Fragility Index of Randomized Clinical Trials in Perioperative Anesthesia: A Methodological Survey of Guideline-Supporting Evidence
Margarita Otalora-Esteban, Juan Camilo Segura-Salguero, Amir Zabida, Veronica Echeverri Mallarino, Lorena Diaz-Bohada, Juan Diego Artistizabal-Mayor, Felipe Muñoz-Leyva, Fabián Gil, Martha Beatriz Delgado-Ramirez, Lehana ThabaneBackground:
The Fragility Index (FI) quantifies how small changes in trial outcomes can alter statistical significance, providing a measure of the numerical stability of randomized controlled trial (RCT) results. Despite its growing use, fragility has not been systematically examined in the perioperative evidence supporting anesthesiology guidelines. This study evaluated the FI, Reverse Fragility Index (rFI), and Fragility Quotient (FQ) of RCTs cited in North American and European perioperative clinical practice guidelines and explored trial characteristics associated with fragility.
Methods:
We conducted a methodological survey of RCTs referenced in clinical practice guidelines (CPGs) published between 2012 and 2022. Eligible trials included parallel-group or factorial RCTs with binary outcomes in adult, pediatric, obstetric, cardiovascular, and regional anesthesia populations. Following stratified random sampling, primary analyses were restricted to superiority trials. FI, rFI, and FQ were calculated for the superiority cohort using the binary outcome supporting each guideline recommendation. Trial characteristics associated with FI were explored using a prespecified exploratory negative binomial regression model.
Results:
Out of 1,868 identified RCTs, 639 met the eligibility criteria, and primary analyses included 161 superiority trials. The median sample size was 120 participants (IQR, 80–310), and the median FI was 4 (IQR, 2–8). Cardiovascular trials had the highest median FI (6 [IQR, 1–13]), whereas pediatric trials had the lowest (1 [IQR, 1–5]). In the exploratory multivariable analysis, obstetric trials were associated with higher FI values than adult trials (incidence rate ratio [IRR], 1.63; 95% CI, 1.02–2.65), whereas single-center trials were associated with lower FI values than multicenter trials (IRR, 0.52; 95% CI, 0.35–0.76).
Conclusions:
Randomized superiority trials underpinning contemporary perioperative guideline recommendations are generally small and demonstrate modest Fragility Index values, indicating that statistical conclusions often depend on relatively few outcome events. Fragility metrics should complement, rather than replace, established measures of trial quality and evidence certainty when interpreting perioperative evidence.