Article Text

Download PDFPDF

Redefining the ‘E’ in EBM
  1. Tom Jefferson1,
  2. Lars Jørgensen2
  1. 1 Centre for Evidence-Based Medicine, University of Oxford, Oxford, UK
  2. 2 Nordic Cochrane Centre, Kobenhavn, Denmark
  1. Correspondence to Dr Tom Jefferson, Centre for Evidence Based Medicine, University of Oxford, Oxford OX3 7LF, UK; jefferson.tom{at}

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

The philosopher of science, Thomas Kuhn,1 would probably have called our reliance on biomedical journal trial evidence a paradigm. It has served us well, allowing the building of the philosophical2 and practical3 backbone of evidence-based healthcare. However, like all paradigms sooner or later it has begun to creak. Our reliance on journal articles needs a redefinition, if not a shift. In the last decade,4 evidence has accumulated,5 across a spectrum of different interventions,6 that journal publications7 cannot be trusted. Article reports of clinical trials suffer from a grave illness which is curable, but needs a concerted approach to prevent the growing threat of reporting bias.8 When some of us started looking at the alternative sources of evidence for our Cochrane review of neuraminidase inhibitors9 for influenza nearly a decade ago, we discovered that below every 10-page trial report lies a far deeper and more complex web of data and information needing attention. That is, if the trial was published in the first place.

The first problem is sheer bulk. For every page of journal article, there may be up to 8000 pages of regulatory data on the same clinical trial.10 We call it a compression factor.

The next problem is bias. We reasoned that even the most faithful servant of evidence would not be able to publish a 10-pager based on a regulatory report without a radical selection of information and data. As we have no idea what the criteria for …

View Full Text