The Skeptical Convert 2/4/13

If you were my patient and I mentioned to you that it wasn’t until recently that I found out that blood circulation was how you get oxygen to your body parts instead of absorbing it from your skin, what would you think of me? I think you’d politely excuse yourself and leave. Because although I specialize in gastroenterology, I as a doctor am supposed to have an understanding of how the whole body works. I’m not supposed to see it as some mysterious black box that I had to learn to deal with by rote.

When I talk to a cardiologist about a case, he may not go as far into physiologic details with me as he would with another heart specialist, but he will refer to general principles that we all learned earlier in our education and training, to orient me to at a reasonable level of understanding about what is going on and what needs to be done.

Right now though this concept doesn’t seem to apply much in the medical computing world.  By way of an example, I direct you to a study and editorial in the January 15 Annals of Internal Medicine. The original study looked at Meaningful Use measurements in practice by going back over the actual records.

The authors documented a statistically “wide measure-by-measure variation in accuracy” that “threatens the validity of electronic reporting.” I know, that’s no big surprise to any regular reader at this site–file it under “Department of Duh.”

The accompanying editorial caught my attention, though.  It was written by a distinguished general internist, trained at top institutions and a university medical faculty member. She wrote very well, and with a knowledgeable authoritative tone, about the problems with getting statistically valid data out of multiple sources, users, and formats.  

Right in a middle paragraph, after a general comment about about how variable use of the EHR by different providers increases “measurement noise”, she noted a striking personal example, and I quote: ”In my own practice, I learned that my lower rates of blood pressure control reflected the fact that I was documenting the patient’s blood pressure in free text rather than using an available structured field.” And then back to the general subject.

Wow. It seems to me that that deserved more discussion. OK, maybe she didn’t know they were tracking blood pressure in the first place. Maybe she assumed the system had the ability to capture that information from a text entry by some sort of NLP process. I’d like to know that, but I’d also like to know if she understood at that point about these things called databases underneath applications — that they store different categories of data, that they treat numeric data differently from text, and how numeric data generally needs to go into structured fields for that to happen.

Because I can tell you, from lots of personal conversations I’ve had, that whether she did understand those basic concepts or not, plenty of other medical practitioners don’t. That was worth discussing at greater length, whichever of those theories or combinations are true or false.

Why? Because if medical practitioners, as users, are going to see HIT as an alien world only approachable by rote training, they’re going to fall into potholes like this all the time, and I see it happening a lot.  There are a lot of lousy EMR designs out there, and a lot of mediocre training, but I can’t help but think that at least some of the problems with usability stem from gaps in basic user comprehension of the bigger picture.

David Brooks said it well the other day. “Change is hard because people don’t only think on the surface level. Deep down, people have mental maps of reality — embedded sets of assumptions, narratives, and terms that organize thinking.”  

That’s what happens when I’m talking to the cardiologist. Deep down, we have a common map of reality in our heads. That’s how we organize things in our minds and how we think. We’re here in the first place because that’s what we’re supposed to be able to do.

I read a lot of naysaying on this site about the computerization of medical practice, written as if it could all go away.  It’s not going to, but what we have right now isn’t working very well. Part of the solution will be for the mental maps of HIT people and physicians to match up better. The physicians do need to accept that their mental maps are going to need some revision. The IT people need to realize that we need explaining to get the training to sink in.  

Robert D. Lafsky, MD is a gastroenterologist and internist in Lansdowne, VA.