Diminishing diagnostic cognitive bias: Anchors away

By Warren R. Heymann, MD, FAAD
Sept. 27, 2023
Vol. 5, No. 39

We have all been there, probably daily. You’re an hour behind schedule, and the patient in front of you has a laundry list of complaints demanding that you address them. You shift to autopilot, using your keen observational acumen and management skills, as you glide through your diagnoses —

“Mrs. Jones, that scaly plaque on your elbow is psoriasis and it should respond well to betamethasone ointment; the brown growth on your back is a seborrheic keratosis — it’s benign and, if you wish, it can be removed as a cosmetic procedure; the yellow bumps on your forehead are called sebaceous hyperplasia — these too are benign and can be treated cosmetically; the scale between your toes is tinea pedis, aka athlete’s foot, and should respond well to ketoconazole cream; the brown patches on your back are tinea versicolor, a different type of fungus which is a yeast living in our hair follicles, but that too will get better with the ketoconazole cream I gave you for your feet; the brown discoloration on your neck is called acanthosis nigricans, possibly related to your recent weight gain that you mentioned. If you can get that weight off, and use a mild tretinoin, it may improve. I think we have covered your list. Do you have any questions?”

“Yes, doctor, I forgot to mention that I’m losing hair!” (I try to be accommodating but “oh, by the way” alopecia mandates a separate visit.)

I am never satisfied after such encounters (the patient probably is not either), unsettled by working so quickly that critical findings and conclusions could have been overlooked.

In a previous DWI&I commentary devoted to optimizing the patient encounter, Dr. Daniel Kahneman’s constructs for two major thinking patterns were discussed — fast thinking (System 1, a gut reaction, habit, “expert” thinking) and slow thinking (System 2, which is more conscious and controlled). “As dermatologists, we are trained to use System 1 most of the time, however, that might not lead to the right conclusion. Nor does reliance on System 2. The reality is that the combination of the two, while being cognizant of our thought processes regarding how we reach conclusions, is likely to provide us with valuable insights.” (1)

Cognitive biases (CBs) are cerebral shortcuts that simplify decision-making. (2) Diagnostic errors occur in at least 5% of the outpatient department and emergency room cases, leading to fatalities in approximately one in every 1,000 cases, resulting in an estimated 40,000-120,000 deaths annually in the United States. (3)

CBs exist at every stage of the patient-physician encounter, from taking a history, to performing a physical examination, generating a differential diagnosis, and in managing the patient. Yesudian and Yesudian developed a new model for categorizing CBs in dermatology, building on metacognition — the art of self-reflection on one’s thought processes. They describe 33 CBs, defining each, while offering “potential metacognition questions” accompanying each CB. (4) What is frightful is that I have experienced all 33, several repeatedly.

My two primary mea culpas are anchoring bias and confirmation bias. Anchoring bias refers to prioritizing information and data that support an initial impression, even when first impressions are wrong. Confirmation bias is the selective gathering and interpretation of evidence consistent with current beliefs while neglecting contradictory evidence. (5) As a non-diagnostic example — I cherish coffee. In actuality, medical evidence touts many benefits. Regardless, from my biased perspective, any study refuting such benefits is automatically classified as flawed!

Jia and Lester present the case of a 56-year-old man with Fitzpatrick phototype VI who ultimately succumbed to a marginal zone lymphoma. In the year before his death, he received multiple intralesional triamcinolone injections for a nodule on his chest for a presumed keloid, despite the continued growth of the nodule. Ultimately a biopsy of the nodule was performed, demonstrating his lymphoma. The authors correctly state “This case illustrates the importance of integrating new or conflicting clinical information and how it should trigger reevaluation of previous diagnoses or treatment plans.” (6)

In a study of 130 physicians responding to a self-reflection questionnaire about their most memorable diagnostic error, the three most common CBs were anchoring (60%), premature closure (58.5%), and availability bias [cases that readily come to mind] (46.2%) (3). Correlating experience with improved diagnostic skills, emergency room faculty have a greater propensity toward anchoring bias than residents. (7)

Is it possible to conquer CBs completely? Short of artificial intelligence, probably not. We are still obligated to try. As Sloan asserts, “Recognition of these biases is the first step in overcoming them.” (2) Debiasing strategies, based on cognitive psychology, include guided reflection interventions (which reinforce behaviors that reduce bias in complex situations) or cognitive forcing strategies (involving conscious consideration of alternative diagnoses). (5) Such techniques should be part of medical school and graduate medical education curricula. Additionally, structural changes in practice allowing enough time for cognition are essential. All practitioners need enough time to think — without that availability, System 1 thinking alone becomes the default.

