Anchoring Bias: Tales from the Front Line:
A thorough history and awareness of the unexpected help overcome anchoring bias.
Tales from the Front Line: Anchoring Bias
The medical notes say he has a swollen, red, painful wrist. As he is bed-bound, he requires a home visit—the differential lists cellulitis as the most likely diagnosis.
“Did you hurt your wrist,” I say.
“No”
I look for any scratches or entry points that indicate infection. The notes state the wrist is warm, red, and painful. He has a temperature, but otherwise, his observations are stable. Something doesn’t add up. It does look like he has cellulitis, but there’s no source for the infection.
In their influential 1974 paper “Judgment under Uncertainty: Heuristics and Biases,” Tversky and Kahneman proposed that when individuals make predictions, they start with an initial value or anchor and adjust their reasoning from that arbitrary point.
Anchoring is a common bias in emergency settings, leading to diagnostic errors. Applying excessive weight to initial information makes adjusting the diagnostic hypothesis challenging even when further information becomes available.
The presence of anchoring bias is a warning sign that can alert clinicians to the increased risk of diagnostic error. In a qualitative study interviewing GPs, we found that early views of the problem are at least partly based on an initial fast response to the issue (System 1 reasoning) and tend to be biased unless one takes time for a slow analytic, reflective response under the influence of System 2 reasoning.
“Do you have any other symptoms,” I ask.
“Well, I have a little bit of tummy pain. Also, my eyes have been a bit of a bother.”
The carer says, “ His urine has been a bit smelly.”
He has a catheter. “When was it last changed,” I ask.
“Two weeks ago.”
Reactive arthritis is joint pain and swelling triggered by an infection in another body part. Reiter's syndrome, also known as reactive arthritis, is the triad of conjunctivitis, urethritis, and arthritis that occurs after an infection, particularly in the urogenital or gastrointestinal tract. In reactive arthritis, the body's immune system overreacts to the infection and attacks healthy tissue, causing inflammation. Hence the swollen wrist. However, the exact mechanism for the arthritis reaction is unknown, and no single test exists.
I test the urine - he has a full house for a catheter-associated infection.
A thorough history and awareness of the unexpected help overcome anchoring bias. I can diagnose the source of the problem and form a plan to manage it before proceeding to the next case.
The full series
Both Carl's post and Helen's comment are revelatory.
I'm an old Chartered Engineer with no medical training, but the "anchoring bias" issue makes absolute sense and corresponds to many experiences I have had, both at work and in my long life.
Of course, this is bad enough with a single doctor making an honest, if incorrect diagnosis about a single patient.
When hugely boosted by malevolent and venal billionaires with their own agendas, on the one hand and "the madness of crowds" on the other, it becomes disastrous.
This issue of anchoring bias was one of my great concerns in April 2020. The global pre-occupation with the new virus created a sudden shift where en masse everyone from the public to healthcare workers saw every febrile or non specific acute illness through the prism of ‘suspected Covid until proved otherwise’. So sepsis, diabetic ketoacidosis, UTI, acute heart failure, bacterial pneumonia, any other virus, delirium in elderly… all probable Covid. And once the vascular complications of Covid were recognised that list soon expanded to include other things- myocardial infarction, acute stroke, pulmonary embolism, weird neurological stuff, sudden death at home with cough medicine or tissues at the scene- Covid. The death certification changes reinforced this anchoring bias. Worse, the advice was that this was an untreatable condition and face to face assessment should be avoided unless symptoms indicated more severe illness, but if the wrong diagnosis is being considered because of the anchoring bias then that advice would be wholly inappropriate for someone whose new onset slight chest pain and breathlessness was not due to a viral infection…
In early April 2020 I wrote to our clinical team suggesting we must not lose sight of all the other stuff that is not Covid-19 in the fear and noise of Covid-19 and that we should provide timely assessment and treatment where indicated. In those first weeks it could be a challenge admitting an older adult with a febrile illness to hospital as it was assumed there was nothing that could be done for them and if they were admitted with Covid they may cause the disease to spread in hospitals (this was the main take home message from Italy- the hospitals were overwhelmed with older people, treatment attempts were futile, the virus spread and infected healthcare workers, and then when younger people presented a week or two later with severe infection there were no beds for them). The NHS took preemptive action to try to avoid that reality. I think the anchoring bias did cause delays in face to face assessment as was predicted. Well never know many came to permanent harm as they are all mashed up in the official Covid statistics.