Dr. David Shulkin has either made a dramatic and long-overdue change, or he’s stomped on a hornet’s nest and unleashed furies that’ll eat him alive.
The new Veterans Affairs secretary certainly showed he’s a decision-maker by announcing VA will abandon its decades-old Veterans Information Systems and Technology Architecture, or VistA. It will adopt the same commercial electronic health record as the Defense Department — a product called MHS Genesis, from a company called Cerner. Cerner is a supplier to many commercial health systems, claiming installations in 25,000 locations. To be sure, it is a top-tier medical systems vendor.
Still, I’m going to go out on a limb here. After watching these types of projects for the last 25 years, I predict this will cost VA 10 times as much and take 10 times longer than its executives think.
There’s a real possibility of total disaster, as much as there’s the possibility of positive transformation. Shulkin clearly had some understanding of what it takes to even select a source for something so large and fundamental to the operation of the department. In trying to make the case for his no-compete, sole-source decision, he noted it took the Defense Department 26 months to go from requirements to contract award.
VA and DoD had tangled for years over ways to make common their record-keeping for service members after they transition to veteran status. Now it looks as if VA has blinked. I don’t question Shulkin’s motives. He’s trying to do what he thinks is ultimately best for veterans, who would be well-served by a seamless record-keeping system from their service days. He brings impeccable credentials and track record to this decision.
But if Shulkin or anyone thinks this will be a quick lift-and-shift, like changing out Lotus 1-2-3 for Excel, or the Lyft app for Uber, think again.
Shulkin’s decision hit the ranks of VA like a thunderclap. I spoke to a programmer at a southeastern regional facility, who said so many questions popped up immediately that the center director has scheduled a series of town-hall question-and-answer sessions for Wednesday. This man has been programming VistA applications in the M language for more than 30 years.
VistA is more than an electronic health record like your dentist might have. It’s more like the VA’s nervous system, with hundreds of applications covering everything from patient dietetics and recipes to facilities engineering. It populates data warehouses, from which VA managers around the country subscribe to reports. Nearly all of VA’s nearly 345,000 employees touch it in one capacity or another. It may sound old-fashioned, because it goes back a half century, but the M language — developed not by VA but by the Massachusetts General Hospital — has technical characteristics that make it surprisingly contemporary for large-scale transaction systems.
I say all this not to sell VistA, but just to illustrate that it’s not some old COBOL application with discrete outlines that you can easily excise and replace. Mismanagement of the replacement could literally bring VA to a standstill. Think of the data migration challenges alone. VA officials will be putting the viability of the department — and its implicit promise to the nation’s some 19 million veterans — into the hands of a vendor, one still largely unproven in the military-veteran complex.
Cerner, I’ve no doubt, is a fine company. It has a strong track record. But it is a company operating in the modern business world. It’s facing a class-action suit for alleged failure to pay overtime. Modern Healthcare reports Cerner is also in a tangled breach-of-contract suit it brought against a three-hospital company in Pennsylvania. What will happen when VA and Cerner come to an impasse over this or that, as they inevitably will?
My programming source says it will take 20 years to replace VistA. He’s only half joking. It took the IRS 15 years to get its 1990s modernization effort out of first gear. DoD, after 30 months of effort, has MHS Genesis running at just a single, small location.
VA also has a human capital issue to deal with, given the large number of employees devoted to VistA and M programming. Many are approaching retirement age and could simply opt out of starting over. Does VA replace them with programmers familiar with the languages used by Cerner? The company lists these as “Java, Ruby, C++, C#, Objective-C, and Python, while leveraging technologies such as Hadoop, Rails, iOS, Android, Eclipse RCP, Chef, Github, Maven, and Splunk.” Or does it contract out everything? If the latter, will it pay by the hour? By the application? By the report? By the update?
Maybe Shulkin made the right decision. But to make it so, no one from VA should walk into this with any illusions about the complexity or cost of the task they’re facing.
That so many news outlets cover federal information technology is in part testimony to a pioneer in the field, Israel Feldman, who died last week at 86. In 1982, following a federal IT career, Feldman founded Government Computer News, which at its height was the largest and most authoritative publication in the field. I joined GCN as editor-in-chief not long after Feldman sold it to a large multi-title publisher. But he continued to maintain an office at our Silver Spring offices, and I got to know him.
Izzy was a visionary, a builder, and an energetic and forceful personality who left an impression on everyone who knew him. He was also a driver behind the formation of the Industry Advisory Council, which became and remains a large and influential force in the $90 billion federal IT market. He was born in Romania, but at the age of three, emigrated with his family to Israel. He made a fortune in the United States and became a major benefactor of the Technion, known as the MIT of Israel.
I liked Izzy Feldman and I’ll miss him. Read more about him here.