Advanced Methods and Software for Trial Sequential Analysis in Living Systematic Reviews - PROJECT SUMMARY Systematic reviews (SRs) and meta-analyses (MAs) are essential tools in evidence-based medicine and comparative effectiveness research, enabling thorough evaluations of treatment benefits and risks. In the context of the COVID-19 pandemic, SRs and MAs have been extensively utilized to provide timely insights on preventive interventions, therapeutic medications, and vaccines, aiding informed decision-making and response strategies. However, traditional SRs have limitations such as redundant publications, inconsistent inclusion criteria, and inadequate investigation of evidence accumulation. Living systematic reviews (LSRs) have emerged as a dynamic approach to continuously update and synthesize evidence, addressing these limitations. LSRs offer benefits such as timeliness, reduction of research waste, identification of research gaps, and integration of the latest evidence. Trial sequential analysis (TSA) is a valuable tool to derive conclusive evidence for assessing the adequacy of studies in LSRs. By providing monitoring and futility boundaries, TSA ensures reliable decision- making regarding a treatment’s effectiveness or futility. Conclusive evidence can save patients from harmful treatments or placebos, and redirect resources to other research areas. However, the current TSA methods have significant limitations, mainly stemming from their heavy reliance on interim analyses of RCTs, where patients tend to be more homogeneous than those in MAs that consist of multiple studies on different populations. This proposed project aims to advance TSA methods by developing innovative approaches to establishing decision boundaries, specifically targeting the improvement of early-stage TSAs and reducing the risk of premature termination of LSRs. By addressing between-study heterogeneity and enhancing statistical methods, this project seeks to enhance the reliability and timeliness of living evidence. Rigorous validation of the proposed methods for TSAs will be conducted through extensive simulation studies carefully designed to evaluate the overall type I and type II error rates of treatment effect estimates. Additionally, the performance of the proposed methods will be assessed using diverse real-world datasets. Furthermore, user-friendly, open-source software will be developed, accompanied by comprehensive instructions and examples, ensuring accessibility and ease of implementation for biostatisticians and clinicians. In conclusion, these novel TSA methods hold broad applicability across medical fields, including infectious diseases and cancers, facilitating more robust assessments of existing evidence, guiding decisions regarding the necessity of new randomized controlled trials, and ultimately advancing comparative effectiveness research and evidence-based medicine.
