VISTA Enterprise Network - Successful Implementation, World Class Support

Thursday, January 14, 2010

It's the Culture, Stupid!

Dear Reader,

ACPE’s 2004 Technology Survey identified one bright spot in the health IT industry: the U.S. Department of Veterans Affairs’s VISTA system (Veterans Health Information Systems and Technology Architecture). VISTA has an unusually high user-approval rate, has won Harvard’s Innovations in American Government Award, and has dramatically and measurably improved the quality of healthcare at VA facilities over the last quarter century. When Hurricane Katrina devastated New Orleans, the only medical records to survive were those in the VISTA system at the Veterans Affairs hospital; VISTA was back online after only forty-eight hours of downtime.

Because VISTA is not only a high-quality health IT system but also a public-domain one, it is increasingly being adopted outside VA. The national health IT movement’s increasing awareness of the importance of IT is contributing to VISTA’s rising popularity both in America and abroad, which is coalescing into an international VISTA movement.

VISTA is not only an unusually successful software product but also an unusual health IT software-development culture that grew from an analysis of not only why most health IT projects fail so often but also why VISTA projects tend to succeed.

This same analysis explains why VISTA is an unusual success story for health IT: it is developed according to an alternative software-development culture that fits the needs of medical culture better because it essentially is the medical culture. VISTA is developed by a community of programmers most of whom began their careers as doctors, pharmacists, lab techs, or other medical professionals. Not only were they not steeped in the stasis-seeking software-development culture, they were steeped in the medical culture, which is used to a continuous state of changing needs.

Accordingly, the alternative software-development culture they created does not seek to achieve a perfect status quo but instead a highly responsive fluxus quo to keep up with the pace of medical change. Instead of seeking to avoid errors, this approach seeks to fix them quickly, something that would be impossible under the staggering quality-assurance (QA) overhead of the dominant paradigm.

Most importantly, in healthcare the stakes are too high to measure software correctness any way except by how well it meets the current state of medical needs; instead of measuring correctness by adherence to specifications this approach measures it by user satisfaction. Since medicine is far too complex for any individual or small body to authoritatively define, authority over what is to be done to the software is put in the hands of all of the users and the developers are given the authority to make whatever changes they need to whenever they need to in order to please their users. In short, authority is decentralized and the future of the software resides in the hands of an ongoing collaboration between the health professionals who use VISTA and the software engineers who develop it for them.

To allow the customization and peer review required in medicine, the source code for VISTA is open and the software itself is free. To avoid imposing any kind of penalty on efforts to make the software better serve medicine, adopters are not charged any kind of fees to report problems or have them fixed, nor are they charged to have improvements made. The economic model is instead (1) fee-for-service to set up a new VISTA site and train its adopters in how to use it, and (2) fee-for-relationship for ongoing support to encourage adopters to make as much use of support as possible, the better to channel their insights into the software development lifecycle.

In these and many other ways, the VISTA software-development culture is essentially the opposite of the dominant paradigm. It has more in common with more recent upstart methodologies like the open-source movement, rapid prototyping, agile programming, extreme programming, and so on, though it has been doing these and many other highly unusual things since long before any of these new methdologies had names.

The upshot of this VISTA analysis of the state of health IT can be summed up as follows.

Even if a perfect health IT solution were installed at a hospital, if the dominant software methodology is followed that software will become less and less able to meet the needs of its adopters as the state of medicine shifts out from under it until it eventually becomes a threat to the health of its patients. That is, good health IT software goes bad over time under the dominant software-development culture. Likewise, even if a dreadful health IT solution were installed at a hospital, if the VISTA software methodology is followed that software will become more and more able to meet the needs of its adopters and will change to take into account advances in the state of medical science. That is, bad IT software turns good over time under the VISTA software-development culture.

Yours truly,
Rick

2 comments:

Rodney H. Kay said...

Do you ever wonder why this has been so hard to figure out?

Rick Marshall said...

Dear Rodney,

Yes, I too often wonder about this.

I believe the more obvious but smaller cause is a misapplied notion of pragmatism that makes people think they can put off wisdom until later to focus on getting things done now. The obvious problem is how is one to know which things should be done if one has poor judgment?

The less obvious but larger cause, I think, is what Heraclitus wrote about twenty-five hundred years ago when he wrote that the Logos (the hidden harmony of the cosmos) is right in front of everyone all the time. They look right at it but they don't see it because of their incredulity. They disbelieve that what steers all things actually does so. They replace the truth about the cosmos with simpler explanations that make them more comfortable, that better fit their past experience.

The more successful someone has been in the past outside the VISTA arena, the more likely they are to do this, because they feel naturally enough that their past successes validate their worldview. Of course, since their mental models describe VISTA very poorly, they are frequently running up against situations that they didn't predict. Rather than concluding that their explanations might be too simplistic, they instead try to reinterpret what happened in terms of their faulty models, and when they can't they disregard these situations as anomalies and then do their best to forget about them.

When too many expensive failures pile up, they change jobs and move on to things they think they can better manage, leaving chaos and destruction behind them for the next misguided soul to try to deal with. It would be interesting to see what would happen if a top-level manager were prevented from leaving, if he were forced to stay long enough to learn the consequences of his actions so he might have a chance of learning from them and changing strategies.

So it is in part the turn-over among the political-appointee layer of management that perpetuates this system by keeping each manager's exposure to VISTA too brief to overturn a lifetime's worth of accumulated confidence. That's a stimulating line of speculation but ultimately futile, because unfortunately VISTA's history demonstrates that when bad managers' noses are rubbed in their past failures in an effort to get them to learn from their mistakes, rather than doing so they insulate themselves from the facts by reorganizing to add more layers of management between themselves and the damage they're causing, and they insulate themselves from responsibility by creating systems of authority that destroy accountability. The end result tends to be an organization with many layers of management in which no can explain who made the bad decisions.

The human capacity to disregard reality inspires awe and dismay in any who study it for long.

To someone who gets what's going on, it looks like all these people are sleepwalking, but where it gets surreal is that the sleepwalkers are generally in charge and making the decisions.

Thus the failures of centralizing VISTA authority lead to further centralization, and the successes of the very different classical VISTA model are waved off as history and therefore irrelevant.

As for the solution?

Well.

I conclude from studying our thirty-three years of VISTA history that no large centralized organization, neither government nor a large corporation, is competent to manage VISTA because sooner or later they will always try to centralize authority over it to fit their own structure. Doing that is committing a primal sin against VISTA that always leads to failure. Always. Since neither the federal government nor any individual corporation will permit itself to be decentralized, VISTA is going to have to be developed elsewhere.

I predict the next couple years will be a very interesting time for our community as we proceed to do exactly that.