Point to Remember: All dermatologists are subject to cognitive biases that could result in diagnostic and therapeutic errors. Embrace metacognition (self-analysis) as a pathway to determine how to be a more effective clinician.

Our expert’s viewpoint

Christine J. Ko, MD, FAAD
Professor of Dermatology and Pathology
Yale School of Medicine

More attention is being paid to the role of cognitive bias in dermatology, as highlighted by Dr. Heymann’s commentary and the cited literature. As stated by Dr. Heymann, in dual process theory, there are two major cognitive thinking patterns — System 1 and System 2, and importantly neither is better than the other. Cognitive bias is part and parcel of how human beings think, and such bias, unlike racism, sexism, ableism and other “-isms,” is not a negative. Cognitive bias is reinforced by our thinking patterns and cannot be completely eliminated, as long as we are thinking.

In other words, the overlearning of dermatologic diseases (recognition, management, and treatment) that is promoted in trainees and in a lifetime of medical practice is channeled to create expert, System 1 thinking. As System 1 thinking is also “hardly thinking” (imagine how easily a diagnosis of psoriasis pops into your mind), it is generally never a bad thing to supplement System 1 with System 2. We cannot always do so — if we thought twice about every decision we make in a given day or patient encounter, we would be highly inefficient and mentally overloaded. Nonetheless, being able to self-reflect, recognize diagnostic error, and promote error recovery (in which reasons for errors are sought out and used to prevent the same error in the future) will make us all better physicians.

1. Heymann WR. Optimizing the dermatologist-patient encounter. Dermatology World Insights and Inquiries. October 5, 2022, vol. 4 no. 4. https://www.aad.org/dw/dw-insights-and-inquiries/archive/2022/optimizing-dermatologist-patient-encounter

2. Sloan B. This month in JAAD Case Reports: February 2022 Availability and Anchoring Biases. J Am Acad Dermatol. 2022 Feb;86(2):299. doi: 10.1016/j.jaad.2021.11.047. Epub 2021 Dec 2. PMID: 34864110.

3. Watari T, Tokuda Y, Amano Y, Onigata K, Kanda H. Cognitive Bias and Diagnostic Errors among Physicians in Japan: A Self-Reflection Survey. Int J Environ Res Public Health. 2022 Apr 12;19(8):4645. doi: 10.3390/ijerph19084645. PMID: 35457511; PMCID: PMC9032995.

4. Yesudian RI, Yesudian PD. A new model for categorizing cognitive biases and debiasing strategies in dermatology. Int J Dermatol. 2023 Feb;62(2):137-142. doi: 10.1111/ijd.16348. Epub 2022 Jul 8. PMID: 35802380.

5. Doherty TS, Carroll AE. Believing in Overcoming Cognitive Biases. AMA J Ethics. 2020 Sep 1;22(9):E773-778. doi: 10.1001/amajethics.2020.773. PMID: 33009773.

6. Jia JL, Lester JC. Anchoring bias and heuristics can perpetuate disparities in dermatology. J Am Acad Dermatol. 2023 Jan;88(1):265-266. doi: 10.1016/j.jaad.2022.05.059. Epub 2022 Jun 4. PMID: 35671884.

7. Dargahi H, Monajemi A, Soltani A, Nejad Nedaie HH, Labaf A. Anchoring Errors in Emergency Medicine Residents and Faculties. Med J Islam Repub Iran. 2022 Oct 26;36:124. doi: 10.47176/mjiri.36.124. PMID: 36447549; PMCID: PMC9700406.

---