Intended for healthcare professionals

Editor's Choice

Diagnosis—the next frontier

BMJ 2006; 333 doi: (Published 24 August 2006) Cite this as: BMJ 2006;333:0-f
  1. Tony Delamothe (tdelamothe{at}, deputy editor

    This week's bmjupdates+—immediately on the left if you're reading the print journal—looks at the causes of temperatures exceeding 41°C in children presenting to US emergency departments. Fewer than half had any sort of infection; viral infections were as common as bacterial infections; and neither white cell count nor absolute neutrophil count helped to distinguish between viral and bacterial infections.

    Once the diagnoses had been made—urinary tract infections caused by Escherichia coli, lobar pneumonias by Streptococcus pneumoniae—the treatment was presumably straightforward. En route to treatment, however, the contribution of the signs (high temperature) and tests (white cell count) to the final diagnosis was anything but straightforward.

    The message I take from this is that while evidence based treatment is well on the way to being sorted out, evidence based diagnosis is still in the dark ages. This week's journal suggests that things are beginning to change. In her editorial Sharon Straus states what should be a self evident truth: “When making a diagnosis in patients who are already ill we should be able to draw on evidence about the accuracy of diagnostic tests” (p 405). Yet just how far we have to go is shown by a study from Susan Mallett and colleagues of reporting and review methods used in systematic reviews of diagnostic tests for cancer (p 413). Lousy methodology means that “even these apparently evidence based studies are flawed,” comments Straus. Relief is at hand: those repositories of evidence based treatments—Clinical Evidence (published by the BMJ Publishing Group) and the Cochrane Library—are turning their attention to diagnosis.

    Perhaps it was their backgrounds in ion channel biophysics and mathematics that sensitised third year residents Matt Bianchi and Brian Alexander to the sloppy way that doctors think and talk about diagnostic tests. “Quantitative reasoning is neither intuitive nor well understood,” they discovered on the wards (p 442). Yet understanding “the limitations of inherently imperfect diagnostic tests” is an important aspect of evidence based medicine, and the authors provide practical guidance.

    Testing can become almost an end in itself. A diagnosis that eluded batteries of diagnostic tests haunts a German general practitioner's account of her 3 year old daughter's life (p 430). Leading her list of what was important was “to be protected from specialists who propose more and more tests but cannot admit they do not know what is wrong.”

    And here's a sign that's still awaiting validation, but which might be useful if you're considering joining the crowds that dash up Mount Everest each year. If you're not ascending 100 metres in 1-1.5 hours, then go back (p 452). Slower than this, there's probably something wrong, and your chances of survival are less. The sign is obvious enough for the French consul in Kathmandu to say of a mountainside fatality, “A 14 hour climb—it is too long. All the files we get of those that die on the mountain, c'est toujours la m®me chose—they take too long to reach the summit.”

    View Abstract