View as “PDF version”
Cognitive biases unfortunately often impact our diagnostic and therapeutic decisions in a negative manner. One such bias coloring our thinking is the “availability” bias (also known as the “recency” bias). This is where the most recent or most memorable cases from the past narrow our thinking about a current patient. We discussed this (and a variety of other biases) in our August 12, 2008 Patient Safety Tip of the Week “Jerome Groopman’s “How Doctors Think”We all know how a previous bad experience with use of a medication may influence us not to use it again, even when we know the medical evidence tells us we should use it (one of the reasons so many patients with atrial fibrillation are never placed on coumadin). We often encountered this when trying to get surgeons to use DVT prophylaxis. Some would remember vividly one of their patients who had a hemorrhage when on DVT prophylaxis, so they would be reluctant to use DVT prophylaxis again.
Heuristics are simplified decision tools or shortcuts that are sometimes used in more complex decision-making scenarios. One researcher recently looked to see if decisions in the delivery room were influenced by such heuristics (Singh 2021). Singh explains that this behavior can be loosely predicted by a “win-stay/lose-shift” heuristic, according to which the decision-maker either switches strategies if the last outcome was a “loss” or continues with the same strategy if the last outcome was a “win.”
Singh examined records of over 86,000 deliveries from an urban and a suburban academic hospital over a 20-year period to see whether complications in the prior patient’s delivery mode (whether a cesarean or vaginal delivery) make the physician more likely to switch to the other delivery mode for their next patient, and the effect of this heuristic on the patient for whom it is used. The analysis suggested that, if the prior patient had complications in one delivery mode, the physician will be more likely to switch to the other—and likely inappropriate—delivery mode for the subsequent patient, regardless of patient indications. Moreover, there was evidence that this heuristic has small, suboptimal effects on patient health.
Singh was able to identify delivery mode and L&D (labor & delivery) complications using ICD (International Classification of Diseases) procedure and diagnosis codes in a large electronic database. The data showed that the probability of switching delivery modes increased as the number of complications in the prior delivery increases. When there are no prior complications, physicians are more likely to “stay” with the delivery mode they used for the prior patient. And, if the preceding delivery had complications, the odds of the physician switching to the other delivery mode increased.
Though it is difficult to assess from administrative data whether a delivery mode is appropriate or not, Singh does attempt to show that some of these delivery mode decisions were likely inappropriate for patients. According to Singh (Singh 2021b), complications during a vaginal delivery increased the likelihood of a subsequent C-section by up to 3.6%. That was about 23 potentially inappropriate C-sections per year per hospital studied. Complications during a cesarean increased the likelihood of a subsequent vaginal delivery by up to 3.4%. That’s about 50 potentially inappropriate vaginal deliveries per year per hospital studied.
In the editorial accompanying the Singh study, Li and Colby (Li 2021) emphasize that she found that “more experienced physicians use this decision rule more often, even though physicians who rely more on this rule have worse patient outcomes over time. The result demonstrates that the use of heuristics or simplified decision rules is a common human tendency even among smart, well-intentioned, highly trained doctors.”
Li and Colby also point out three well-established cognitive biases as potential causes of this maladaptive switching of delivery modes, whether individually or in concert: recency, affect heuristic, and confirmation bias. We’ve already noted the recency (or availability) bias above. That is where the most recent or most memorable cases from the past narrow our thinking about a current patient. The “affect heuristic” notes that people rely on their affective response to gauge how large a risk is and that a highly emotional response to a previous delivery complication would lead a physician to overestimate the risk in a subsequent delivery. Thirdly, our frequent nemesis – confirmation bias – may also come into play. They note that a physician may inadvertently seek and interpret evidence in a way that is consistent with his/her fears or concerns about a potential complication, which helps them feel comfortable that modality switching is appropriate for the current patient.
Singh gives another common example of this “win-stay/lose-shift” heuristic – a physician may be reluctant to prescribe a certain drug if a previous patient suffered an adverse event related to that drug (Singh 2021b). She notes that at least 2 interventions we use (clinical decision support algorithms and nudges) may help reduce use of this heuristic. The algorithms in clinical decision support systems suggest what might be best for a particular patient. The “nudge” example is placing a preferred drug at the top of a drop-down list and placing a drug that might best be avoided near the bottom of that drop-down list.
Our own experience would fit with that in Singh’s study. For years, hospitals have tracked VBAC (vaginal birth after cesarian) rates. The push was to get more VBAC’s done and reduce the C-section rates. We would see a cyclicality to VBAC rates. There would be a trend toward more VBAC’s. Then, after an obstetrician encountered a ruptured uterus in a patient with a planned VBAC (or even heard of such a case), the VBAC rates would plummet again.
The Singh study is one of few that have actually studied real-life use of heuristics in medical decision making by physicians. A systematic review on cognitive biases and heuristics in medical decision making (Blumenthal-Barby 2015) concluded that most of the studies on biases and heuristics in medical decision making are based on hypothetical vignettes, raising concerns about applicability of these findings to actual decision making. It also concluded that biases and heuristics have been underinvestigated in medical personnel compared with patients.
As physicians, we rarely are aware of cognitive biases underlying our medical decisions. And we also typically are not aware of when we are using a heuristic that may be inappropriate. And our patients are often subject to similar cognitive biases and inappropriate heuristics in their own medical thinking. One wonders if these phenomena have increased in the “age of disinformation”. We agree that having more real-life examples of maladaptive use of these would be helpful.
Some of our prior columns on diagnostic error and cognitive biases:
Singh M. Heuristics in the delivery room. Science 2021; 374(6565): 324-329
Singh M. People use mental shortcuts to make difficult decisions – even highly trained doctors delivering babies. The Conversation 2021; October 14, 2021
Li M, Colby H. Physicians’ flawed heuristics in the delivery room. Science 2021; 374(6565): 260-261
Blumenthal-Barby JS, Krieger H. Cognitive Biases and Heuristics in Medical Decision Making: A Critical Review Using a Systematic Search Strategy. Medical Decision Making 2015; 35(4): 539-557
Print “PDF version”