Performance of molecular tests for diagnosis of bloodstream infections in the clinical setting: a systematic literature review and meta-analysis.

Abstract

Background: Rapid identification of bloodstream pathogens and associated antimicrobial resistance (AMR) profiles by molecular tests from positive blood cultures (PBCs) have the potential to improve patient management and clinical outcomes.

Objectives: A systematic review and meta-analysis were conducted to evaluate diagnostic test accuracy (DTA) of molecular tests from PBCs for detecting pathogens and AMR in the clinical setting.

Methods: .

Data sources: Medline, Embase, Cochrane, conference proceedings, and study bibliographies were searched.

Study eligibility criteria: Studies evaluating DTA of commercially available molecular tests vs. traditional phenotypic identification and susceptibility testing methods in patients with PBCs were eligible.

Participants: Patients with PBCs.

Tests: Commercially available molecular tests.

Reference standard: Traditional phenotypic identification and susceptibility testing methods (standard of care, SOC).

Assessment of risk of bias: Study quality was assessed using Quality Assessment of Diagnostic Accuracy Studies-2.

Methods of data synthesis: Summary DTA outcomes were estimated using bivariate random-effects models for gram-negative bacteria (GNB), gram-positive bacteria (GPB), yeast, GNB-AMR, GPB-AMR, and specific targets when reported by ≥ 2 studies (PROSPERO CRD42023488057).

Results: Seventy-four studies including 24 590 samples were analysed, most of which had a low risk of bias. When compared with SOC, molecular tests showed 92-99% sensitivity, 99-100% specificity, 99-100% positive predictive value, and 97-100% negative predictive value for identifying total GNB (43 studies), GPB (38 studies), yeast (24 studies), GNB-AMR (35 studies), and GPB-AMR (39 studies). For individual pathogen targets, 93-100% sensitivity, 98-100% specificity, 86-100% positive predictive value, and 99-100% negative predictive value were estimated. Five of seven AMR genes had 91-99% sensitivity and 99-100% specificity. Sensitivity was lower for IMP (four studies; 62%; 95% CI, 34-83%) and VIM (four studies; 70%; 95% CI, 38-90%) carbapenemases, where genes were not detected or were not harboured in Pseudomonas aeruginosa (i.e. low prevalence). Performance of molecular tests in detecting AMR was generally comparable when grouped by geographical region (Europe, North America, and East Asia).

Discussion: High DTA support the use of molecular tests in identifying a broad panel of pathogens and detecting AMR in GNB and GPB.