Health IT

By Alice Lipowicz

Blog archive

Baker responds to VA VistA report

Roger Baker, assistant secretary for information and technology at the Veterans Affairs Department, today thanked an industry panel for its “exhaustive” study on modernization of the VA’s VistA electronic medical record system.

The panel, convened at Baker’s request, was sponsored by the American Council for Technology-Industry Advisory Council and chaired by Ed Meagher, a former deputy CIO at the VA. It recommended on May 3 to move VistA to an open source platform and phase out its reliance on the MUMPS computer language.

“The VA is pleased with the efforts of the ACT-IAC to review the option for moving VistA into an Open Source endeavor and especially the dedicated efforts of the members of the work group and leadership of ACT-IAC,” Baker said in an e-mailed statement today.

“Clearly the group did an exhaustive study covering the significant technical and governance aspects important to making decisions in the Open Source arena. We are currently reviewing the many recommendations covered in the report and look forward to a rich dialog and exchange of ideas and concepts with industry. VistA is the preeminent example of what a patient centered Electronic Health Record [EHR) can do, and its improvement will take the joined efforts of clinical experts, government and industry. The VA looks forward to continuing this review,” Baker wrote.

The 34 vendor panelists were said to be unanimous on the recommendation; however, reaction in the broader health IT has been divided.

VistA is considered a superior system with strong clinical content and effectiveness, but it operates with the relatively unpopular computer language known as MUMPS. Some supporters and practitioners of MUMPS strongly laud its qualities, while detractors assert that MUMPS is outdated.

I’m certainly no expert on MUMPS, so I will offer only one brief observation.

From what I’ve heard, one of the elements that makes VistA superior to many commercial EHRs is that it was created with substantial input from physicians who would be using it. It has been described to me as “clinically rich.” Because the doctors are — or should be -- the ultimate arbiters of how their own records should be kept, it seems to me it would be important to hear from physicians on whether the clinical richness would be threatened by moving to an open source development.

I asked Meagher about physician input into the industry committee’s work, and he told me the panel spent two months consulting with VA doctors and other VA officials about their needs. He also said VA physicians and “end users” would be included in the open source development process.

A principle of modernization is that “whatever is done must be at least as good as what we have,” Meagher said. “We want to make the system as robust as possible.”

Good intentions are certainly a necessary first step. It would be great if the VA physicians could get involved in this public discussion about a very significant issue that may have much broader implications for health IT implementations.

Posted by Alice Lipowicz on May 10, 2010 at 12:14 PM


Reader comments

Tue, Nov 16, 2010 Milkyway Mapping USA

MUMPS and its integrated schemaless sparse array database (aka Globals is a mystery to the uninitiated) is being used by the European space agency to map the entire milkyway (our own galaxy) that has millions of stars and billions of planets and other phenomena. Consider the complexity of describing planetary bodies and their attributes in motion via a 4 dimensional computer model that spans light years in all directions. Try to do that with Oracle, MSQL, Java or C# and it would be a certain failure!

Tue, Jul 13, 2010 Chris California

I take great exception to the statements about MUMPS in Tom Love's description. The VA has had more than twenty years to replace MUMPS in VistA. Where is their solution? What makes them thing they are going to find it now after 20 years and gobs of money thrown at their goal of "Kill MUMPS First". Actually, MUMPS is very heavily used in Banking, Travel services, the European Stock Exchange is a MUMPS system (or at least it was last time I spoke with some of the folks who support it), libraries, nearly all of the Yellow Pages were published using MUMPS. What is he talking about? He really should read about "No SQL" a new movement that is ripping holes in the relational model. MUMPS is the oldest and most venerable of the "No SQL" Model. Yes, MUMPS can store the date in a variety of formats, but VistA standardizes the data types very nicely and provides LAYGO (Learn As You Go) technology which allows your database application to grow as needed without any pre-allocation. VistA is a fully Enterprise Wide Database solution which is fully scalable from a no bed clinic to a thousand bed hospital without any major modification to the basic code. VistA is a tool kit that allows the gradual expansion of the technical services required at a facility. It is easy enough for nurses, ward clerks, lab techs, and pharmacists to master. This is why the VA still can't rid itself of VistA, it is too competent. The people at the point of care are making the decisions rather than software specialists who have never spent a single night in a hospital ward making policy for the doctors and nurses. That is a huge difference. You can download a VistA Appliance from Source Forge and install it and see CPRS running in less time than it takes to download it. Go take a look for yourself.

