October 5th, 2011
Does Intuition Lead to Bad Medical Decisions?
Discussing how medical practitioners use intuition and cognitive shortcuts (heuristics) to make decisions can elicit strong reactions. Some people heartily agree that reflecting on their use is informative and helpful; others believe that to entertain this topic is to condone sloppy thinking and to renounce rationality and hard science. These critics are concerned that heuristic shortcuts are quick and dirty, favoring speed over accuracy. But research in cognitive psychology suggests otherwise.
Take, for example, a study in which Dawes and Corrigan compared the accuracy of the tallying heuristic (discussed in my previous post) with that of statistical linear regression. The investigators’ conclusion: For most practical decisions, a sum of unweighted variables beats models that use regression equations (Psychological Bulletin 1974; 81:95). Specifically, they found that linear regression can “over-fit” the data, resulting in perfect explanations of the past but imperfect predictions of the future. So, simply tallying in our heads is often better than making complicated calculations. As Dawes and Corrigan wrote, “The whole trick is to decide what variables to look at and then to know how to add.”
Herbert Simon coined the term “satisficing” to describe how we modify our decision making for speed and simplicity. Rather than performing exhaustive searches for the best answer to a problem, we stop when we reach a sufficiently adequate solution. A medical example of this heuristic is a mental process called early-hypothesis generation. We know from the work of Arthur Elstein that expert physicians generate 3 to 5 hypotheses very early in the evaluation of a complicated patient. They learn through experience that early-hypothesis generation improves the speed and accuracy of such an evaluation: It leads to targeted questioning and testing that reduces the random effects of a broader “shotgun” approach.
Other critics of intuitive reasoning say that we should rely only on deductive reasoning and evidence-based medicine — that we should simply follow the scientific rules. Unfortunately, that is usually not possible. Tricoci and colleagues recently examined the 16 current ACC/AHA guidelines and found that only 275 (10%) of the 2711 recommendations have the level of evidence (Ia or IIIa) that would enable the use of deductive logic (JAMA 2009; 301:831); another 26% are just instructions based on expert opinion (Ic or IIIc); the remaining 64% (the rest of the level designations) require judgment and reasoning. Furthermore, guideline recommendations pertain to only a fraction of medical practitioners’ daily decisions. In general, psychologists estimate that 9 out of 10 decisions are made with intuition — a figure consistent with the data in Tricoci et al.’s analysis of cardiology guidelines.
Whether heuristics are helpful or flawed has been a subject of hot debate in the psychology literature for the past several decades. Just like more “rational” modes of thinking, heuristics have their advantages and limitations (see my previous post for a discussion of the work of Gerd Gigerenzer and of Daniel Kahneman and Amos Tversky, respectively). The bottom line: Heuristics are here to stay because they help us effectively manage the uncertainty that is inescapable in medical practice.
I doubt that “rational” analytical methods and computers will someday replace our use of intuition and heuristics. After all, there is still no computer program that can reliably pick stocks or predict economic downturns. Computers remember things better than we do and can aid in decision making, but overall they don’t reason any better than the human mind. Practicing medicine involves science and common sense. To improve our thinking, we need a common understanding of how we use common sense. We must continue to advance science, but we also need to give greater attention to how we use medical reasoning to apply the science in daily practice.
Where do you stand on the question of how we do — and how we should — make decisions in our daily practice?