Abstract

Background: Meta-analyses and randomized controlled trials (RCTs) play pivotal roles in evidence-based medicine. However, meta-analyses are increasingly criticized for overestimating treatment effects and lacking agreement with large RCTs, potentially resulting in misleading or premature conclusions that influence clinical guidelines. Small, early-phase trials and publication bias contribute to type I and type II errors, raising concerns about the strength of meta-analytic findings. Trial Sequential Analysis (TSA) is a statistical tool designed to assess the robustness of cumulative evidence by adjusting for random errors and required information size.

Objective: This study evaluates the agreement between meta-analyses and subsequent large RCTs in perioperative medicine published between 2015 and 2022. Additionally, it investigates whether TSA alters the interpretation of meta-analytic findings.

Methods: A systematic search identified large RCTs (≥1,000 participants, with at least one major dichotomous clinical outcome) and their corresponding preceding meta-analysis. TSA was applied to each outcome to determine whether the meta-analysis had reached a reliable conclusion and to classify results into distinct evidence zones.

Results: Of the 23 outcome comparisons assessed, 78.3% of meta-analyses correctly predicted the results of the corresponding large RCTs. However, TSA reclassified several initially ‘accurate’ predictions as inconclusive or potentially false positive, particularly under assumptions of higher relative risk reductions.

Conclusion: Although meta-analyses often align with subsequent RCTs, they carry a substantial risk of false positives, especially when based on small studies. TSA adds important nuance by identifying when cumulative evidence is insufficient for firm conclusions. These findings support a cautious interpretation of meta-analyses in clinical decision-making and emphasize the need for large, well-powered RCTs before changing clinical practice.