New methods to deal with publication bias

Published and retrievable reports of research do not necessarily represent the research performed, and some research findings never reach the consumer. Positive findings—or those that are perceived to be positive—are more likely to be published, and in more prestigious journals, than are negative findings.1 This so called publication bias is an important factor to consider when searching for data and using it to make evidence based decisions. In the linked study (doi:10.1136/bmj.b2981), Moreno and colleagues test the performance of new methods of detecting and correcting for publication bias.2

Data collated for a meta-analysis can be investigated for possible missing studies using a funnel plot,2 3 on which Moreno and colleagues’ method is also based. It is a plot of effect size (for example, relative risk) versus the precision of a trial’s effect estimate (for example, number of patients or standard error). Typically, data points lie in a funnel shaped area (figure⇓). Results of smaller …