Rapid Antimicrobial Susceptibility Testing Using Direct Centrifugation Significantly Reduces Turnaround Time While Maintaining High Diagnostic Accuracy in Bloodstream Infections

Abstract

Background/Objectives: Bloodstream infection-associated sepsis and septic shock remain major causes of in-hospital mortality. Because conventional antimicrobial susceptibility testing (AST) requires 48–72 h, earlier susceptibility results may facilitate timely antimicrobial optimization. This study evaluated the performance and clinical utility of a direct centrifugation-based AST method compared with the standard workflow. Methods: A total of 328 monomicrobial positive blood culture isolates, comprising 168 Gram-negative bacteria and 160 Gram-positive cocci, were analyzed. Essential agreement (EA), categorical agreement (CA), very major errors (VMEs), major errors (MEs), minor errors (mEs), and turnaround time (TAT) were compared between the two methods. Results: For Gram-negative bacteria, the direct method achieved an EA of 97.77% and CA of 98.53%, with VME, ME, and mE rates of 0.81%, 0.14%, and 0.60%, respectively. Particularly high concordance was observed for Escherichia coli and Klebsiella pneumoniae. For Gram-positive cocci, the EA and CA were 93.29% and 95.91%, respectively, with VME, ME, and mE rates of 3.47%, 1.08%, and 0.94%, respectively. Although overall agreement remained high, the GPC VME rate slightly exceeded the predefined 3% criterion, and elevated species-specific VME rates were observed for several organisms. The direct centrifugation method significantly reduced the mean TAT from 48.74 to 25.13 h, corresponding to an approximately 23.61 h reduction. Conclusions: The direct centrifugation method substantially shortened AST reporting time while maintaining high overall agreement with the standard workflow, particularly for Gram-negative bacteria. Further validation is warranted for Gram-positive and organism–antimicrobial combinations associated with elevated VME rates before routine clinical implementation.

The authors' abstract, as published at the source. Diagnostics, 2026 · DOI ↗