A new paper on risk of bias assessments from the Meta-analysis Group at CTU has been published in the BMJ. The paper evaluates the reliability of risk of bias assessments for randomised trials in cancer. The study used the Risk of Bias tool developed by the Cochrane Collaboration, which is mandatory in Cochrane systematic reviews, most of which rely on using published reports of trials.
The paper describes how, compared with information obtained during the process of individual patient data meta-analyses, assessments obtained from publications only can be misleading. Using the outcomes of assessments based on published reports alone leads to more trials being judged as unclear or high risk of bias and may lead to trials being inappropriately left out of meta-analyses. This open access paper can be read online by everyone.
The paper concludes that obtaining additional information from published protocols or directly from investigators on trials would help inform risk of bias assessments and improve their reliability. It also suggests that the guidance around completing such assessments could be improved.