• By Dartington SRU
  • Posted on Tuesday 11th August, 2009

Does the way of the hunter lead beyond Bradford Hill?

Conceding rather ruefully earlier this year that the meaning of technical concepts will sometimes bend in the wind of changing attitudes, British Medical Journal columnist Christopher Martyn gave as an example a series of reflections on causation widely known nowadays as “Hill’s (or Bradford Hill’s) criteria”.Their author, epidemiologist Austin Bradford Hill (1897-1991), is famous for his pioneering work on the relationship between smoking and lung cancer. He was also the father of the randomized controlled trial.Hill’s criteria are generally regarded as a necessary and, by some, sufficient test of scientific method, social, medical – or forensic. They seem to provide an escape from the profound difficulty of deciding, as Martyn put it, whether “an observed association between exposure to an environmental variable and the occurrence of disease can be taken as indicating causality”. The first criterion is strength of the association. So, for example, if chimney sweeps are found to be 200 times more likely to die from scrotal cancer than workers not exposed to tar or mineral oils, the association is simply too compelling not to be considered causal. But here Hill added an ominously cautious note: “We must not be too ready to dismiss a cause and effect hypothesis merely on the grounds that the observed association appears to be slight. Relatively few persons harboring the meningococcus fall sick of the meningococcal meningitis. Relatively few persons occupationally exposed to rat’s urine contract Weill’s disease.Second is consistency. “Has the association been repeatedly observed by different persons, in different places, circumstances and times?” Third, specificity. “If the association is limited to specific workers and to particular sites and types of disease and there is no association between the work and other modes of dying, then clearly that is a strong argument in favor of causation.” Fourth temporality. “Which is the cart and which is the horse? Does a particular diet lead to disease or do the early stages of the disease lead to those particular dietetic habits?” Fifth biological gradient a.k.a. the dose-response curve. “The fact that the death rate from cancer of the lung rises linearly with the number of cigarettes smoked daily, adds a very great deal to the simpler evidence.” Sixth plausibility. “It will be helpful if the causation we suspect is biologically plausible. But … what is biologically plausible depends upon the biological knowledge of the day. As Sherlock Holmes advised Dr Watson [of whom, more later], ‘when you have eliminated the impossible, whatever remains, however improbable, must be the truth.’” Seventh coherence. “On the other hand, the cause-and-effect interpretation of our data should not seriously conflict with the generally known facts of the natural history and biology of the disease.” Eighth experiment and a reminder – or foretelling – of the usefulness of prevention science. “Occasionally it is possible to appeal to experimental, or semi-experimental, evidence. For example, because of an observed association some preventive action is taken. Does it in fact prevent?” Last analogy. “With the effects of thalidomide and rubella before us, we would surely be ready to accept slighter but similar evidence with another drug or another viral disease in pregnancy.” Christopher Martyn’s first observation about current readings of the Bradford Hill criteria was about context. Consider them in condensed form, and they have the ring of laboratory regulations, when, in fact, they emerged from scientific “stand-up” – a lecture Hill gave in his retirement in 1964 to the Section of Occupational Medicine newly formed by the UK Royal Society of Medicine.More important, he was not asserting that the nine observations amounted to any reliable test of causality. Quite the contrary: “I do not believe,” he insisted, “that we can usefully lay down some hard-and-fast rules of evidence that must be obeyed before we can accept cause and effect. None of my nine viewpoints can bring indisputable evidence for or against the cause-and-effect hypothesis and none can be required as a sine qua non. “What they can do, with greater or less strength, is to help us to make up our minds on the fundamental question – is there any other way of explaining the set of facts before us, is there any other answer equally, or more likely than cause and effect?”Christopher Martyn called his BMJ column “Fighting a lost cause?” and was content to highlight the dangers of allowing meanings to somersault in common usage. “No formula or checklist can establish or refute the presence of a cause and effect relation,” he concluded. Each case must be attended to. End of story.But having traveled this far, what are we now to do? Here is the lifeless body of X; and here – weak or strong, plausible, consistent, specific or coherent – is the evidence that Y is the culprit. Who dares to say who done it?

Elementary, my dear Freud

In a famous essay about clues, method and the “fruitless opposition between rationalism and irrationalism,” the historian Carlo Ginsburg connected the “scientific connoisseurship” of Italian art historian and authenticator Giovanni Morelli, the detective rigor of Sherlock Holmes and the early psychoanalytical impulse of one who admired and was influenced by the achievements of both – the young Sigmund Freud. He did so in order to discuss ideas about decipherment “the ability to construct from apparently insignificant experimental data a complex reality that could not be experienced directly.” (Morelli diagnosed the authorship of paintings by comparing the rendering of details – ears and nipples, for example.)Ginzburg was attracted to the observation that, after a fashion, all three were medics: Freud was a physician, Morelli had a medical degree, Conan Doyle, a GP before he started writing, modeled “Holmes-Watson” on one of his Edinburgh professors, a diagnostician. But he likened their truer talent to the capacity to construct a narrative out of a jumble of trivia - narrative being a coherent rendering of causality. It was akin to the hunter’s primitive certainty, based on the disposition of birds and animals and other signs, that “someone passed this way”.Ginzburg wrote of it as a cognitive leap – ultimately equivalent to the difference between Bradford Hill’s modest assertion that no foolproof system of logic allows the inference of causality, and his prodigious achievement as an epidemiologist.Ginzburg worried about using the “dangerous term” intuition, and avoided it out of preference for the Arabic quality of firasa – whereby perception becomes a source of certain knowledge. Ancient applications of firasa, he noted, extended to the art of finding one’s way in the desert, the art of discovering springs, the art of forecasting rain and divination by means of footprints.See: See: Martyn C “Fighting a lost cause?” BMJ 2009:338:b1621 and Ginzburg C (1990) “Clues: Roots of an Evidential Paradigm” in Myths, Emblems, Clues, Hutchinson Radius • To read Austin Bradford Hill’s lecture, “The Environment and Disease: Association or Causation?” from the Proceedings of the Royal Society of Medicine, 58 (1965), 295-300, visit the website of the information designer and statistician Edward Tufte

Back to Archives