Abstract

Background: Many meta-analyses in anesthesiology are small, heterogeneous, and prone to random error.

Methods: We reappraised the conclusiveness and statistical robustness of 1,300 Cochrane anesthesiology meta-analyses using trial sequential analysis (TSA), which adjusts for repeated significance testing and calculates a required information size (RIS) for 80 % power. For each meta-analysis we assessed whether the cumulative evidence crossed monitoring boundaries for benefit or futility, determined whether the required information size (RIS) was achieved, and quantified underpowered syntheses together with their type I and II error risks.

Results: Only 17.9 % of meta-analyses showed a confirmed treatment effect, 27.4 % crossed futility boundaries, and 54.7 % remained inconclusive; overall, 85.9 % failed to reach the required information size (RIS). Among underpowered analyses, 62.8 % of conventionally significant findings were vulnerable to type I error and 63.9% of non-significant findings were vulnerable to type II error.

Conclusions: Most anesthesiology meta-analyses lack definitive evidence, and TSA frequently downgrades apparently positive or null conclusions. Incorporating sequential monitoring and information-size considerations can enhance the reliability of evidence synthesis in anesthesiology.

Trial Registration: None.