Abstract
<title>Abstract</title> <p>Background: Despite decades of sepsis randomized clinical trials (RCTs), few therapeutic interventions have demonstrated reproducible mortality benefit. Whether this phenomenon reflects failure of individual therapies or broader limitations in sepsis RCT design remains unclear. We performed a domain-specific, meta-analysis of adult sepsis RCTs and evaluated the distribution of mortality effect estimates across therapeutic intervention domains using a complementary landscape analysis framework. Methods: We conducted a meta-analysis of adult sepsis and septic shock RCTs published between 2001 and 2025 (PROSPERO CRD420261289221). Using PubMed/MEDLINE, Embase, and CENTRAL, two reviewers screened studies and extracted data. Trials were grouped into six therapeutic domains: initial resuscitation interventions, intravenous fluid quantity, intravenous fluid type, vasopressors, corticosteroids, and vitamin C–based interventions. Separate random-effects meta-analyses estimated pooled risk ratios (RRs) and 95% confidence intervals (CIs). We also performed a complementary landscape analysis using Python to evaluate the distribution of mortality effect estimates relative to study sample size as well as a structured thematic analysis to identify recurrent trial design constraints across therapeutic domains. Results: Across 60 RCTs (n = 29,337), pooled mortality effects demonstrated no mortality benefit across each of the six therapeutic domains: resuscitation interventions (RR 0.88, 95% CI 0.77–1.01), intravenous fluid quantity (RR 0.97, 95% CI 0.88–1.06), intravenous fluid type (RR 0.98, 95% CI 0.88–1.08), vasopressors (RR 0.96, 95% CI 0.88–1.05), corticosteroids (RR 0.92, 95% CI 0.84–1.00), and vitamin C-based interventions (RR 0.95 CI 0.83–1.09). Sensitivity analyses excluding studies at higher risk of bias did not alter the pooled estimates or interpretation of findings. A landscape analysis demonstrated progressive convergence to no mortality benefit with increasing study sample size across therapeutic domains. A structured thematic analysis using a prespecified trial-constraint framework identified recurrent design limitations across therapeutic domains, including delayed intervention timing, biologic and clinical heterogeneity, broad eligibility criteria, reliance on fixed-time mortality endpoints, treatment contamination, and trial design and infrastructure constraints. Conclusions Across major sepsis therapeutic interventions, RCTs have not demonstrated consistent reductions in mortality. Recurrent trial design limitations may contribute to these findings and can inform future trial design emphasizing earlier enrollment, biologic enrichment, adaptive platforms, and mechanism-aligned outcomes.</p>