Wednesday, November 30, 2011

CPOE So Easy A Caveman Surgeon Could Use It!

Whilst cavorting with old friends at lifeIMAGE (see the previous post), I had the chance to connect with another old friend, Kang, formerly one of the technical gurus of AMICAS, now CEO of his own little company, HealthFortis. (He's working with another AMICAS alum, Dmitry, whom I sadly didn't get to see this trip.)

Kang, having created some wonderful stuff AMICAS over the years, needed a new challenge, and he picked a big one: CPOE, aka Computerized Physician Order Entry. (Some say "Provider" instead of Physician, which clearly tells us physicians that we are no longer held in particularly high regard, but whatever.) CPOE is clearly a tough nut to crack, as one must create software that physicians (even surgeons and, yes, even orthopedic surgeons) will have to use to order stuff for their patients. Keep in mind, these are guys (and gals) who are used to scribbling something illegible on a piece of paper over the course of three seconds, and then faxing or maybe throwing it (literally) at someone with the full expectation that their intentions will be telegraphed magically. Of course, sometimes they will lower themselves to simply barking said orders at someone, in person or by phone, with the same somewhat unrealistic expectations of completion.

Now that EMR's and such have taken over, computerizing this process in the form of CPOE is felt desirable, and even necessary. But no one asked the physicians about this, and therein lies the path to big trouble.

Our largest hospital system instituted Cerner Millenium CPOE over the past few years, and the physicians to a man (and woman) seem to hate it with a passion. I'm on the CPOE committee for one of our other hospital systems, and we are struggling with the joys of trying to crowbar a entry process into the ancient legacy MediTech Magic program, you know, the one that ports a 1980's green Data General window to Windows. I'd rather be the bagel delivery-boy for the Gaza district.

Kang nicely outlines what is wrong with pretty much every CPOE product out there: it tries to make docs do things differently, and, trust me, docs do not want to do things differently. Why are we forcing them to take 5-20 minutes and 59 mouse-clicks to accomplish what they once did with a piece of paper and a pencil in 3 seconds? The intelligent approach is to first make it easy for the physician to use CPOE, and then leverage all the nice things that an electronic approach can deliver. This, of course, includes "Decision Support at the Point of Decision." Brilliant!

HealthFortis takes a very simple approach. There are just a few points of entry, but they spawn everything appropriate to taking the order, and all of the entry boxes allow for free-text, much like Google, with a list of possible entries building and then narrowing as you type. To find your patient, you might simply type Do Da, which would bring up Doctor Dalai, among other less interesting people. Select the patient and then a simple window appears, wherein you enter the diagnosis and, if you wish at that point, the exam to be done. Here's where the magic starts.  Suppose you enter "AA".  The program gives a few possibilities, such as "AAA", which we then select. You are then given a list of possible exams, ranked by ACR recommendation codes from 9 (good choice) to 1 (you have to be kidding!). Clicking the exam you want spawns an order in HL-7 to be delivered back to your HIS/EMR.  If the condition/symptom isn't quite so specific, the program brings up more data and options to help you decide. Of course, you can still override this and forge on ahead with an arteriogram for little toe pain, but you will definately get a "1" for that choice, and you will be told just why that is inappropriate.

There is included some nice stuff like searches for recent orders of the same type (did you really want to repeat the CT for the 10th time this month?)

Right now, the system is in its infancy, having been online for only a few months, but it is growing, and I'll predict there will be rather wide-spread acceptance. The order-sets for the moment include more radiology exams than anything else, at least as I understand it. Kang did outline a heuristic learning function, which will help grow the database; as more and more entry-pairs are collected, the system will learn which are being used most frequently, and make them more easily available.

This is one of those offerings that is elegant in its simplicity and usability. I'm not sure it will be ready for full hospital use in time to derail MediTech, but it possibly could be deployed at least to physicians ordering stuff from outside the hospital. I'll take that for now.

Make it easy and they will come. Guaranteed.

1 comment:

Anonymous said...

Very good observations..