• Taina Mattila, Violeta Stoyanova, André Elferink, Christine Gispen-de Wied, Anthonius de Boer, and Tamar Wohlfarth.
• Medicines Evaluation Board, The Hague, The Netherlands. tk.mattila@cbg-meb.nl
• Eur Neuropsychopharmacol. 2011 Jul 1;21(7):500-7.

AbstractSelective publication can have a deleterious effect on evidence based medicine, health policy decision making and treatment guidelines. Using the European Public Assessment Reports (EPARs) as reference, this study examined selective publication and selective reporting of efficacy and safety of insomnia medication. EPARs of with three insomnia medications were used to identify all clinical trials that were performed between 1998 and 2007 for the purpose of registration in the EU. The matching publication for each trial was searched through a systematic literature search. Accuracy of information in the publications was examined by comparison to the information in the EPARs. Only 55% of the trials with insomnia medications identified in EPARs were published. Positive trials were approximately two times more likely to be published. The lag time from study completion to publication was shorter for the positive compared to the negative trials. Sample size did not correlate with publication of negative trials. The meta-analysis of the effect size of insomnia medication was 1.6 times larger in the published data compared to the complete data. While the primary end points of the trials were reported reliably in the publications, remarkable inconsistencies were detected in the reporting of the secondary end points, methods, results and, especially safety. In conclusion, selective publication and reporting lead to an overestimation of efficacy and underestimation of safety of insomnia products. Authors of treatment guidelines should be aware of this bias. EPARs/FDA reviews provide a more unbiased view of the benefit-risk balance of insomnia and other medications and hence these documents should be consulted by e.g. authors of meta-analyses and of treatment guidelines.Copyright © 2010 Elsevier B.V. All rights reserved.

17 keywords...

hide…