Reducing the effect of bias

Unfortunately it is all too common for individual trials, which have produced the most striking results, to be emphasised in the medical literature. Perhaps the most important reason for performing systematic meta-analyses is to avoid this biased viewpoint. Conclusions on the relative merits of therapies should be based on the results of all relevant trials, irrespective of whether or where they have been published, or how much publicity they have received. In addition to this publication and reporting bias (Section 11.4.2), there are many aspects of the design, conduct, analysis and reporting of individual trials that may be subject to potential biases [12] and which could influence the trial results (see Chapters 4, 5 and 9). Any systematic review or meta-analysis should therefore appraise trials critically and any potential biases and problems should be noted and if possible rectified. If important problems remain, careful consideration should be given as to whether it is appropriate to include such trials in the systematic review.

0 0

Post a comment