As a clinical neuropsychologist I make mistakes, and I am not alone. Researchers interested in clinical decision-making estimate that across all medical fields diagnosis is wrong 10–15% of the time.
In many instances clinical errors are underpinned by one of a number of cognitive biases. For example, the “availability bias” favours more recent, readily available answers, irrespective of their accuracy; the “confirmation bias” fits information to a preconceived diagnosis rather than the converse. In the time-restricted milieu of emergency medicine, where I work on occasion, particular biases compound: “the commission bias”, a proclivity for action over inaction, increases the likelihood of “search satisfying” – ceasing to look for further information when the first plausible solution is found, which itself might be propelled by “diagnostic momentum” where clinicians blindly continue existing courses of action instigated by (more “powerful”) others.
I can identify such failures in my own work, which may, counterintuitively, guarantee its relative quality: research on the “blind spot bias” indicates that doctors who describe themselves as excellent decision-makers perform relatively poorly on tests of diagnostic accuracy.
Intellectually, I understand that biases are part-bug, built into the brain’s learning preferences to short-circuit complexity in the face of rapid, evolutionarily advantageous decision-making. Igualmente, they can be caused by different types of contextual imperfections; the lack of statistics and mathematical reasoning in medical epistemology; the absence of heuristics to identify how sociocultural norms – ethnicity, género, wealth, mental health – are integrated or excluded from decision-making. But that intellectual knowledge does not translate to understanding what it’s like to be on the receiving end of error. And that lack of understanding might be the most profound bias of all.
When my daughter was a toddler she keeled over face first into her Rice Krispies one morning. A terrible dash to the nearest hospital followed, only for her to revive – wondering what had happened to her breakfast – in the triage queue at A&mi. Having checked her over the paediatrician decided she had probably experienced a one-off seizure caused by a lingering cold. It wasn’t unusual in his experience.
I went to my work on a neurosurgical ward at a national children’s hospital. That afternoon when I described the morning’s events to a surgical colleague, he insisted I have my daughter reassessed, fearing the possibility that the seizure had been caused by an undiagnosed tumour. It wasn’t unusual in his experience. I rushed my daughter back to the original hospital where the paediatrician refused to scan for a tumour.
We were caught between two radically different diagnoses. Both doctors spoke with utter conviction about what was commonplace for them. Perhaps the difference had been caused by “a framing effect”: that the different emphasis I placed while retelling the story had helped to create the discrepant diagnoses. More likely, it was caused by “base rate neglect” – where the underlying incident rates in the relevant population are ignored: the surgeon moved in the rarefied waters of a national hospital where tumours were run-of-the-mill; he never saw febrile seizures, which were relatively common in a local setting. Thankfully, the paediatrician turned out to be right – it was a one-off event, an unforgettably frightening day for our family but nothing more.
Specific cognitive biases can be more or less corrected for by retraining and environmental support, or in more wholesale fashion by replacement with AI systems that use machine learning to improve diagnostic accuracy. But the bias that neglects or foreshortens the experience of the patient is part of what Wittgenstein would call the background “picture” of medicine itself. The picture paints expert, highly specialist clinicians capable of making disengaged, illusion-free decisions about something, even when aspects of it may be fundamentally mysterious to them. In other words, the picture creates perspectival distortions of its own, which can have catastrophic consequences.
Some years ago, a 75-year-old lady, who had lost her husband nine months before, came to my clinic reporting minor episodes of forgetfulness. After my formal memory assessment, the findings were inconclusive. But considering my report alongside her MRI – which showed hyperperfusion (subtle reductions in the blood supply) of the frontal poles of her brain – her neurologist decided she had the early stages of Alzheimer’s.
At our next consultation six months later, the episodes she reported were no longer minor: she’d flooded the kitchen three times in a fortnight; she got lost in a neighbourhood she’d lived in for 30 años; when her phone rang she tried to answer the television remote. But the profile of her memory showed no signs of deterioration, and this time the MRI indicated that the hyperperfusion had disappeared altogether, leaving her with a typical-looking brain for someone halfway through their eighth decade. There was no indication of Alzheimer’s.
The neurologist’s original diagnosis was clearly wrong: transient changes in blood flow – probably relating to grief at the loss of her husband – had been mistaken for a neurodegenerative process. The neurologist admitted the error and corrected the diagnosis that same afternoon. But the woman’s condition continued to deteriorate over the coming months despite a clean bill of neurological health. She was passed on to psychiatry with no sign of a solution. Something about not being “seen” properly in the first instance, compounded by a gross diagnostic error, had intractable consequences for her mental health.
Neurological trauma can change everything for a patient in a moment. And yet as acute clinicians we never see patients either side of a small window of care, neglecting who they might have been prior to it, and only giving short shrift to who they would become afterwards. Years after the fact, I found myself profoundly disturbed by how little I knew about some of my patients.
The framing effect of the medical picture, a failure to consider what it is like to be the person in front of us, means that clinical encounters are doomed to remain between strangers. Cognitive biases inevitably give way to emotional ones, restricting the possibility of empathy. The whole thing is somewhat preordained: clinicians are selected for their knowledge, their problem-solving skills, not for their loving kindness. This is the starting point for my book The Case for Love, a series of case studies whose starting points are failures of imagination.
Imagination, a skill considered the province of storytelling, can broaden the perspective, enhance the picture, by deepening our humanity. Properly applied by clinicians, it may help correct certain biases, changing our patients’ lives for the better.