Development and evaluation of a direct disk diffusion, rapid antimicrobial susceptibility testing method from blood culture positive for Gram-negative bacilli using rapid molecular testing and microbiology laboratory automation

Rapid antimicrobial susceptibility testing (RAST) using disk diffusion (DD) from positive blood cultures (BC) can aid initiation of effective antimicrobial therapy. Rapid molecular blood culture identification panels (BCID) and microbiology laboratory automation (MLA) provide an opportunity to optimize RAST. This study aimed to evaluate and optimize RAST in combination with BCID and MLA for <i>Escherichia coli</i>, <i>Klebsiella pneumoniae,</i> and <i>Pseudomonas aeruginosa</i> positive BC. Monomicrobial BC positive for <i>E. coli</i>, <i>K. pneumoniae,</i> and <i>P. aeruginosa</i> by BCID between February 2020 and June 2021 were included. Images were captured by the MLA system after 4, 6, and 8 hours of incubation. The European Committee on Antimicrobial Susceptibility Testing (EUCAST) RAST breakpoints were used to interpret the zones of inhibitions measured by the WASPLab Halo Recognition software. Categorical agreement (CA) and error rates (minor [mE], major [ME] and very major [VME] errors) were calculated using automated microbroth dilution (aMBD) and DD as comparator methods. Ninety-one BCs met the inclusion criteria. CA between RAST and aMBD was >90% for most antimicrobials tested at all three time points for <i>E. coli</i> and <i>K. pneumoniae</i>. Overall, mE rates were <4%, ME rates ranged from 1% to 6% and VME rates ranged from 2% to 9%. Performance of RAST was lowest for piperacillin/tazobactam (71%-86%). For <i>P. aeruginosa,</i> CAs were mostly <90% with significant error rates. Combining rapid organism identification by BCID with automated RAST set-up and interpretation can facilitate the generation of early, preliminary antimicrobial susceptibility results for <i>E. coli</i> and <i>K. pneumoniae</i>.IMPORTANCERapid antimicrobial susceptibility testing (RAST) for Gram-negative bacilli can quickly aid in the optimization of therapy for patients with bacteremia. In this study, a RAST workflow that includes identification using a molecular blood culture identification (BCID) panel, followed by manual set up and incubation on microbiology laboratory automation (MLA) with automated reading of zones of inhibition, was developed and evaluated. This workflow could provide AST results as soon as 6 hours after blood cultures flagged positive with categorical agreements > 90% for most antimicrobials tested. Additionally, the software used for automated measurements of the zones of inhibition provided accurate readings in 75% of the recorded measurements.