VA weeds out poorly performing programs

Roger Baker, chief information officer at the Veterans Affairs Department, drew widespread attention when he announced in July 2009 that he was suspending 45 underperforming information technology projects under his new Program Management Accountability System (PMAS). Since then, 32 of those projects have restarted, 12 were canceled, and one remains suspended.

PMAS, pronounced “P-Mass,” is now in effect for all 280 of the department's IT projects. Baker recently spoke with Federal Computer Week staff writer Alice Lipowicz about PMAS and other issues.

FCW: Tell us more about PMAS. Is it a formal process?

Baker: It is becoming a very formal process at the VA. We now have a dashboard to track projects on a public Web site. We have staff dedicated to the PMAS as a process. I am not ready to declare victory on it, but we have made a lot of progress.

The PMAS is a set of principles, combined with good private-sector practices of incremental development.

In the PMAS, I am being draconian about dates. In the private sector, every project gets banged up against the great wall of profits. If it is not generating results, eventually someone says it is a dumb expenditure and it should stop.

In government, no one is willing to say, “Let’s stop it.” Having a schedule is a rock to be pounded up against. PMAS has been very useful in getting project teams and the customer to work to resolve problems instead of dragging things out.

FCW: Is there a PMAS handbook? Can others in government learn from your experience?

Baker: We have not been focused on documenting PMAS for the outside world. Frankly, we are focused on doing it well at the VA.

The first time we introduced PMAS, it was in a five-paragraph e-mail. There was a memo in February referencing a PMAS guide. I have not read the guide. There are lots of good people on it.

At this point, PMAS is running really well. We have statistics documenting that. We are 10 months into it. We still need to know exactly how the decisions work and what conditions constitute a milestone prior to delivery to a customer. We are working on milestones for each project.

The PMAS has a firm set of principles: Deliver software in six months or sooner so that it can be evaluated by the customer. The date is paramount. There are various ways in how it needs to be applied. Nowhere in our PMAS guidance does it say when the software must be in production. The customer is in control. The PMAS gets the project to the point of being accepted by the customer.

FCW: What is the latest update on the 280 projects in development?

Baker: [As of April 20], 92 IT projects are “active” and able to spend down their budgets, 93 are “in planning," 53 are “not started," 21 are “closed," and 21 are “other.”

The 92 active projects have the go-ahead to spend fiscal 2010 funds.

It is clear we are not spending a lot of money right now. We are being very tight on which projects we are starting to work on.

FCW: On March 1, the VA shut down its bidirectional system between VistA and the Defense Department’s digital health record system, AHLTA, due to errors popping up. You’ve had some difficulties getting it back online. Any lessons learned for the VA/DOD Virtual Lifetime Electronic Record (VLER)?

Baker: We have learned that some bugs are really tough to find. The only place the errors occur is in production. We were unable to replicate it in a test environment. Last week we bought up the product environment without users to do testing.

The bidirectional system is a very old system. No part of what we are doing in our future work on VLER will be on it.

The point of the VLER is to move the VA/DOD health data exchange to the Nationwide Health Information Network. We are replacing 1990s technology for a 2010 technology.

FCW: VA and DOD’s first joint health facility is set to open in north Chicago in October. What lessons do you hope to learn from that project?

Baker: At [the North Chicago center], we are doing something that no one else has done: physician order portability between two record systems. We have heard people say you cannot do it.

On the operational side, I think there will [be] a lot of lessons, like putting orders in AHLTA, then into VistA, and back to AHLTA. We will learn how two organizations work together as one operational model.