Correspondence
Selective Publication of Antidepressant Trials
N Engl J Med 2008; 358:2180-2182May 15, 2008
- Article
To the Editor:
The importance of the study reported by Turner et al. (Jan. 17 issue),1 on selective publication of antidepressant trials, can hardly be overstated because it shows how researchers and clinicians are deprived of accurate data, resulting in a wrong understanding of antidepressant efficacy. The most serious implication of this study is that guidelines for the treatment of depression may be inaccurate, since they are often based on meta-analyses of published data alone.2,3
This current system makes it impossible for scientific journals to provide a valid picture of the efficacy of antidepressants. Therefore, we propose that meta-analyses be accepted only when they include an adequate analysis of the potential bias due to exclusion of unpublished studies (beyond Egger plots, since large negative trials may also remain unpublished).
3 ReferencesPeter de Jonge, Ph.D.
Claudi L. Bockting, Ph.D.
University of Groningen, 9700 RB Groningen, the Netherlands
peter.de. jonge@med. umcg. nl 1
Turner EH ,Matthews AM ,Linardatos E ,Tell RA ,Rosenthal R . Selective publication of antidepressant trials and its influence on apparent efficacy. N Engl J Med 2008;358:252-260
Full Text | Web of Science | Medline2
Moncrieff J ,Wessely S ,Hardy R . Active placebos versus antidepressants for depression. Cochrane Database Syst Rev 2004;1:CD003012-CD003012
Medline3
Katzman MA ,Tricco AC ,McIntosh D , et al. Paroxetine versus placebo and other agents for depressive disorders: a systematic review and meta-analysis. J Clin Psychiatry 2007;68:1845-1859
CrossRef | Web of Science | Medline
To the Editor:
Turner et al. discuss the question of how accurately the published literature conveys data on drug efficacy to the medical community. To address this question, they performed a literature search. They searched PubMed and the Cochrane Central Register of Controlled Trials. I would like to make the following suggestions to improve their literature search. First, always search EMBASE, in addition to PubMed, when executing a systematic search, especially in fields such as pharmacology. Not only does EMBASE have a focus on the pharmacologic literature, but it also covers 2000 journals not indexed in PubMed. Second, consider other bibliographic databases, depending on the scope of the research question (e.g., CINAHL for nursing, PsycINFO for behavioral medicine, and Web of Science for meeting abstracts).
In short, other bibliographic databases can complement PubMed searches considerably. These suggestions might lead to the identification of relevant, previously unknown references,1 thus supporting the statement Turner et al. make: that evidence-based medicine is valuable to the extent that the evidence base is complete.
1 ReferencesJan W. Schoones, M.A.
Leiden University Medical Center, 2300 RC Leiden, the Netherlands
j.w. schoones@lumc. nl 1
Schoones JW . Improving literature searches. Radiology 2007;243:301-302
CrossRef | Web of Science | Medline
To the Editor:
We challenge assertions made by Turner et al. Both venlafaxine studies reported as unpublished were “failed” studies. The Food and Drug Administration (FDA) distinguishes between negative and failed studies as follows: “a drug that does not work — and a study that does not work.”1 Failed studies cannot be meaningfully interpreted in isolation, and independent publication does not inform measures of drug efficacy.
The venlafaxine study characterized as published with results that conflicted with the FDA decision was conducted to assess the low end of the dose range. The published manuscript2 reports exactly what the FDA concluded: the “results clearly showed a statistically significant dose–response relationship for venlafaxine.”3
Two issues suggest author bias: disregarding publicly available data in abstracts and reports on multiple studies, and ignoring the limited scientific value of failed studies. Wyeth strongly advocates publication transparency regardless of the trial outcome and has followed such a policy since 2004.
3 ReferencesPhilip T. Ninan, M.D.
R. Michael Poole, M.D.
Gary L. Stiles, M.D.
Wyeth Research, Collegeville, PA 19426
ninanp@wyeth.com 1
Temple R ,Ellenberg SS . Placebo-controlled trials and active-control trials in the evaluation of new treatments. 1. Ethical and scientific issues. Ann Intern Med 2000;133:455-463
Web of Science | Medline2
Mendels J ,Johnston R ,Mattes J ,Riesenberg R . Efficacy and safety of b.i.d. doses of venlafaxine in a dose-response study. Psychopharmacol Bull 1993;29:169-174
Medline3
Food and Drug Administration. Summary basis of approval for Effexor.
Author/Editor Response
We disagree that failed studies should be ignored (though our results were unchanged when we excluded them). Deciding not to publish any study on the basis of how it turns out amounts to a post hoc litmus test, which is not “publication transparency regardless of the trial outcome.” We agree with Otto and Nierenberg1 that the notion of failed studies is logically problematic. In the unpublished venlafaxine study (study 367), when the active comparator paroxetine did not beat placebo, was it because the study “did not work”? Or since paroxetine could not beat placebo in 9 of 16 studies, might it be that study 367 worked fine, but the drugs in the study did not? If the sponsor wishes to discount such results, instead of keeping them secret, it should disclose them, make the “limited scientific value” argument in the discussion section, and let the scientific community reach its own conclusions.
Pharmaceutical companies have begun posting clinical-trial results on their Web sites, but it is not clear whether physicians are aware of and use this information. Also, none of the nine venlafaxine trials reviewed by the FDA (involving 1675 patients) have been disclosed on Wyeth's Web site (www.wyeth.com/ClinicalTrialResults?query=Depressive%20Disorder). Since venlafaxine was the 12th most prescribed brand-name drug in 2007, with 17.2 million prescriptions (www.drugtopics.com/drugtopics/article/articleDetail.jsp?id=491207), aren't these results relevant to the public health?
We believe that the publication exposure a clinical trial receives should not hinge on how it turned out. There should be no shame in acknowledging that not every clinical trial is a “success.” If transparency is applied selectively, then it's not transparency — it's translucency.
1 ReferencesErick H. Turner, M.D.
Oregon Health and Science University, Portland, OR 97239
turnere@ohsu.edu Robert A. Tell, L.C.S.W.
Portland Veterans Affairs Medical Center, Portland, OR 972391
Otto MW ,Nierenberg AA . Assay sensitivity, failed clinical trials, and the conduct of science. Psychother Psychosom 2002;71:241-243
CrossRef | Web of Science | Medline
- Citing Articles (1)
Citing Articles
1
Joel Lexchin. (2011) Those Who Have the Gold Make the Evidence: How the Pharmaceutical Industry Biases the Outcomes of Clinical Trials of Medications. Science and Engineering Ethics
CrossRef