Wed, May 26, 2010 Srini

I believe Open Source is the right way to go. Regrardless of how innovative, smart an organization (or a company) is, there are more smart, creative, and innovative people outside your organization. There are quite a number of examples that validate this. Look at the success of Linux, MySQL, a suite of Apache Projects, JBoss and SpringSource in the serverside Java. The VA community and the entire healthcare community can benefit from an open source EHR with collaborative development among technologists and domain experts. The success of an open source project depends on a healthy community. Hopefully This can generate enough momentum to create an healthy community.

Wed, May 19, 2010

It is what it is, but it is not what it should be. What VistA has always been is a paperless equivalent of a paper medical record. It stores information quite well for quick retrieval, as if reading a medical record. What it is not, yet what it needs to be is a patient care management system. As with other large software systems, it was designed to fulfill a particular need, electornic recording, but has not evolved as other sectors of the industry have. Coming from the private sector into the VA I have been quite impressed by the ease at which the large volume of information is recorded and retrieved by VistA, but also dismayed as I experience the long delays and arcane processes to retrieve actionable knowledge from the system. I also heartily agree that there is a surfeit of staff within the system who could tackle the herculean task of moving this platform forward. However, I dread the low-bid solution that will be forced upon us in due course.

Thu, May 13, 2010 Tom Love

As a member of the ACT IAC executive committee for the VistA project, let me help clarify a couple of issues. MUMPS is an esoteric (rarely used) programming language designed in the 1960's. Like Basic, it was comparatively simple, interpreted, and has one very unusual feature -- built in "persistence". So it was a language in which medical information could be easily stored and retrieved. It has persistence, but not a database management system. AND it so happened that in order to work on the computers of its day, it had to store and retrieve this information quickly.

But MUMPS does not have any of the features that subsequent programming languages have found invaluable (some really important; others proved to be just plain worthless). Here are a couple of examples. MUMPS is a typeless language so you can create a variable called date and store in it either 20910 or 9Feb2010 or Tuesday. MUMPS is not an object-oriented language -- the largest, most important, and least understood feature of modern languages. Object languages have routinely resulted in 5:1 reductions in code bulk, dramatic increases in reusability, and many fewer errors.

What is truly unique about VistA is that it was built using what today we would call agile development with extreme co-location of developers and users. Often one doctor and one programmer sat together in an office after normal working hours to design, implement, and validate new applications. More than 108 such applications now exist in VistA -- all largely self contained with a minimum of sharing between the applications.

So the committee said, "what if we had a modern day VistA designed, architected, and implemented using state of the art software?" We concluded that the users would be delighted as long as it was easy to modify, continued to run as fast and as reliably as today's VistA. This may not be easy to do, but if it could be done and done in 1/5 the code with much higher quality, VA would be delighted because maintenance costs would be lower and new features and enhancements could be delivered far more frequently.

Thus the recommendation "use VistA as a functional specification" and "reengineer it using state of the art software technology, agile methods, extreme colocation, and improved quality processes throughout".

We did not recommend a conventional modernization project -- most (if not all) of which have failed. Instead we are saying, "take what you have, make it much, much better from an engineering perspective, but change as little from a user's perspective as possible.

It's like reengineering a vacuum tube radio to one built with semiconductors -- that's reengineering. And only then begin thinking about improving that semi-conductor radio to accommodate stereo, satellite receivers, and TV surround sound.

Scope management is everything in the enterprise software business…

Show All Comments

Please post your comments here. Comments are moderated, so they may not appear immediately after submitting. We will not post comments that we consider abusive or off-topic.

Please type the letters/numbers you see above