バイアスを可視化する

抄録

In systematic reviews, the evaluation of the risk of bias and indirectness on effect estimates is essential to assess the certainty of the body of evidence. A program in the programming language R was developed to execute a meta-analysis adjusting for the effects of biases after assessing bias magnitudes, directions, and uncertainty. The program visualizes the effects of biases by creating an overlayered forest plot of bias-adjusted and -unadjusted values. When performing meta-analyses and sensitivity analyses, this helps assess the certainty of the body of evidence by setting various estimates for biases and indirectness of individual studies as risk ratios and standard deviations or 95 % confidence intervals.