Abstract
Bloodstream infections (BSIs) have become increasingly challenging to treat due to emerging antimicrobial resistance (AMR). As rapid administration of appropriate antimicrobials is crucial to positive patient outcomes, clinical alternatives to slow phenotypic antimicrobial susceptibility testing (AST) must be developed. We developed an automated real-time, colorimetric assay (ChroMIC) that phenotypically determines minimum inhibitory concentrations (MICs) directly from positive blood culture. Gram-negative bacteria within positive blood cultures from 83 patients were directly analyzed with ChroMIC. Both ChroMIC MICs and the much slower VITEK 2 MICs for each of seven antibiotics were compared to broth microdilution (BMD). Direct-from-blood-culture ChroMIC MICs were obtained without the ~18-hour growth and isolation steps needed to obtain both VITEK 2 and BMD MICs. ChroMIC assays yielded >90% categorical and essential agreement within 7 h from the start of the experiment and within 4 h of initial growth detection. ChroMIC minor, major, and very major errors were all comparable to, or in the case of VMEs, better than, those from VITEK 2 ASTs-the latter being performed only after an additional ~18-hour subculturing delay. As these patient blood samples were run in parallel with VITEK 2 MICs, a retrospective impact analysis on patient care suggests that ChroMIC could have facilitated faster therapeutic decision-making in nearly all cases, saving an average of >35 and >43 h for antimicrobial escalation and de-escalation, respectively, compared to VITEK 2. IMPORTANCE: The often sluggish pace of phenotypic antimicrobial susceptibility testing (AST), relative to sepsis progression, limits flexibility in altering patient treatment. We report a new direct-from-blood culture phenotypic AST that delivers excellent results within ~7 h of blood culture positivity. This rapid and accurate determination of optimal bloodstream infection treatment was compared in a retrospective study on patient blood cultures that flagged positive for Gram-negative pathogens. Although standard clinical methods were used to guide treatment, our testing was performed in parallel and could have enabled correction of the treatment course ~40 h earlier than was actually performed. Once clinically implementable, such innovative, low-labor, automated, and accurate susceptibility determinations hold great potential for improving patient outcomes and lowering overall healthcare costs.