I remember like it was yesterday the example used in my MBA economics course to teach the Law of Diminishing Returns. A fishing boat captain can get dramatic benefit by adding a second fisherman, nearly as much so by adding a third, but then less and less as he hires more fishermen. The reason is that they start to get in each other’s way. Plus, there’s only so many fish to catch.
In more formal words, marginal production decreases as scale increases. That means that unit costs actually go up instead of down since productivity has been reduced. There’s an optimal point at which further expansion is a bad idea.
I think this rule applies to big healthcare IT projects. Department-level rollouts work. Bigger ones often don’t.
Ambitious HIT failures in Canada and England seem to bear this out. At some point, increasing IT project scope causes a rapid increase in cost even as the marginal benefit decreases, especially when the government is involved.
(There are exceptions, like the VA’s VistA. It was still wildly expensive, but was successful because it was given an unprecedented strategic focus).
For the most part, the productivity of a given knowledge worker should increase if you give them a PC and an application specific to their job. The same holds true when you arm a workgroup or a department with technology. A focused hospital department such as lab, rad, or pharmacy could not function without their technology tools.
The returns seem to diminish, however, when you multiply that times 500 or 25,000 users in a big department (nursing) or a big group of hospitals. I’ve been trying to figure out why that’s the case.
I’m blaming diversity for starters. Bigger numbers mean a wider range of user types, processes, organizational priorities, and locations. That makes it hard to wring value out of a one-size-fits-all system that was dumbed down to meet the needs of a wide variety of users and organizations. Hospitals are not good at standardizing their practices, so enforcing a single set of rules never works (even when computers aren’t involved). Airline people do fine with a standardized reservation system only because there are only so many ways you can make a reservation and only so many types of workers doing it.
Change management is also a big problem, stretching the capabilities of most organizations to communicate. The “turn the battleship” metaphor is real, especially in big hospitals. It’s hard to implement fast and long-lasting change without having a dangerously maniacal yet passionate autocrat in charge (think Steve Jobs) who is willing to fire the foot-draggers. You don’t find many of those running hospitals.
I think that’s the Achilles heel of big HIT projects: they primarily exist to mandate conformity that humans naturally resist. Enforced conformity, in the absence of authoritarian rule, always fails. It’s easy to find a group of lab, rad, or pharmacy people who think and work nearly identically. That same organization’s 1,000 nurses will wildly disagree how to their jobs should be performed.
The result is diminishing returns. The cost per user or cost per discharge paradoxically goes up instead of down.
Imagine if your Internet experience had been designed by HIT software vendors. Everything would require formatted data entry. Security would be enforced the point that nobody could see much of anything. Instead of Google, searches would require querying using the data entry forms. Standardization and risk reduction would be the ultimate goals. All user experiences would be intentionally identical, regardless of their individual needs or wants.
So that’s my theory and I’m sticking to it. Big IT projects fly against human nature and organizational behavior, which is always a risky bet. It’s tempting to envision a massively deployed, universally adopted, centrally managed IT system that enforces business rules while delivering positive return on investment. Unfortunately, that almost never happens.
Perhaps the good news is that we’re not trying that in this country. Individual physician practice adoption of electronic medical records is really more of a workgroup-scaled project. Data sharing via health information exchanges or the Nationwide Health Information Network follows the Internet model, which works.
Those projects should not have the risk of an all-or-nothing EMR rollout in a large hospital or a multi-hospital group. That remains the greatest risk of diminishing returns when it comes to HITECH money.