, was already six years behind schedule. The long delay prompted the obvious question: Should HealtheVet be reviewed and, possibly, scrapped?
"It's a fair question," Baker said. "I just haven't had the time yet to dig into the whole HealtheVet project. It's certainly something I've got to do."
HealtheVet, Baker said, is, based on size alone, "a tremendous project and it's going to take a couple of months, once I've focused on it, to be able to answer that question. We have the best electronic health system out there in VistA, and we have a project to replace it in HealtheVet. I think we have to be sure that if we spend $10 billion on a replacement, what we get will be better than what we have. If it's not going to be better than VistA, then why do it?"
One thing Baker said he'll look at is how the HealtheVet vendors are collaborating with the ultimate end-users. For example, he said, "Is the interface development (of HealtheVet), of having doctors involved like they were with VistA?
"What makes VistA great is, when you go to a VA hospital, everybody there is doing everything electronically. Finding any paper at a VA hospital is difficult," he said, because virtually all operations are performed by clinicians on the EHR. "They're doing it because it makes it easier to do it electronically than on paper. All of those things were in there because physicians were in the loop when they designed it and ensured they were in there."
In an earlier interview about the Class III memo, Baker said he and the VA are facing a balancing act between change and control.
"How do you create an environment where the innovators have the ability to innovate, and where you can still release a certified medical system with confidence at all of the hospitals?" Baker asked, rhetorically. "In centralizing IT we've decreased the amount of innovation at the local level."
Meanwhile, Baker said, in speaking with veteran VA programmers and viewing their handiwork, he recognizes "something that was very tangible was created out of the way they did things."
"The most important thing is," Baker said, "we absolutely have to bring innovation back to this. I am absolutely, 100% committed to addressing that issue. I don't know how I'm going to do it yet, but I can tell you I have complete secretarial support to address” it.
Many of the early programmers who developed the VistA system remain actively involved with developing the software, even though some no longer work at the VA.
One of them is Frederick D.S. "Rick" Marshall, who in June 1984, at age 18, started working with computers at the VA hospital in Walla Walla, Wash., a week after graduating from high school, launching a 25-year career with the VistA software. One of his earliest jobs was to crowbar the hospital's first mainframe computer from its shipping crate.
Marshall worked in IT at the VA for 19 years, including seven years of local computer support and programming, and 12 years of programming at the national level. He left the VA in 2003 to co-found WorldVistA, a not-for-profit organization overseeing open-source development of VistA for use outside the VA. He left WorldVistA at the end of 2006 to found the VistA Enterprise Network, an IT consultancy. Marshall says he is writing a history of what he calls the "VistA software lifecycle." It is, in essence, an account of how the VistA software code was developed within a decentralized IT management system in which programmers worked over a period of years in close collaboration with clinicians at multiple hospitals within the VA system.
Marshall said he looked over Baker's list of tabled IT projects and reviewed his slide presentation on PMAS. "I think, based on this, it's possible to triangulate on Roger Baker and come to some conclusions," Marshall said.
"What he's doing is necessary; stopping the bleeding is necessary," Marshall said. "All of the worst offenders are on this list to be reviewed. It shows a clear desire to not fail on the scale we have been failing. It may not be a path to success, but at least it is not a path to failure. A lot of people in his position in the past have worked to create a system in which there is no route to accountability." With PMAS, in contrast, "the buck stops here."
"I'm seeing intelligence and good intentions, which is a huge step forward, but he doesn't yet grasp what he's up against," Marshall said. Centralizing IT management won't work in an organization as large as the VA with a software system as complex as VistA and with as vibrant and fast changing an environment as healthcare, Marshall said. Still, he said, "He might be open to hear what is needed to move forward. I'd like to talk to him."
What do you think? Submit a letter to Your Views. Please include your name, title, company and hometown. Health IT Strategist reserves the right to edit all submissions.
Also, please share your thoughts by taking our latest HITS reader poll.