Veterans enrolled in the Veterans Health Administration, part of the Department of Veterans Affairs, have to answer a variety of questions regarding compensation, pension status, military service and more. Based on those answers, a veteran is given a priority group status label (from 1 to 8) within the VHAs continuous enrollment system.
It sounds simple, but that wasnt always the case. There are 40 questions, and they must be implemented by the system and answered by the veteran in a very specific order. If not, the worst-case scenario happens: Veterans who should have stayed enrolled become unenrolled, warned Lynne Harbin, associate director at the Health Eligibility Center for the VHA, in Atlanta.
It took Harbins team six months to understand and fix all the issues with continuous enrollment. Problems ranged from coding issues, unclear business logic and improper sequencing. Harbin admitted it took an incredible amount of software development time to rebuild the process to make sure it worked.
Policies were ingrained in the VHA—the largest health care organization in the United States, with 7.6 million enrolled participants. Continuous enrollment was just one of many of the VHAs rules-based applications that were in constant jeopardy of breaking down.
The collapse of rules resulted from their lack of visibility.
“It was very difficult to maintain. Theres very little documentation,” said Harbin. “Everything is hard-coded.”
Quite simply, “Its difficult to identify what the changes are in the context of our existing system,” Harbin said. “The system is poorly documented, and memory only goes so far. Thats where we tend to run into trouble.”
To help reveal the business specifications of rehosting and modernizing its legacy enrollment system, the VHA brought in Philip Matkovsky, principal with Macro Design Group, of Arlington, Va.
Matkovsky is a systems consultant with plenty of experience in the public sector. More important, he has worked with the VHA. It was his job to discover problems with the rules engine. Given that there was a major gap between documentation and code, this job wasnt going to be easy.
“The rules sit all over the place,” said Matkovsky. When changes were made, documents often were not updated. Just figuring out what the current rules are in the VHAs embedded system became a challenge in itself, he continued.
Resolving problems has been a daily chore for the VHA. Unfortunately, a simple change in the past was unnecessarily complicated. Given the current system-coded, IT-driven, rules-based environment, it is impossible to be agile among the VHAs world of changing policies.
For about seven years, Harbin and the team had looked at deploying a BRMS (business rules management system) to take control of the rules and regulations of the VHA.
If they chose a BRMS, it could extract rules-based coding from their current system layer, giving the VHA greater visibility into what the rules are, thus allowing them to deploy changes faster than they currently can.
By 2004, Harbin and Matkovsky started seriously reviewing a handful of leading industry applications, looking intently at Ilogs JRules, a BRMS. At the time, there was another VA office in Colorado that was testing JRules. Its there that the VHA got a firsthand look at the tool that had a business-friendly focus while still satisfying the demands of the VHAs IT department, Harbin said.
“The whole idea around business rules management is making it easier to manage business logic,” said Pierre-Henri Clouin, director of strategy at Ilog, in Mountain View, Calif.
When you deploy a BRMS, business logic is unbundled from the rest of the system, creating another level of specialization, which, in turn, facilitates the VHAs massive system upgrade, Matkovsky said.
That overall system upgrade is being managed by Electronic Data Systems. According to Clouin, EDS spent two years trying to build its own BRMS and failed, primarily because EDS couldnt keep up with the pace of the VHAs ever-changing regulations. Clouin said that his company was brought in because the VHAs rules engine project was reaching a crisis level.
JRules automatically creates an interpretable documentation from which the rules are directly derived. Analysts can choose to read the rules in natural “if then” language, as a decision table or in a decision tree, Clouin said.
With the new JRules system, Harbin and the team will be able to test rules in a sandbox environment before theyre deployed. To be able to see the effects of changes and be able to tweak them will greatly improve development time, allowing them to implement regulations at a much faster pace. Theyre nearing the end of their integration testing and hope to have the new system up and running by early in the first quarter of 2007, Harbin said.
While Matkovsky said he enjoys the value of being able to first work in the sandbox environment, he said he believes speed to deployment is dependent on building specifications upfront and not waiting until you get to black box [final] mode.
“Adding precision to the analysis activity reduces the overall life cycle for the long term,” said Matkovsky. “By the time you get to QA [quality assurance], the net effect of those changes is far more expensive than [adding precision to the analysis activity].”
Clouin offers a similar recommendation, advising that people not wait until a project gets mired in the issues inherent in legacy systems. Its a complex process that requires expertise. Separating business logic from other aspects is not easy.
Hire someone who has done it before, Clouin said.
And lastly, Clouin said that everyone needs to realize that a full-fledged business rules approach requires organizational changes. It can be worrying for the IT department while businesspeople infiltrate their territory, he said.
Throughout the past year and a half that the VHA has been working with business rules management, Harbin said she feels validated that the VHA chose the right approach. “It takes miles of hard codes [to do what] you can do easily on a decision table,” Harbin said.
David Spark is a freelance writer in San Francisco. Contact him at [email protected]
Request for Comments
Have a comment or suggestion? Please e-mail Solutions Series Associate Editor David Weldon at [email protected]
Case File: Veterans Health Administration
- Organizational snapshot: Largest health care organization in the U.S. with 7.6 million enrolled participants
- Business need: Needed more visibility into its rules-based system that determines health benefit eligibility for veterans
- Technology partner: Ilog, of Mountain View, Calif.
- Recommended solution: Use Ilogs JRules to extract the rules management from the system into a separate, more visible business rules management layer; the project timeline was two years