For healthcare professionals only

Editor's Choice

Simple problems please, and one at a time

BMJ 2005; 330 doi: (Published 19 May 2005) Cite this as: BMJ 2005;330:0-g
  1. Fiona Godlee, editor (fgodlee{at}

    Great is our desire as doctors to identify a single simple problem that has a single simple solution. This approach usefully fits the constraints of the 10 minute consultation, but its roots lie in Western medicine's biomedical model of disease. Doctors are still trained to diagnose single well defined disorders and to offer single targeted treatments. We prefer if possible to deal with one problem at a time.

    The very name post traumatic stress disorder (PTSD) fits this picture: a single problem with an identifiable cause and potentially therefore a single solution. But this is deceptive says Simon Wessely (p 1215). Since the Vietnam war, when PTSD was first characterised, psychiatry has had to accept that “the invention of the disorder did not reverse half a century of knowledge and that the person exposed to the trauma matters just as much as the trauma itself.” It has proved surprisingly difficult, for example, to predict who will develop PTSD, and the results of trials of the disease specific intervention psychological debriefing have been disappointing.

    The apparent simplicity of PTSD as a concept may be to blame for what the World Health Organization sees as the disproportional resources it receives in disaster areas. As van Ommeren and colleagues point out, other more mundane conditions such as anxiety and depression contribute most to the burden of mental illness after a disaster such as the Asian tsunami (p 1160). Under these circumstances, the most useful interventions are likely to be social and economic, such as getting children back to school and supporting economic development.

    Randomised controlled trials (RCTs) are a construct of the biomedical model and one of its key pillars. No wonder then that they too thrive on simplicity. Gunn Elisabeth Vist and colleagues give them a clean bill of health—patients in trials do as well as those given the same treatment outside a trial and the results of trials can be generalised to people not in trials (p 1175). But Charlotte Paterson and Paul Dieppe explain convincingly why RCTs are no good at evaluating complex interventions, such as acupuncture and psychotherapy (p 1202). Acupuncture is interwoven with so called non-specific factors, such as talking and listening, which are in fact part of the therapeutic relationship. The process of diagnosis is also hard to separate from the therapy, emerging as it does throughout treatment rather than being an isolated preceding event. The fact that these non-specific but probably therapeutic elements are usually in both arms of trials explains (to my satisfaction at least) the otherwise puzzling fact that, despite its long history and widespread use, acupuncture has never found convincing support from RCTs. It also explains why both real and sham acupuncture shows benefit in trials. Simplicity is appealing but can, if we are not careful, be misleading.


    View Abstract