Free Markets, Free People
Another software disaster in the making?
Over at The Corner, there’s a discussion going on about medical records, prompted by this sentence in the “stimulus” plan:
“Computerizing every American’s health record in five years, reducing medical errors and saving billions of dollars in health care costs.”
There are the usual (and valid) privacy concerns. But Iain Murray goes further, and wonders:
I’m not sure why insurance companies haven’t insisted on it, but my guess (and I stress it is a guess) would be some regulations related to privacy, which was the source of the AMA and ACLU’s opposition in the past.
From someone sitting inside the world of healthcare software, perhaps I can enlighten things a bit. Privacy is only one of a host of challenges. A bigger obstacle is the difficulty involved. Creating the software that manages patient data electronically is, to put it bluntly, beyond the capabilities of almost all software developers. It’s really, really hard.
Here are some of the challenges in creating such software:
1. The data is very complex. It’s not just numbers and text; it includes all kinds of media, which needs to be interpreted and annotated.
2. The data evolves rapidly over time. New tests are constantly being created.
3. How the data is interpreted varies rapidly over time. Today’s rule might be “you need a prostate exam if you’re over 50 and blah, blah”, but a cheaper, less invasive test next year might mean it changes to “you need a prostate exam if you’re over 40, period.”
4. The users are very difficult to please. Doctors are the most difficult users I’ve worked with in an entire career of software development. They won’t sit still for two weeks of training. If it’s too hard to use, they just won’t use it. They’ll keep using paper. (Given the responsibilities with people’s lives they have, that’s understandable.)
5. There are laws (HIPPA) concerning privacy that are difficult to design for. The rules are not prescriptive, so you don’t really know if you have satisfied the law until some auditor tells you whether you have.
6. Existing systems are very fragmented, and typically include only a small minority of information such as prescriptions. But that data must still be brought in. So transitioning to a new system is very, very hard because all kinds of weird data must be imported. That transition has to be right; errors introduced during transition would be a huge legal risk.
I’m not sure it was even possible to satisfy all these constraints with technology until the very recent past. We now have much better technology for user interfaces, and better technology for transporting records around. But it still takes extreme architectural and design skills to create a system that can incorporate entirely new types of data and rules by clinicians without the involvement of a programmer.
That means in particular that this can’t be a big government job. The IRS and their four billion dollar debacle showed the problems government has with creating large systems. I simply don’t think any government effort could attract and keep the talent needed for this task.
Even private entities in the healthcare world have trouble with complex systems. HCA Healthcare attempted to write a next-generation patient accounting system, and wrote off some $130 million, and I’m pretty sure the actual amount of money spent (on a system that was thrown away) was much higher than that.
It’s easy for liberals, and even some “compassionate conservatives” to see the opportunity for saving lives and saving money, and just want to pass a law to make it happen. I don’t think I have to tell the people who frequent this site why that’s a bad idea. We could end up wasting tons of money.
But there’s a potential outcome that’s even worse. If an inadequate, buggy, brute force, low tech system were rushed into being, and its use was mandated, that would block the adoption of an innovative, more modern, better system that could be developed later. We would effectively be frozen into using that system, just as the air-traffic control system was frozen on old, obsolete technology for decades.
To sum things up, there is enormous opportunity to improve healthcare by applying technology to clinical patient data. But it’s a huge challenge too. And the more government tries to push it before it’s ready, or to command it into being the more likely that the potential won’t be reached.