Philadelphia Reflections

Electronic Medical Records

The electronic medical record had a great flurry of excitement about 1980, and I was one of its earliest proponents. I wrote my own program in the Basic language to produce bills and insurance claims forms, and to serve as the basis for adding diagnoses, lab work, and prescriptions. It took about a year of my spare time, worked very well in my office, and is available for anyone who wants it. It had two flaws, and it still has two flaws. After a fruitless effort to simplify physician input, I abandoned that particular effort as taking too much of my time.

That's still the case, thirty years later. Programmers have a habit of telling the boss something can't be done when it is merely inconvenient to do it, and doctors absolutely will not tolerate doing something some lesser-paid person can do for them. Some variant of the Google search engine might suffice for coping with physician input, particularly if combined with Dragonfly voice recognition. Eventually, someone will conquer this beast, but it turns out to be harder than it looks, and nerdy doctors have turned their attention from data entry to Big Data. Meanwhile, programmers have avoided the task of simplifying something which could be simplified, just because it requires acknowledging that somebody else's time is worth more than their own. The time for excitement about data entry has passed, while the problem remains incompletely solved. If you were going to win a Nobel prize, it wasn't going to be for data entry perfection.

The second physician obstacle is that computers generate far more information than anyone has time to read. The white blood count is vital when appendicitis is being considered, but it just bulks up the chart thirty years later. So what is needed is an automatic summarization system, with hooks back to the original data in the rare instance where retrieval is needed. Since medical care is constantly changing, the summarization algorithm must change with it. It's a big job, and somehow billions of dollars have been expended trying to do something else the doctor never asked for. The new danger is that some malpractice lawyer will discover a point vital to his case has been omitted from the summarization (but not the bulk record) without the doctor's knowledge, even though it seems no longer vital to the treatment of the patient in the future. For a long time, I merely smiled when people told me the EMR was more trouble than it was worth. But recently I have read of the astounding amounts of money (thirty billion dollars have been mentioned) which have been devoted to this mess, much of which is pretty cute but nobody asked for it. Apparently, the doctors who advised the program let the big-shot millennials who actually wrote the code do the real directing because it seemed worthwhile to spend billions to accomplish their personal goal of payment by diagnosis instead of by procedure. Maybe that's how you subdue a bucking broncho, but the doctors work around you until you seem to be winning. And then they quit.

If someone had the bad judgment to put me in charge of this circus, I would immediately limit the archiving to data which can be automatically generated, and rest content with reports of lab and x-ray reports from this source. And then I would advertise for someone to produce workable physician data entry, as well as generate periodic automatic summarization. Until these two features pass approval by seasoned physicians, we would just have to get along with paper records. This is essentially what I told some meetings of 1980 enthusiasts. But it hasn't happened, yet, so the system progressively antagonizes a group they cannot command and cannot do without. Herding cats, I believe it is called.

Originally published: Monday, August 29, 2016; most-recently modified: Thursday, May 16, 2019