Abstract
INTRODUCTION: Benzodiazepines represent the first-line management for status epilepticus; however, heterogeneity in study designs has resulted in conflicting efficacy estimates. We aimed to conduct a target trial emulation meta-analysis to quantify design-induced bias and estimate the efficacy of benzodiazepines for adults in out-of-hospital settings under an ideal double-blind randomized controlled trial (RCT) setting. METHODS: Following PRISMA 2020 guidelines, we searched PubMed, Scopus, Web of Science, Cochrane Library, and Google Scholar up to November 25, 2025. Studies evaluating benzodiazepines for status epilepticus in adults were included. The T3-Meta framework was utilized to model design features as bias covariates, with the target trial effect (θ*) representing seizure cessation rates under ideal RCT conditions. Network meta-analysis was used to evaluate comparative effectiveness. RESULTS: Fourteen studies (2,803 adult patients) were included. Traditional random-effects pooling demonstrated a seizure cessation rate of 68.8% (95% confidence interval [CI]: 63.2-74.0%, I(2)=82.0%). Target trial analysis revealed significant open-label bias (β = 0.757, odds ratio [OR]=2.13, P=0.048), with the bias-adjusted θ* of 64.9% (95% CI: 53.9-74.6%, I(2)=41.5%). Design features explained 49.4% of the between-study heterogeneity. Network meta-analysis demonstrated midazolam superiority over lorazepam (OR=1.60, P=0.001) and diazepam (OR=2.21, P=0.002). Midazolam achieved the highest P-Score (96.2%), followed by lorazepam (50.9%) and diazepam (2.9%). CONCLUSIONS: Traditional meta-analysis was found to overestimate benzodiazepine efficacy by 3.9 percentage points due to open-label design bias. Intramuscular midazolam demonstrated superior effectiveness compared to intravenous lorazepam and diazepam for status epilepticus in adults.