E is for equivocal in EBM

BMJ 2004; 328 doi: https://doi.org/10.1136/bmj.328.7447.1077 (Published 29 April 2004) Cite this as: BMJ 2004;328:1077
  1. Matthew R Kiln, principal in general practice (lkiln{at}AOL.com)
  1. Rosendale Surgery, London SE21 8EZ

    EDITOR—Straus asks what's the E for EBM (evidence based medicine).1 E is for equivocal because that's the best assessment of most published, controlled, double blind research studies (no matter the medical journal).

    The evidence for EBM is inherently unreliable for four reasons.

    Firstly, a large volume of research funded by drug companies is not allowed by them to be submitted for publication.2 3

    Secondly, totally unsuitable patients are sometimes recruited into sensitive drug trials, and only a few such patients are needed to invalidate the conclusions of studies. There is no way when doing peer reviewing or reading the paper after publication that this can be detected. I have noticed this only from reading the paper, carefully going through a patient's notes, and talking to the patient at length. I have seen this in diabetes research, but it could be common in areas of medical research because no one usually has the time, the position, and the information to check for such practices.

    Thirdly, some papers are ghost written.3 4 We will never know how many research papers this practice applies to but it is likely to be substantial. In addition, some chosen professors or doctors who put their names on research papers have never seen the raw data, let alone know how accurately they have been written up.

    Fourthly, bias has an effect. Whether bias is intentionally or unintentionally introduced by researchers, doctors, statisticians, or ghost-writers, the effect is difficult to measure. Declared and undeclared interest is one element,5 but bias occurs in so many other ways that to publish a double blind research paper without it seems almost impossible.


    • Competing interests MRK is cochairman of a patient support charity.


    View Abstract

    Log in

    Log in through your institution


    * For online subscription