The Doctor's Surgery
If two heads are better than one, are three heads better still, or 33?
A somewhat opaquely-written paper about collective intelligence, recently published in the Journal of the American Medical Association, suggests that groups of doctors are better at diagnosis than individual doctors. Quite what this would mean in practice is not easy to say.
At least the opening sentence of the paper is clear, if not altogether reassuring to the poor patient: ‘Diagnosis is central to the practice of medicine, yet studies suggest that misdiagnosis is prevalent even for common conditions, with potential morbidity and mortality.’
Case histories with laboratory results were given to individual doctors and doctors working in groups. When on their own, medical students, doctors still in training and senior doctors got the diagnosis right in 55.8, 65.5 and 63.9 per cent of cases respectively. When they worked in groups of nine, accuracy rose to 85.6 per cent, increasing with group size.
What does this mean? Does it mean, for example, that you should always seek a second opinion? Clearly not: a second opinion is merely that of another doctor working alone. It will only confuse you if the opinions you are given are different.
Should you instead convene a committee of doctors to decide your diagnosis? And should it be a committee of nine, or would six do? Or indeed would 12 be even better? I can’t help recalling George Urban’s book about the Cultural Revolution in China, The Miracles of Chairman Mao, in which it is recounted that hospital cleaners armed only with The Little Red Book were able at meetings to come up with solutions to cases that baffled the most experienced bourgeois professors.
The authors conclude, ‘We found that collective diagnosis of groups was associated with improved accuracy over individual diagnosis. Given the few proven strategies to address the high prevalence of misdiagnosis, these results suggest that collective intelligence merits further study in a real-world, clinical setting.’
But what would such a real-life setting be? It is hard to imagine, outside a dystopian novel. The authors tells us that ‘The modern proliferation of smartphones and widespread use of the internet can potentially enable near real-time gathering and pooling of group decisions.’
In real life, however, at least for the time being, diagnosis is not a straightforward matter of deduction from a few objective data. What is rather chilling about this paper is that it seems to envisage a model of medicine in which doctor and patient do not interact in person, and doctors communicate with one another solely by electronic means.
In fact, doctors still derive a great deal of implicit information from interactions with the patient, which is one of the reasons that continuity of care is not only reassuring for the patient but efficient. A doctor who knows his patients well knows whether they are stoical or hypochondriacal and can often tell just by the way they come through the door whether they feel better or worse.
As medicine becomes ever more impersonal, no doubt with many advantages, so implicit knowledge is lost. Such knowledge cannot be reduced to writing on a computer or captured in laboratory results or those of scans.
Am I just a Luddite, fearing the advance of technology that will eventually make doctors, even in groups, as redundant as pre-industrial hand-loom weavers? Instead of talking on buses about what the doctor said to them, will people talk of what the computer told them?