On diagnosis

Abraham Vergese, an infectious disease physician, gave a talk here in Edinburgh last week. It was a very mixed audience, but I suspect the many students who were there enjoyed it. I have not read any of this books — nor looked at his TED talk — but his Wikipedia entry gives you a flavour of how interesting he is, and how varied a career can be — when you have courage.

One issue that came up tangentially, was the history of diagnosis, and there were some opinions ventured by the audience in terms of when diagnosis was historically established. I may have missed key points, but I found it hard to accept that the idea of diagnosis was something you could date except in very broad terms, even less that you could associated it with the 1870s or with the idea of stethoscopes being a key marker of when modern ideas of diagnosis were established. For instance — and since the lecturer was an ID physician — my first thoughts turned to scabies. The scabies mite was identified in the 1690s, and it was recognised as the cause of the disease ( I am not quoting primary sources so let me know if……) So here we have a clear linking of symptoms, signs, causality, a causal agent, and a broader theory about pathogenesis and epidemiology. So, this it got me thinking about how I view the topic of diagnosis.

Diagnosis is the mapping of one state with another, with the two states being linked by a network of attributes. Diagnosis is a suitcase term: it may contain lots of different tools, tools suited to various purposes, and tools for which we may find different purposes over time. Diagnosis represents an attempt to classify the world into particular states with often the goal of making some predictions about some other state. Most of the time, we think in terms of prediction, about what might happen to that person with or without some intervention. If you see these physical signs (burrows) and the patient describes particular symptoms (itch), then the ‘state’ is scabies. If the diagnosis is correct, you can say something about what causes the state, what might happen, and what effect a particular intervention (permethrin / malathion etc) might have. If you are lucky, you can feel happy with causal arrows linking much of what you say and think. Prediction is important but it is of course not the only quality we want in a theory. We tend to prefer some theories to others, even when they why make similar predictions. Think of Copernicus. We tend to prefer one of the following, irrespective of whether both allow the same quantitative clinical predictions:

1. Sunbathing causes skin cancer: if you increase exposure by X then incidence goes up by Y
2. Sunbathing increase the dose of UVR, UVR is mutagenic and in particular cause very specific types of mutation, cancer is a result of the accumulation of mutations, and therefore we will see particular mutational spectra in skin cancers

Our suitcase of diagnostic concepts have changed over time, however. For instance, even in modern medicine, causality is often lacking. We may use proxy or associated factors to define particular states. We may use simple heuristics as our guide to action, even though we have little idea of where the causal arrows are going. Think much of psychiatry. This does not mean we are powerless, just that we are more ignorant than we would like. We are of course wedded to particular metaphysical systems.

Diagnosis might have been used in the absence of knowledge about particular interventions to attribute blame, as an explanation. If a patient behaved in this way or suffered some state, it was a divine punishment for some behaviour. Now, I may not agree with this world view, but this too is diagnosis. The theory my seem wrong, it may seem primitive, but then my ideas of physics are primitive too if they are applied to the world of the very small.

Galen thought in terms of the mean, and the treatment by opposites (hot treatments for cold; moist treatment for drying diseases etc). This all sounds slightly crazy to modern ears (although dermatologists among you will point out the latter has definite therapeutic merit within very particular skin states). Or how about the idea of therapeutic ‘signatures’. This is from Ian Hacking [1]:

Syphilis is signed by the market place where it is caught; the planet Mercury has signed the market place; the metal mercury, which bears the same name, is therefore the cure for syphilis.

As Hacking points out this allowed Paracelsus to kill lots of people simply because he knew that mercury worked. But whatever the metaphysical system linking two states, the idea of diagnosis was firmly established. Just as Newton got most things right in his physics, and most of us ignore what came after — except when we use the GPS.

Diagnosis was not limited to medicine. Our ancestors spent their lives making diagnoses about what to eat and what not to eat. Making diagnoses about what particular weather states would do to crops etc. Plumbers make diagnoses, as do any humans trying to make sense of an environment that is not static, and where we value intervention.

What may have been specific to medicine was our hangs up about whether there was something special about humans, and whether the simple rules, experimentations and demonstrations of efficacy that allowed other types of human technological progress or indeed much of everyday life, applied in the domain of disease. Successful interventions or demonstrations will have had an effect on metaphysical beliefs in the long term. And of course much of this story is tied up with the growth of that particular branch of formal knowledge we call science. 1870 is just a little late.

[1] Hacking I. The emergence of probability : A philosophical study of early ideas about probability, induction and statistical inference. Cambridge: Cambridge University Press; 1984.

[2] Jonathan Rees: Why we should let EBM rest in peace. Clinics in Dermatology (2013) 31, 806–810