HIV and ID Observations

Electronic medical records (EMRs) are much on my mind, as last week at Medical Grand Rounds Robert (Bob) Wachter, chief of the medical service at UCSF, gave a brilliant talk on the unanticipated consequences of our move towards what he calls the “Digital Doctor.”

Bob has thought a lot about this issue, so much so that he’s about to publish a book on the topic. In his talk, after a brief history of how we got to where the vast majority of U.S. physicians use EMRs, he focused on three main consequences:

1. The fact that doctors now interact as much (if not more) with screens as they do with patients — the “iPatient” phenomenon. The no eye contact problem. The lack of doctors on the medical wards, as we gravitate toward “work rooms” full of computers. You know how pediatricians sometimes get drawings from their school-age patients that include the doctor? He showed a remarkable example, in crayon of course, of a doctor facing away from the artist (the child), the MD staring at a computer screen and typing. From the book: “I’m guessing this one didn’t make it onto the doctor’s Wall of Fame.”
2. The loss of interaction between doctors when the data are digital rather than something you can hold. Remember that brilliant radiologist who used to go over all chest films on your medical team? Now a radiologist may be reviewing films at home overnight, or in India, reports filed digitally and not requiring any human-to-human contact with the ordering doctor. Radiology rounds are slowly disappearing, along with the time for clinicians to pause — and think collectively — about what the images mean.
3. The potential for automated systems to amplify medical errors. We’ve grown increasingly reliant on computers to help with decisions, for better and worse. In a taut, complex story involving a series of increasingly unlikely errors, he described how a child received a massive overdose of medication during hospitalization — all the indirect result of how a poorly designed systems can usurp clinician autonomy.

What he didn’t have time to cover (but does so in the book — he shared the excerpt with me), is the powerful effect EMRs have had on clinical notes.

It’s a fact that the note as means of communicating how the patient is doing has all but been destroyed. Notes even from the best clinicians routinely have the following features:

• A massive amount of repetition. Cut/paste phenomenon #1.
• “Required” elements that serve no clinical purpose. How useful is a lengthy review of systems? And isn’t a history-directed, targeted physical examination of far greater value than a comprehensive one “done” merely to meet higher billing criteria?
• Giant chunks of computer-generated data. Cut/paste phenomenon #2. It’s mostly lab and imaging results, with no interpretation of what the data mean.
• Factual errors. Cut/paste phenomenon #3. In the ambulatory record, one of my favorites is that some children never age:  “Has three children, a son age 10, daughters ages 8 and 1” — which is then written unchanged in the social history over the next five years. Reminds me of The Simpsons — Bart, Lisa, and Maggie never age either. On the inpatients, we routinely see this: “ID consulted, considering pneumonia, UTI, C diff, disseminated fungal infection as cause for fevers” — then these same words are repeated for many days after some or all of these diagnoses have been ruled out.
• Sentences whose sole purpose is to avoid getting sued. You know ’em when you see ’em. They sound defensive, are depressing to read, and communicate no useful clinical information.
• Boilerplate text of highly dubious relevance to the individual case. During a mandatory “compliance” review of my notes (shudder — is there anything in modern medicine more painful?), I had someone suggest I add the following phrase to all of my notes: “More than 50% of this 30-minute visit was spent counseling the patient on the chronic nature of his/her condition, the rationale behind the laboratory tests ordered, the importance of taking medications directed, and the directions for making follow-up visits. Contact information provided, and patient’s questions answered.” The rationale? “You don’t do a procedure, so you need to improve the documentation of what you’re doing with your time.” Lovely.

The genesis of this problem, of course, is that the medical note is trying to do too many things at once. Previously a way of summarizing the clinical course of the patient, both for our own individual use and to communicate with other clinicians, it now has other masters with different motivations. Facilitated by EMRs, the note has subsequently evolved into a Jackson Pollock-like canvas of disjointed text, much of it of marginal or no clinical significance, with sections held together only loosely by the name and medical record number at the top of the page or screen.

Here’s a solution that will never happen — let’s have the medical note evolve even further, breaking it down into distinct sections based on their primary purpose. Imagine three tabs on the top of the note; you get to read only the one you want or need:

1. Clinicians, here’s your section — it includes the stuff you really want to know, such as the history, exam, and lab/imaging results that matter (not all the labs/imaging, thank you), plus what the clinician writing the note thinks is going on, and what he/she plans to do.
2. Billing compliance folks, read this part — it will have the required review of symptoms (most of them irrelevant), lengthy rubber-stamp documentation of counseling and education, and whatever other parts are required by whatever payor this patient has. And it will be inserted there by someone who’s not a doctor — or even better, by some automated bot — because successfully generating this kind of documentation is not why we went to medical school.
3. Medicolegal guys, this is for you — lots of defensive phrases here, none of them of any clinical relevance, but they’re here just in case something untoward happens and the case ends up in court.

Have fun.