Reason knows nothing: how biases infect medicine

Doctors are not creatures of fact. Despite the medical profession striving for logical progression, a limiting factor is the orchestration by humans. We, as a species, are tainted by cognitive errors. Our decisions are influenced by a myriad of biases – both consciously and subconsciously tweaking everything we do.

The names Amos Tversky and Daniel Kahneman may not be familiar to readers of the Journal of the Royal Society of Medicine. When asked to name medical pioneers – a modern day Osler, Jenner or Blackwell – the names of two Israeli-American psychologists do not immediately jump to mind. Tversky and Kahneman spent three decades focusing on judgement and decision-making under conditions of uncertainty.¹ This started with simple, psychology experiments at Hebrew University in Jerusalem, cleverly cracking shells in our societal mirage of rationality. A culmination of this work resulted in the Nobel Prize for Economics, awarded to Kahneman in 2002.²

Medics make a constant stream of decisions tempered by uncertainty. Heuristics – the strategies by which we all form decisions – are fallible. One such example is the representative heuristic, whereby judgement is predicated on how much a novel situation/patient fits a known situation/patient. Physical representations of patients had a greater effect, when sorting patients into groups, compared with other factors such as known probabilities.³ For example, a study asked nurses to differentiate between scenarios suggesting a patient had a cardiac arrest or stroke.⁴ Additional situational information was given in some scenarios – whereby those in the cardiac attack group had been fired from their job and those in the stroke group had alcohol-smelling breath. The study found that nurses who had been given situational backgrounds were more likely to use this information to choose a less serious diagnosis – a highly significant effect reducing the accuracy of diagnosis. When probabilities were given (for example, 30% of patients were in the stroke group), the use of situational information superseded that of probabilistic objectivity. It is important to note that the process of diagnosis is complex, relying on a variety of factors including medical history, background, epidemiology and social factors. However, the paper reasoned that situational information, an example of the representative heuristic, bypassed more relevant factors in the decision-making process. These findings were remarkably similar regardless of training status implying such a bias was more deeply rooted than mere experiential naivety.

Cognitive errors in medicine are in the process of being identified with no indication of exhaustion – the bandwagon effect, default bias, anchoring bias, decoy effect, ambiguity aversion, etc., the list is long.⁵ A study by Tversky investigating one such error, the cognitive framing allusion, appeared in the New England Journal of Medicine in 1982.⁶ A theoretical patient was given lung cancer with a short life expectancy. However, an experimental surgery, with the potential of a cure, was offered. One group of doctors was told the patient had a 90% chance of surviving the surgery. The other group was told the patient had a 10% chance of dying. Those in the first group were nearly twice as likely to recommend surgery despite the given percentages resulting in identical outcome (mortality vs. survival rates). The ethical implications behind this cognitive error are important to consider. Doctors financially incentivised to operate can use similar distortion techniques to influence the decision of patients. ‘This is a nudge in the medical world’ states Michael Lewis, author of The Undoing Project, a book centred around Tversky and Kahneman’s friendship.⁷ Nudges are subtle changes which have the potential to significantly influence choice architecture and resultant behaviour. Examples include an opt-out organ donation registry (generally considered good) and the creation of unnecessary additional paperwork needed to vote in the US (generally considered bad).^8,9

We are complex. Our brains are a hodgepodge of functionally distinct areas. The motor cortex allows us to battle against Newton’s gravity and stretch stringy sinews of muscle to their mechanical limit. Our visual system, a symphony of electrical fibres, beams of light and liquid distortion, is orders of magnitude more complex than any man made variation. Yet, cognitive errors are as natural as our ability to move and see. So how does medicine deal with these findings? Tversky and Kahneman’s work is increasingly well established. Medical journals eat up articles shedding light on natural stupidity – a tongue-in-cheek quip by Tversky, using the antithesis of ‘artificial intelligence’. Yet, the clinical realm remains generally unchanged. With medical error shown to be the third most common cause of death in the United States, a systematic change is needed.¹⁰ Medical schools forgo lectures on this most important of lessons – the fallibility of the individual. Specialty training and fellowships emphasise the importance of experiential improvement alone, when in fact the explicit teaching of universal cognitive errors – such as the representative heuristic and framing allusion – would improve the consistency of clinical decisions.

Aesop’s fable, the boy who cried wolf, is often told to children to emphasise the tragedy that befalls liars. Yet, far more pertinent is the Greek myth of Cassandra – daughter of King Priam of Troy – who was given the gift of prophecy, only to be cursed by universal disbelief after angering the god Apollo.¹¹ She was the antithesis of the boy who cried wolf – somebody who only told the truth yet was never believed. The problem of arrogance and insularity are well advertised in medicine as are the gradual improvements to counter these most ancient of conditions. Medicine is becoming less hierarchical and more multidisciplinary. Yet, lessons unearthed decades ago are still buried beneath a tectonic shell of status quo. Tversky and Kahneman do not deserve to be treated as modern day Cassandras, if only to shield patients from the fate of the boy who cried wolf.

Declarations

Competing Interests

None declared

Funding

None declared

Ethics approval

Not applicable

Guarantor

SP

Contributorship

Sole authorship

Provenance

Not commissioned; peer-reviewed by Bharadwaj Chada