Sapping VistA’s soul

In the past I’ve spoken highly of VistA, the Veteran’s Administration computerized
health records system, — and with good reason. VistA has a lot going for it. In 2006, it won an “Innovations in American Government Award” from Harvard. Studies show that use of VistA has improved VA productivity by 6 percent a year since nationalVa implementation was achieved in 1999. In a time of sky rocketing health care costs, VA care has become 32 percent less expensive than it was in 1996 in part thanks to VistA. The computerized system also has helped the VA reach an amazing prescription accuracy rate of over 99.997 percent. And last — but certainly not least — VistA is a flexible program that allows for much independent tinkering in the name of improvement, both by techies outside of the VA and those within the administration.

Given all these pluses, you’d think that the government would be happily throwing its weight behind VistA and ensuring that the system is firmly institutionalized for the long term. But in fact, just the opposite is happening. VistA is under attack; and it’s the federal government that’s leading the assault.

According to Dana Blankenhorn, a writer at ZDNet (a much-trafficked techie website),  VistA is dying “of starvation and neglect.” Its demise comes in part from an unlikely source: the Department of Defense (DoD). In 2005, the DoD introduced it’s own computerized health records system, called AHLTA. The system was developed by Integic, a private firm that was acquired by defense giant Northrop Grumman a mere nine months before AHLTA’s formal roll-out.

So why would the DoD contract out the development of a health records system instead of co-opting VistA, which can be reworked for different contexts? It’s not because of it’s too difficult, that’s for sure. Blankenhorn quotes Phillip Longman, a senior fellow at the New America Foundation and an outspoken champion of the VA  noting that the government “could wire Walter Reed or Bethesda (the two biggest military hospitals) for VistA in an afternoon. Technically there’s no big problem….”

Yet still, the DoD created an entirely new system — one which has only limited interoperability with VistA. Longman, the author of "The Best Care Anywhere: Why VA Health Care is Better Than Yours," explains how bad things are: “I just gave 11 (speeches) to front-line VA employees in the last few weeks, and I heard over and over again their frustration over not being able to get to the people at the (DoD) making the hand-offs (of patients between departments). Not only can’t the computers talk to each other, they can’t get the Army doctor in Germany on the phone to answer a simple question.”

This may seem silly but not outright harmful to VistA’s prospects — until you consider the fact that the DoD and the VA need to be able to work closely in order to coordinate soldiers’ care. When you introduce a new computer system, you also introduce the need for them to interface. Suddenly the issue isn’t expanding VistA, but merging it with AHLTA. The homogenization of a once-great open source platform becomes inevitable.

Indeed, the government is already working to fuse the two systems together—with help from the private sector, of course. This past September, the government hired the consulting firm Booz Allen Hamilton to “determine the feasibility of having a common electronic medical record system (across VistA and AHLTA) and what it would look like.” In November, Booz Allen and Gartner, a tech consulting firm, began helping the Army analyze industry suggestions on how to merge the two systems after the military issued a formal Request for Information to the private sector.

Of course, none of this would be necessary had the DoD used VistA as the basis for its technology rather than creating a new system from scratch — let alone one whose code and data is proprietary, and thus not easily reconcilable with other systems. (For-profit companies such as Grumman like to keep a tight grip on their products). But it seems that dishing out contracts has become more important than common sense. Last year the DoD awarded Northrop Grumman a $10.3-million contract to help deploy a common data pool for both VistA and AHLTA. (Northrop is a favorite of the federal government: the defense giant received a whopping $16 billion in federal government contracts in FY 2007, behind only two other military-industrial giants, Lockheed Martin and Boeing).

What’s going on here? Why is the DoD strangling VistA? Why is the government wasting time and taxpayer dollars on contracts while ignoring the potential of a proven, high-quality IT system like VistA? Longman has some ideas. For one, he says, “there are DoD people who have built their careers on AHLTA and want people to switch to their system.” Further, he notes, “the recent political appointees to the VA … are people with DoD backgrounds. And the DoD culture is ‘procure everything’ – they don’t make anything themselves, they procure it. When they get to the VA they don’t appreciate the open source culture.”

Longman may be right, but the DoD’s stance on IT is actually part of a broader paradigm shift that’s eroding VistA. The DoD wants what techies call a centralized IT system — that is, one with consolidated channels of authority and technological infrastructure. Traditionally the VA has a decentralized IT system, with 130 regional data centers and multiple IT teams working in local capacities in hospitals and administrations. Think of the distinction this way: the DoD wants one big IT department, and the VA historically relies on lots of little ones.

A centralized system provides more information security — after all, when you have a more standardized technology it’s easier for central management to protect the data because it’s housed in fewer places. A desire for this sort of consolidation lends itself to procuring commercial software, because it’s easier to buy a self-contained, out-of-the-box solution in order to ensure uniformity.

