Presentation (CTI)

OBJECTIVES: Rapid diagnostic platforms can shorten time-to-optimal therapy in bloodstream infections (BSI), but standard decision-tree and Markov models may inadequately represent the continuous antibiotic exposure, sequential diagnostic switching, and pathogen-specific pathways involved in their evaluation. This study describes a discrete-event simulation (DES) developed to evaluate the cost-effectiveness of DRAiGON, an AI-guided whole-genome sequencing identification and antimicrobial susceptibility testing platform compared with conventional diagnostics for suspected BSI caused by ESKAPEE pathogens, comprising Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, Enterobacter spp., and Escherichia coli.
METHODS: A DES was developed in R (Simmer package) to track individual patients from blood culture collection through microbiological workup to hospital discharge or death. Three features distinguish it from published diagnostic CEAs. First, a three-phase antibiotic stewardship pathway is explicitly modelled: gram-stain-directed therapy at culture positivity, species-directed therapy following identification, and susceptibility-guided therapy following AST, with the DRAiGON arm collapsing phases two and three into a single simultaneous result and clinical decision point. Second, antibiotic costs accrue continuously and reset at each clinical decision point, enabling attribution of cost and usage differences to diagnostic timing. Third, mortality is modelled from baseline setting-specific risk, empiric therapy appropriateness, and a per-day delay multiplier to optimal therapy, calibrated to published time-sensitive mortality estimates. The model incorporates 12 ESKAPEE pathogen categories with source-stratified distributions of suspected infections.
RESULTS: In the preliminary base-case analysis, DRAiGON dominated conventional microbiological diagnostics, yielding incremental costs of −€50.80 and incremental QALYs of 0.0058 per patient. Cost reductions were driven primarily by shorter hospital length of stay, while QALY gains reflect averted mortality attributable to earlier initiation of optimal therapy.
CONCLUSIONS: This DES framework embeds continuous antibiotic stewardship dynamics and time-sensitive mortality within a patient-level model, offering a methodological basis for evaluating rapid diagnostics whose value depends on the timing and sequence of clinical decisions