I woke up on Monday morning with a 2 inch diameter bright red rash on the inside of my right ankle. It was mildly pruritic (itchy). I could not recall any exposure to insects or trauma of any kind and it did not appear to be infected, so I applied some topical corticosteroid cream and went to work. That night at home the rash seemed very mildly improved but it still itched. I decided to get some medical input at that point. The usual choices in my area are the Emergency Department or Urgent Care, but recently my health plan started to offer online consultation through a combination of limited diagnoses and procedures, an algorithmic set of questions, the ability to upload images, and consultation with a nurse practitioner. I looked at the list of conditions they were set up to diagnose and treat, noted that "rash" was one of them and logged on.
Health care IT is still in its infancy so nobody should be surprised that it took me much longer than expected to log in to the appropriate interface. At first the program suggested I could just use my existing login and that would also integrate previous test results and conditions into the current evaluation. After needing to call them I established a separate login and password for this episode. Rather than the expected details up front, the program started to ask me all of the usual questions about the rash. There were 28 screens in all, including some that forced an answer. That question was "What do you think is causing the rash?". Possible answers were: insect bite, infection, allergy exposure, poison ivy, etc. There was nothing on that list that seemed likely. That was after all the reason I was calling in. I could not proceed past that point without giving an answer so I clicked "insect bite". After completing 28 screens there was a text field and I entered: "Even though I answered "insect bite" on question #8, I only did that because I could not proceed if I did not provide an answer."
Next came the expected demographic data. I live in a town that the U.S. Postal Service never gets right. If I list a Zip Code the wrong town name pops up. This software was no exception. It took me extra time to enter and reenter data that was already there somewhere in my healthcare company's database. The final screen was the billing and financial data including credit card information. More data that my healthcare company has know for the last five years. At this point I am about 20 minutes into the process and it is time to upload the photos. I had 4 photos of the ankle and the program accepted 3 of them. Sign off occurred at the 25-30 minutes mark. As I waited for the return e-mail or call, I marvelled at how health care companies have transferred all of this clerical work to physicians over the last 20 years and now they are transferring it to the patient. I just did the work of the intake person and financial person in any clinic or hospital.
In 20 minutes I got a call from the nurse practitioner. She said that although it was clear that I had a rash, it was not a rash they could diagnose in the system. I told her that I was applying a potent corticosteroid and she said to just keep doing that but to go into a primary care clinic and get it checked out by my primary care physician. Within 2 minutes, I got an e-mail from them:
Thanks again for taking the time to talk with us on the phone. Your health and safety is our top priority. Based on the information you shared with us, we think that an in-person visit is the best way to handle this specific condition. And, please know that you will not be charged for your visit today.
We're sorry we couldn't help you this time, but please keep us in mind the next time you're feeling ill. Thanks for choosing us.
Good to know I guess, but no diagnosis or specific treatment plan. I continued the corticosteroid and the next night after work I stopped into an urgent care clinic after work. I saw a family medicine physician who inspected and palpated the rash, took my pulses and determined that they were good in the area, and asked me clusters of questions that were clearly designed to rule in/out various pathological processes. His conclusion: "Well it's not an infection and its not due to trauma, but it clearly is an inflammatory process like atopic dermatitis. So at this point I would keep applying the corticosteroid." He asked me for questions. My mind was preoccupied with tales of devastating spider bites lately so I blurted out: "This does not in any way look like a brown recluse spider bite does it?" He laughed and said: "Absolutely not."
So what have I learned from all of this and how do those lessons apply to psychiatry? First off, it appears that human diagnosticians are safe for now. Keep in mind that the system is set up to diagnose and treat a restricted list of conditions that are considered to be the least complicated in medicine. Second, the human diagnostician's superior capabilities depend on pattern matching and that in turn depends on experience. It reminded me of a course I taught for 15 years on how to avoid diagnostic errors and pattern matching was a big part of that. The two examples were rashes and diabetic retinopathy. Dermatologists were much faster and much more accurate in classifying rashes from pictures than family physicians. Ophthalmologists are much more accurate using indirect ophthalmoscopy than family physicians using direct ophthalmoscopy in diagnosing diabetic proliferative retinopathy. In fact, the family physicians were slightly better than chance.
The lessons for psychiatry are two fold. Remember the idea of a restricted list of conditions that are not considered complex? It turns out that depression and anxiety are on that list. Even though there is no call center where you can call and complete the paperwork like I did, it would probably not be much of a stretch to say that many if not most primary care clinic diagnoses of depression and anxiety are keyed to some rating scale. Like the studies of Dermatologists and Ophthalmologists, there are no expert pattern matchers looking at the patient. That can result in a diagnosis that is essentially dialed in.
The second aspect here is the design of the algorithm and its implications. My rash algorithm had a forced choice paradigm. I could not proceed to the end unless I picked an answer that was clearly wrong. That is the way it was set up. That is the problem with so-called "measurement based" care. There is the appearance of a quantitative result. The Joint Commission called the 10-point pain scale "quantitative" in the year 2000 with their pain treatment initiative in the year 2000. I have spent a good deal of my adult life talking with patients about their moods, sleep and appetite patterns, and other symptoms. The most important part of my job is coming up with a plausible scenario for their current distress. I can say without a doubt that over half of the people I see cannot describe discrete episodes of mania or depression. The usual description of depression I get is that it is life long with no remissions. Certain personality characteristics predict descriptions of symptom severity in the initial interview. Some people completely minimize symptoms and other people will flat out tell me that they do not want to discuss their inner thoughts even if they are experiencing thoughts that may place them in danger. Map those response patterns onto a psychiatrist and hopefully that will result in a diagnostic formulation and a plan to deal with the nuances. Map those response patterns onto a PHQ-9 and suddenly you have a number that somebody believes has meaning. Looking only at Question 9:
"Thoughts that you would be better off dead or of hurting yourself in some way."
Suddenly people are alarmed with the person with a personality disorder and chronic suicidal thinking or chronic obsessions involving suicidal thinking endorses "nearly every day" as their response. We are falsely reassured when the patient who has a significant personality change and depression endorses "not at all". We have forced them to make a choice and they have, rather than using all of the information necessary to make an evaluation.
As a discipline - we should be moving in the direction of using all of the relevant information in clinical situations and not less. My rash today is an example of what can happen in an organ governed by much less genetic, metabolic and signalling information than the human brain. Even in that situation a diagnosis with no clear etiology or diagnostic features can present itself.
Forcing choices reduces the information flow rather than facilitating it. If primary care physicians find this checklist approach to diagnosing anxiety and depression useful I would see no problem with that, but it might be useful to look at the medications being used based on the PHQ-9 and the kind of impact this approach is having on medication utilization. It also might be useful to have a seminar or two on the problem of over prescribing medications. The correlation between overprescribing opioids and the use of a "quantitative" scale to measure everyone's pain is undeniable.
The question that applies in all of these circumstances is whether a number on a subjective rating scale is ever enough of a reason to prescribe a medication.
You already know what I have to say about that.
George Dawson, MD, DFAPA
