Trial sequential analysis has the potential to improve the reliability of conclusions in meta-analysis

摘要

Editor: A recent publication in this journal [ 1 ] explored the ability of Trial Sequential Analysis (TSA) to draw firm conclusions in meta-analysis. The authors of this article have recently presented a new command in stata software [2], allowing TSA to be implemented and including the option to analyse hazard ratios, and we commend them for this work. Having ourselves developed software to implement TSA for binary and continuous outcomes (www.ctu.dk/tsa) and begun an exploration of its implementation in empirical studies [3-6], we read the paper by Miladinovic and colleagues with much interest. The authors claim that their evidence shows that "TSA may lead to a premature declaration of statistically significant treatment difference, when further accumulated evidence suggested otherwise". To further the discussion, we suggest that their conclusion, as it is worded, may not be justified by the evidence they present, and may be misleading.