Reply from author of reviewBMJ 1996; 312 doi: https://doi.org/10.1136/bmj.312.7046.1611 (Published 22 June 1996) Cite this as: BMJ 1996;312:1611
- John R Absher
EDITOR,—Jammi N Rao and J D Middleton mention “deficiencies” of my review in Evidence-Based Medicine of the trial of interferon beta-1b. They raise two basic points that deserve a response. Firstly, they caution that my review was not “sufficiently critical of the evidence” and support this claim by noting the “almost diametrically opposite conclusions” reached by me and an anonymous reviewer. Secondly, they criticise “the approach taken by Evidence-Based Medicine” and “warn” readers who rely on this type of publication to be on the lookout for reviews that may be similarly deficient.
The first attack is exaggerated. Two of the main conclusions of both my review and the anonymous one were that the drug seemed to improve measures of disease progression and that further research was needed. I noted in my commentary that there was objective (secondary) evidence of benefit (the percentage change in the area of lesions on magnetic resonance imaging from baseline to one year); high dose treatment was superior to placebo, although dropout rates were similar (14/123, 11/125, and 17/124) and high relapse rates in the placebo group may have blunted evidence of an effect. The anonymous reviewer emphasised “the number and severity of clinical relapses” as the disease measure that was “possibly reduced.” My review is favourable despite flaws in the study, but readers of the BMJ surely realise that type I error is just as important as a type II error. If clinical practice depends on overly timid interpretations of the current best evidence then many patients will be denied access to important new treatments.
Regarding the second point, BMJ readers should note that the review process adopted by Evidence-Based Medicine fosters the integration of methodological critique and clinical acumen. Reviewers strive to extract the clinical relevance of the research through careful consideration of both statistical and clinical significance. The trial of interferon beta-1b satisfied me on both measures, despite its weaknesses. The reviewers for Evidence-Based Medicine may omit methodological details that do not seriously undermine the conclusions of the research, or the implications for clinical practice, so that the publication will be more practical. I consider this enlightened policy to be a great strength of the review process, and am happy to be permitted to participate in it.