Algorithmanic episodes

Jerome Groopman has been all over NPR the last few days giving interviews about his new book, How Doctors Think. This morning he landed a seven-minute session on Morning Edition, and recently spoke at length with Terry Gross on Fresh Air. The book sounds like a well-constructed inquiry into the algorithmic processes of doctor think. How do physicians make sense of patients' stories and arrive at a diagnosis?

Or a mis-diagnosis. Groopman discusses the experiences of physicians and patients--even his own experience as a patient, to point out where doctor think goes wrong. The diagnostic process, relies on information from the patient about symptoms, personal and family history, risk factors. Doctors then collect relevant "objective" data (physical exam, labs, studies), and assemble a list, or differential diagnosis we think might explain what the patient is experiencing. Ideally, it's an exhaustive list, and we narrow it with directed studies to rule things in or rule them out. We work hard at it, but for various reasons we don't always get it right, and the consequences can be severe. Dr. Groopman explains to Steve Inskeep on Morning Edition:

"Usually doctors are right, but conservatively about 15 percent of all people are misdiagnosed. Some experts think it's as high as 20 to 25 percent. And in half of those cases, there is serious injury or even death to the patient."

Why do you think that doctors would be wrong that often?

Well, you know, it's very hard to be a doctor. We're working under tremendous time pressure, especially in the current medical system. But the reasons we are wrong are not related to technical mistakes, like someone putting the wrong name on an X-ray or mixing up a blood specimen in the lab. Nor is it really ignorance about what the actual disease is. We make misdiagnoses because we make errors in thinking.

We use shortcuts. Most doctors, within the first 18 seconds of seeing a patient, will interrupt him telling his story and also generate an idea in his mind [of] what's wrong. And too often, we make what's called an anchoring mistake — we fix on that snap judgment.

An anchoring mistake? Sounds like what my people call early closure, a very common problem. Making big differentials is hard, and many things influence how quickly we decide we know the diagnosis. Often, it just seems to fit...mostly...more than other stuff. But is our list comprehensive? Did we consider only common diagnoses? Is is time to trot out the rare, eponymous syndromes we memorized in med school but have never seen? Deciding early to attach a diagnosis to a symptom feels good, especially when there are other patients with symptoms waiting to be seen.

There's something about categorization that appeals to me. Calling something anchoring mistake, when you could just as easily call it laziness, gives me a kind of hope. If I miss a diagnosis and because I'm lazy and the remedy is to "try harder," well, I'm screwed. The problem and the solution are both pretty vague, and my fear around messing up again takes over. Fear motivates, for sure, but I would argue the outcomes aren't so great.

If I can categorize my apparent laziness with nifty terms like anchoring mistake (in which I make a snap judgement) or representativeness error (wherein I tell myself that common things occur commonly), my type-A doctor mind has something productive and familiar to do: memorize a list. In this case, knowing the ways my thinking might not serve me well helps me see potential pitfalls in diagnosing and treating--also known as helping--patients.

I haven't read the book yet, but I will. I did, however read Dr. Groopman's article from January 29, 2007 in the New Yorker. The writing is excellent and the examples compelling. If the book is anything like the article, I'll be reading all weekend.