But ready-made, one-size-fits-all centralization lacks integration. You get a single type of platform, but it only performs limited functions. Unlike VistA, which contains modules with different functions (prescription drug ordering, electronic patient records, etc.) that nonetheless can communicate with each other, the commercial model lets vendors selling one small, specialized part of the IT chain. As more of these niche programs are purchased, the space for innovation grows smaller and smaller. It’s IT by way of assembly line.

VistA, of course, is famous for being open source software that is adapted by various IT teams and used for many different purposes. Unfortunately, the VA is moving away from this model — and not just because of the DoD. In November, the VA signed a contract with the Cerner corporation to replace its VistA laboratory software with Cerner’s proprietary platform. A specialized subset of the VA’s IT needs is now in corporate hands. The VA will no longer be able to tinker with lab software to produce home-grown improvements.

Admittedly, the VistA lab module was born in the 1980s and it needs to be updated. But do you really need to throw the baby out with the bathwater? As Scott Shreeve, founder of Medsphere, a company that adapts VistA for private hospital use, noted online, the fact that VistA’s old lab software is widely distributed and (still) functional in today’s computing environment — even if it’s not cutting-edge — is actually a testament to the program’s quality. Surely there’s something to work with here?

Nonetheless, Cerner got its foot in the door. (It’s worth noting that, the DoD contracted  with Cerner before the VA did.) Shreeve has called the Cerner grab “the first amputation in a long and steady surgical removal of VistA from the VA. Piece by piece, subsystem by subsystem, the VA appears to be taking" specialized software programs and string them together at the expense of “the beautiful and inherent advantages of a single, integrated software solution.” The future of the VA, it seems, is “a patchwork of [industry] solutions.” Shreeve continues: “Cerner now is positioned to wipe out VistA Radiology, VistA Pharmacy, Vista Registration, and ultimately the entire VistA clinical suite. Not only is Joe Public going to lose his several billion dollar investment in the largest and most successful implementation (of electronic health records) to date, but he is going to be paying even more in the future for all those Cerner licenses.”

This switch-over, notes Roger Maduro, an IT and open-source wizard who edits a VistA Newsletter, is “the first major departure from the VA’s successful development methodology for VistA. Over the past 30 years VistA has been developed in-house at the VA using its own development methodology.” But now the VA is turning to buying technology instead of developing it on its own. Its reasons are the same as the DoD’s: centralization and security. But at what cost?

As Government Health IT, a trade publication, noted in February, the VA’s “historically decentralized management and IT structures have created one of the most successful health care systems in the world.” VistA has been remarkably flexible and responsive thanks to its decentralization, and “centralizing the IT infrastructure could compromise VistA’s strengths by constraining creativity and introducing proprietary software into an open system.” The free exchange of ideas is effectively quashed.

Fred Trotter, an open-source medical software programmer and an IT consultant, spoke to Government Health IT about how the VA’s old open system used to work:  “historically, each hospital hired programmers to solve that hospital’s needs. Other hospitals then adapted those solutions to their own needs. [But] with the centralization process, all VistA programmers will be working for a central bureau. This could stop 30 years of innovation in which the best local innovations were taken national.”

It’s no exaggeration to say that we’re seeing the corporatization of the VA — not just in the sense of privatization, but also in the broader ethos of highly structured management systems. This means that the features that have made VistA such a success—flexibility, customizability, and openness—are in trouble. The balance between security and innovation is a very sensitive one, but in the past, we have figured out how to keep information secure. What’s harder is figuring out how to foster creativity. Centralized bureaucracy is rarely the answer.

Curiously, the government assault on VistA comes at a time when the system is making headway in private hospitals around the nation. Recently, for example, Shreeve’s Medsphere has implemented VistA in places such as Century City Doctors hospital in L.A., one of the largest physician-owned hospitals in the United States. Other medical centers that use VistA include the Lutheran Medical Center in Brooklyn, N.Y., Memorial Hospital of Sweetwater County in Rock Springs, Wyo., and facilities in the West Virginia Department of Health and Human Resources. The system is also in use at Midland Memorial Hospital in Midland, Texas, and at the U.S. Department of Health and Human Services’ Indian Health Service. Clearly, this is a system that works in a lot of different contexts—so why is the government abandoning it?

Healthcare IT is an incredibly complex field, and it’s easy for the average citizen to tune it out. But when a program like VistA,  is being phased out, we should pay attention — and ask questions. In 2009, we expect to see changes at the VA.  Health Beat will be keeping an eye on whether corporatization is hurting returning veterans.

Categories: Uncategorized

Tagged as:

8 replies »

  1. This whole talk is ridiculous. If it isn’t broke, don’t fix it. In the time it would take (YEARS) to rewrite the massive amount of functionality in VistA to a more “modern” programming language, who’s to say that new language wouldn’t be antiquated too. Like Peter said, who cares if the government and DoD loves wasting money on their favorite political contributors, VistA (DHCP) has completely transformed the VA healthcare system.

