Hum Reprod Update 2008b;14:289–291.

ratio itself. The effect of this last choice also becomes visible, if you should try to extract a number-needed-to-treat (NNT) from each of the meta-analysis: Gelbaya NNT 26.5 and Ruopp NNT 23.5. All together, these choices affect the quality of a meta-analysis. Even Cochrane reviews that are held to a strict protocol to help pre-serve this high quality standard display in 11 % of their systematic reviews a non-properly conducted meta-analysis (Jorgensen et al., 2006). This series of meta-analyses illustrates that meta-analyses are the result of the choices made by the authors and therefore are a reflec-tion of their opinion on the best available evidence. When reading meta-analyses, one should be aware of the fact that these choices were made along the way. One should consider whether these choices are agreeable and acceptable. It may be that meta-analyses are not as objective as one would hope