Identifying and minimising data bias, consequences of oseltamivir studies

Toby Lasserson and David Tovey commented in an Editorial of 18 Jan 2012 the issue of The Cochrane Library  Jefferson et al 2012 ”Neuraminidase inhibitors for preventing and treating influenza in healthy adults and children (review)”. The review of unpublished data on oseltamivir unveiled inconsistencies.

Previously unpublished material from15 studies on oseltamivir and 10 studies on zanamivir were used in this study, however, 42 studies could not be reviewed because of insufficient information. [1]

The Jefferson et al 2012 review of  oseltamivir studies [2]
The Jefferson study could not access the entire set of documents. The available data confirmed that oseltamivir shortened the time to alleviation of symptoms by around 21 hours compared with placebo. Many uncertainties, however, remain unresolved and question the mechanism of action of these drugs, in particular that of oseltamivir. Further, the study doubts the hypothesis that oseltamivir does not affect antibody production.

The editorial of Lasserson and Tovey stresses that the study of Jefferson et al demonstrates that systematic reviews can be biased when only restricted research data are included. Bias is known in scientific publications, and is an inclination to present or hold a partial perspective at the expense of other also valid alternatives. A common bias in scientific publications is that 'positive' results are more likely to be published and published sooner, than those with 'negative' results. To avoid misleading results it should be made sure that systematic reviews avoid bias.

The effects on other studies
The Jefferson et al review rises question which other review need to be re-evaluated, what measures should be taken to correct errors, and how to include data obtained from regulatory agencies? The Cochraine Library awaits the maturation of trial registration repositories and better disclosure of data.

The Cochrane Collaboration is supporting a new research project that aims to identify when and how exploration of regulatory and licensing agency data is needed. Systematic reviews should adopt the methods developed by Jefferson et al 2012. To improve the  methodology of systematic reviews the Cochraine Editorial Unit presented methodological expectations. [3]  Link

[1] Lasserson T, Tovey D. Neuraminidase inhibitors for influenza: methods change, principles don't [editorial]. The Cochrane Library 2012 (18 January).
http://www.thecochranelibrary.com/details/editorial/1442857/Neuraminidase-inhibitors-for-influenza-methods-change-principles-dont.html

[2] Safety and efficacy of the anti-influenza drug oseltamivir (Tamiflu) questioned
http://www.ourfood-news.com/2012/01/21

[3] Cochrane Editorial Unit: Methodological Expectations of Cochrane Intervention Reviews (MECIR)
http://www.editorial-unit.cochrane.org/meci