  2. Matt’s right. You can’t shoot down Vista for its reliance on MUMPS without hitting at least half of the other big systems out there that use it or one of its contemporaries. And if you think finding someone who knows MUMPS (modernized as CACHE) is hard, try finding someone who knows the guts of AHLTA. They all work for Integic.

  3. Like it or not, MUMPS (or Cache’) are at the heart of MEDITECH, EPIC, and other widely-used, commercially available EMR systems. Re-platforming is not a trivial task…especially if accomplished by a beltway bandit…badly…over the course of several years…at a cost of billions (a la AHLTA).
    AHLTA is a one-off, proprietary system that has cost the taxpayer billions. It’s not COTS per se. It was built for DoD. And is now an albatross.
    This is less about programming language and more about open vs. proprietary. The more money that’s spent on building new software, custom interfaces, replatforming, etc, etc….the less that’s left for implementing systems, re-working processes, and actually transforming care.

  4. Agree w. Alan. Coding in MUMPS is like writing a novel in Sanskrit. You will not find a MUMPs coder under age 50. Moreover, VistA has limited interoperability between the VA’s VISNs, its care regions. VistA is antiquated, and needs to be rewritten in code with enables modern database management and migration of data and function to the Internet. Niko’s post was a nostalgic political diatribe ideologizing the turf conflict between DoD and VA. Not an original contribution to human knowledge.

  5. Alan,
    While I grasp the wisdom of going with a modern language for the chassis, isn’t there an advantage to going with a one-off language for reasons of security? The fewer people familiar with the language, the fewer people capable of hacking it?
    I agree with Peter, though. One can only hope that the corporatization of legitimate government functions and things-that-work will cease under a new administration. But since the tentacles reach into Congress as well, it may be with us for good. Anyone got a connection with Obama’s policy shop?
    I also note that Humana in 2007 beat six other competitors to win all four contracts for something called Project HERO, a VA pilot that has gotten next to no publicity. Ostensibly it appears to be a test to coordinate referrals of VA members to providers outside the Veterans Affairs Health System. Again, it appears to be another camel’s nose under the tent.
    Learn more here:

  6. From a computer science standpoint, I would argue that building on any system that uses (MUMPS) is a very poor design choice.
    MUMPS, for those of you who may not know is the underlying operating environment, database, and programming language on which Vista is built. I searched on Amazon for books on MUMPS. I found 4. Published dates are: 1997, 1989, 1983, & 1981.
    Try doing a search for programming tools such as C, PHP, Ruby, Java, XML, Oracle, MySQL, etc, etc. You’ll find a multitude of resources on Amazon, Google, wikis etc.
    Here lies at least in part why the DoD or anyone may not be so keen on building off of the Vista source code.
    My point is that using such an antiquated, niche, development tool means that support personnel (i.e. people who know MUMPS ) are extremely hard to come by. Also this poses integration issues with modern popular architectures (i.e. Resource Oriented Architectures (ROA), REST, SOAP, XML, XSD, etc.
    The spirit of Vista is right on, but it needs to be easily accessible to your average software developer. This means in order for Vista to have staying power, it needs to be rebuilt from the ground up using modern languages such as Ruby, Java, or C, and use modern database systems such as MySQL and Oracle.
    I might add that data formats are the key to interoperability not so much the system itself. As far as open formats go I think that Harvard has the right open source approach with their PHR document standard.
    See=> http://www.indivohealth.org/pages/docstandard
    No I don’t work for these guys, but I think their approach to having schema definitions as public XSD is the correct modern approach to interoperability. Now if we could just agree on the schemas! Why doesn’t HL7 make schemas available on the Internet?
    Just my 2 cents.
    (Health IT Interoperability Champion and CEO of VIDENTITY SYSTEMS, INC.)

  7. DOD vs VA sounds a bit like the old Army Air Force vs.the Air force…..Competing governmental agencies which actually have different missions. Rather than having a central integrated system, why not have an ‘interoperable patch” such as Crossflo which is used by different portions of the justice department, law enforcement and used to harmonize different agencies for Homeland Security.
    Have DOD write off their ‘investment’ of my dollars and go with VISTA upgrades. They think nothing of sending billions on airplanes, ships and systems that are never implemented.

  8. “VistA is dying “of starvation and neglect.” Its demise comes in part from an unlikely source: the Department of Defense (DoD).”
    Typical of the Bush/Rove/Norquist/Repubican mindset – starve government departments then claim government can’t do anything. Oh, and put incompetent party hacks in charge just to make sure.
    “In 2005, the DoD introduced it’s own computerized health records system, called AHLTA. The system was developed by Integic, a private firm that was acquired by defense giant Northrop Grumman a mere nine months before AHLTA’s formal roll-out.”
    Out-sourcing has become the new method of rewarding your political contributors and turning tax dollars into political contributions. The dishonesty runs deep.