Assessment of risk of bias in randomized clinical trials in surgery.

Abstract

Meta-analysis of randomized clinical trials (RCTs) with low risk of bias is considered the highest level of evidence available for evaluating an intervention. Bias in RCTs may overestimate or underestimate the true effectiveness of an intervention. The causes of bias in surgical trials as described by The Cochrane Collaboration, and the methods that can be used to avoid them, are reviewed. Blinding is difficult in many surgical trials but careful trial design can reduce the bias risk due to lack of blinding. It is possible to conduct surgical trials with low risk of bias by using appropriate trial design. The risk of providing a treatment based on a biased effect estimate must be balanced against the difficulty of conducting trials with very low risk of bias. Better understanding of the risk of bias may result in improved trials with a closer estimate of the true effectiveness of an intervention.