Retiring the Term “Weighted Mean Difference” in Contemporary Evidence Synthesis

作者：Lifeng Lin, Xing Xing, Wenshan Han, Jiayi Tong · 发表于：Cochrane Evidence Synthesis and Methods · 年份：2025 · DOI：10.1002/cesm.70051 · 被引用次数：4 · 研究领域：Meta-analysis and systematic reviews

Evidence synthesis frequently involves quantitative analyses of continuous outcomes. A cross-sectional study examining Cochrane systematic reviews identified 6672 out of 22,453 meta-analyses (29.7%) involved continuous outcomes [1]. The primary effect measures employed in meta-analyses of continuous outcomes are the mean difference (MD) and standardized mean difference (SMD) [2]. The MD is appropriately applied when all included studies measure outcomes using identical scales (e.g., body weight in kilograms). In contrast, the SMD serves as a solution when studies utilize different measurement scales (e.g., varied questionnaire scoring methods). Although alternative measures (e.g., the ratio of means) exist [3], their application remains relatively infrequent. Despite this conceptual clarity, the term “weighted mean difference” (WMD) appears frequently in the systematic review literature [4], which can lead to confusion about its relationship to the MD. In this article, we first clarify the distinction between MD and WMD, then describe the historical factors underlying the term's adoption and persistence, discuss why contemporary methods render it unnecessary, illustrate examples of misuse, and conclude with practical recommendations for clearer reporting. The MD represents the straightforward difference between group means (e.g., intervention vs. control) for a continuous outcome. Although the true MD value relates to unknown population-level differences, practical research r...