The Veterans Affairs Department is starting to deliver applications for its new HealtheVet medical information system, which means the end-date for the venerable Veterans Health Information Systems and Technology Architecture (VistA) is finally on the horizon.
Delays in developing HealtheVet have pushed back completion of the program to 2018, six years later than originally planned, according to VA officials.
They say VistA is outdated and can’t be easily modernized and that HealtheVet will fit better with the information technology architecture VA and the Defense Department are developing. However, critics of the plan say VA is giving up too much by retiring VistA.
“Can [HealtheVet] still provide the same care for veterans that is as good as they get through VistA?” asked Claudine Beron, chairwoman of the VistA Software Alliance, which promotes the use of VistA by non-VA organizations. “The answer to that right now is no.”
Beron said VA’s ongoing effort to centralize systems and consolidate VistA databases at four data centers will pose problems because standardizing the base code means the agency might have to do away with local VistA programmers and developers, who happen to be the clinicians who use VistA.
That grass-roots knowledge has enabled VistA to become the force it is today and allowed VA to provide the level of care it does, Beron and other VistA supporters say.
VA began using VistA in its hospitals and clinics in 1984, and officials have since added many functions and a graphical user interface. The system is now a collection of more than 100 integrated software modules that cover a wide swath of clinical applications, such as patient records, admissions, discharge and scheduling, and nonclinical applications, such as billing, accounts receivable and engineering.
It’s platform-independent, and because its applications are integrated at the database level, information can easily be shared among applications and the various institutions that use VistA.
HealtheVet is intended to modernize and enhance VistA through the use of new information tools and software languages. It will rely on the IT systems approach that the entire federal government is heading toward, called service-oriented architecture.
SOA involves making a set of standard software services or modules available to all users. It can quickly pull information from multiple sources and deliver it just about any way users desire.
Block by block
With HealtheVet, VA officials are seeking a more flexible and patient-centric system in which users can quickly assemble, add to and take away various Web-based services to build a collection of applications that suits their specific purpose. And each of those applications can come from a different source.
In contrast to that loosely coupled approach, VistA is a single, tightly integrated system, and any addition to the code must be carefully tested to make sure it doesn’t interfere with other parts of the program.
VA’s plan involves breaking the functional content of VistA into blocks that will be rewritten for HealtheVet’s SOA environment, said Dr. Paul Tibbits, deputy chief information officer at VA’s Office of Enterprise Development. As HealtheVet blocks are turned on, the corresponding VistA blocks will be turned off.
Two HealtheVet modules have already been developed. The one for patient enrollment is in alpha testing, Tibbits said, and tests of the scheduling module are due to begin soon.
The new approach doesn’t necessarily mean the future will be a totally HealtheVet one, Tibbits said. Some modules are three or four years from being launched, which means pieces of VistA will continue to be enhanced during that time.
Critics contend that the agency is swapping a proven system developed largely by knowledgeable programmers a t VA’s hospitals and clinics for one that might not be as user-friendly or efficient.
On the other hand, they say non-VA users of an open-source version of VistA are improving the system faster than those inside the agency. Open-source vendors used the publicly available VistA source code to develop versions for state and local governments in the United States. In addition, other countries are increasingly turning to open-source versions of VistA as inexpensive and proven solutions for their health systems.
“You can’t just replace VistA with another system,” said Fred Trotter, leader of SynSeer, a developer of open-source clinical IT software. “There are certain features that HealtheVet gives you, but if the VA would open up and let outsiders in, then they could probably get those features in and working at their hospitals in a couple of years.”
Critics also say doing away with the expertise of the clinician programmers who built VistA will inevitably lead to a poorer product in HealtheVet.
“Tibbits has said he is trying to nurture a creative environment at the VA,” Beron said. “The trouble is that lots of ideas don’t come to the top [in a centralized system], so the only way a programmer can get something in is to put it in and not report it.”
Change control is essential
Tibbits said he wants to retain the creative environment that enabled VistA to develop to its current level, but the improvements VA officials are seeking require the use of strict change control for IT processes.
For VistA and HealtheVet, that mechanism involves the Veterans Health Administration nominating products developed at VA sites that VHA wants in all its hospitals, he said. VA gives that product code to an inspection team, which evaluates whether it is safe to use in a production environment or might have a negative impact on resources such as bandwidth.
If it passes the tests, the code is included in a release package that is sent to all VA hospitals. About 20 such products are already under assessment through that process, Tibbits said.
Furthermore, he said, the open-source community should not assume that VA officials do not want to include its products. “In fact, there are some inside the VA [who] very much would like to do that,” he said. “It’s conceivable that as we become more mature in our change control, we could apply those same processes to code developed elsewhere. I see no reason why we couldn’t have that in place within the next year for the open-source community.”
