Tuesday, January 23, 2007

Heuristics invaluable but need to be understood and monitored

I was impressed by and wrote about an article by Dr. Croskerry who has studied how physicians think and how they can think better diagnostically when they understand their basic human tendency to think by using ad hoc rules of thumb, cognitive shortcuts or heuristics. DB's MEDICAL RANTS recently (jan. 22, 2007) referenced a great article by Jerome Groopman entitled "What's the Trouble?" which draws upon Coskerry's insights.

A more detailed treatment of these thoughts can be found here in an 2005 article by Croskerry.

I came across another related article entitled "Diagnosing Diagnostic Mistakes" which also talks about diagnostic "errors" and make the interesting and , I think, valid, point that everything called an error might not be.

Dr. Clement J. McDonald authored an excellent article over ten years ago in the Annals of Internal Medicine entitled Medical Heuristics,The silent Adjudicators of Clinical Practice",
It is available in full text on line and well worth reading.

McDonald talks about common medical heuristics some of which have been codified into aphorisms, For example Occams razor which advises to choose the simplest hypothesis (or a single disease process)to explain a set of observations.Contrasted to this is the less well know Hickam's Dictum which says a patient can have as many diagnoses as he darn well wants. Another is " treat the patient not the numbers. This adage which was quasi-dogma in my house officer days has been diluted by the development of various lines of evidence that we should in fact treat the numbers, at least when we are considering blood sugar in diabetes and blood pressure values and- at least for secondary prevention of coronary artery disease-the cholesterol level.

A more recent article by Dr. Donald Redelmeier also appears in the Annals, "The Cognitive Psychology of Missed Diagnoses"

He speaks of common heuristics: The availability heuristic-we judge the likelihood of a case by how easily examples spring to mind. The anchoring heuristic-we tend to stick with our initial impression ( this runs counter to the well established scientific principle of checking for evidence that would disprove our working hypothesis).This is also called premature closure.

Humans may well have biologically ancient neurological mechanisms that organize incomplete data into a sensible whole ( our visual system seems to do a great job of that) and gives us an reassuring sense of control and an illusion of a more accurate and more complete understanding of a given situation that we sometimes really have. These and other heuristics have survived perhaps because they do work reasonably well much of time but the missed diagnoses are reminders that much of the time is not all of the time and hopefully awareness of their existence and their potential fallibility will enable us to mentally over ride them when appropriate. This could be a simple as reconsidering the diagnosis as more clinical data becomes available.

No comments: