2026/07/01 by Evdoxia Kyriazopoulou, Konstantinos Patras, Symeon Metallidis +9
Medicine · #Neutropenia and Cancer Infections #Sepsis Diagnosis and Treatment #Nosocomial Infections in ICU
paper · doi:10.1016/j.ijantimicag.2026.107937
ABSTRACT Background The diagnostic accuracy and clinical benefit of molecular methods for pathogen identification and antimicrobial susceptibility testing were assessed through a systematic review and meta-analysis. Methods The protocol was prospectively registered with PROSPERO (CRD420251044218). A comprehensive search of MEDLINE (via PubMed), the Cochrane Library, and ClinicalTrials.gov was conducted to identify randomized controlled trials, non-controlled trials, and observational studies. The review investigated rapid diagnostic tests performed on blood samples for pathogen identification and susceptibility testing. The primary outcomes were Diagnostic Test Accuracy (DTA) for microbiological evaluation and time to pathogen identification, as well as to optimal/ adjusted/ appropriate antimicrobial treatment for clinical assessment. Secondary outcomes included 30-day mortality and total hospitalization cost. A random-effects model was used for quantitative synthesis. Results Of 1353 articles retrieved, 49 and 33 records were included in microbiological and clinical analysis. Pooled bivariate sensitivity for pathogen identification ranged from 0.97 to 0.98 (95% CIs approximately 0.94–0.99), with specificity consistently 0.99-1.00 across Gram-negative bacteria, Gram-positive bacteria, and yeasts. The mean time difference for pathogen identification was -21.43 hours (95% CI -29.15 to -13.72; P<0.0001; I 2 = 99%), and for optimal/ adjusted/ appropriate treatment -13.78 hours (95% CI -18.06 to -9.5; P<0.0001; I 2 = 97%). Total hospitalization costs were reduced; 30-day mortality was indifferent (OR 0.97; 95% CI, 0.88 to 1.07; I 2 = 23%). Conclusion Evidence indicates that early, actionable diagnostics, particularly those from positive blood cultures, are essential to effective sepsis care.