Doctors are not scientists

BMJ 2004; 328 doi: (Published 17 June 2004)
Cite this as: BMJ 2004;328:0.9
  1. Richard Smith (rsmith{at}, editor

    Some doctors are scientists—just as some politicians are scientists—but most are not. As medical students they were filled full with information on biochemistry, anatomy, physiology, and other sciences, but information does not a scientist make—otherwise, you could become a scientist by watching the Discovery channel. A scientist is somebody who constantly questions, generates falsifiable hypotheses, and collects data from well designed experiments—the kind of people who brush their teeth on only one side of their mouth to see whether brushing your teeth has any benefit. Most doctors follow familiar patterns and rules, often improvising around those rules. In their methods of working they are more like jazz musicians than scientists.

    Questioning whether doctors are scientists may seem outrageous, but most doctors know that they are not scientists. I once asked a room of perhaps 150 medically trained educators which of them thought of themselves as scientists. About five put up their hands.

    If doctors are not scientists then it seems odd to supply them, as medical journals do, with a steady stream of original scientific studies. Teachers and social workers are not sent original research. Nurses are sent some, but are they simply aping the illogical ways of doctors?

    The inevitable consequence is that most readers of medical journals don't read the original articles. They may scan the abstract, but it's the rarest of beasts who reads an article from beginning to end, critically appraising it as he or she goes. Indeed, most doctors are incapable of critically appraising an article. They have never been trained to do so. Instead, they must accept the judgment of the editorial team and its peer reviewers—until one of the rare beasts writes in and points out that a study is scientifically nonsensical.

    Sometimes readers will alight on an article as a bee alights on a flower to suck a little honey. They will alight, I suspect, for reasons that are more personal than scientific. I am interested in the study showing a steady rise in hospital admissions for acute pancreatitis from 1963 to 1998 (p 1466) because my brother had pancreatitis—maybe, indeed, that link had something to do with the study making it into the journal just as it's been suggested that the BMJ publishes on toenail fungus because so many of the editorial team suffer from it. The authors note that the prognosis of acute pancreatitis is poor and that mortality after admission has not fallen since the 1970s—reflecting the absence of innovations in treatment.

    I am attracted as well to the study on whether the uncertainty principle is violated in clinical trials (p 1463). The principle says that you shouldn't conduct a trial if you think that one treatment is likely to be better than another. The study looked to see if trials more often favoured the experimental treatment. I guessed they would—and, indeed, they did. The authors, however, judge that the trials do satisfy the uncertainty principle. I'm unconvinced.

    To receive Editor's choice by email each week subscribe via our website: