Addressing Continuous Data for Participants Excluded from Trial Analysis: A Guide for Systematic Reviewers

The document from 2013 provides a guide for systematic reviewers on managing missing continuous outcome data in meta-analyses to reduce bias. The authors proposed four imputation strategies of varying stringency and applied them to two systematic reviews—one on cognitive behavioral therapy for depression with a 21% median missing data rate, and another on finasteride therapy for androgenetic alopecia in men with a 14% median missing data rate. The impact of missing data varied, with effect estimates in the first review losing significance under more stringent imputation, while in the second review, they remained significant. The authors concluded that the risk of losing significance due to missing data is higher for small effects and larger percentages of missing data. The recommended approach includes using the median standard deviation for imputation and aligns with GRADE/Cochrane handbook guidance, aiming to improve the handling of missing data in systematic reviews. However, the approach is limited to reviews with studies using the same measurement instrument for continuous outcomes.