Antimicrobial Susceptibility Testing of Non-Standard Direct-from-Blood Cultures: Optimization of Performance through Breakpoint Adjustment

Abstract

Abstract Disk diffusion susceptibility testing inoculated directly from positive blood culture bottles (direct Kirby-Bauer; DKB) allows for preliminary susceptibility results approximately a day sooner than standard antimicrobial susceptibility testing (AST). Our current laboratory workflow utilizes non-standardized inoculum from positive blood culture bottles. To minimize errors associated with inoculum variability, we report preliminary susceptibility results using breakpoints as follows: susceptible — standard-CLSI breakpoints; resistant— a cautious zone size of 6 mm; any zone between these extremes are reported as “pending”. DKB results are reported a median of 23.7 hours earlier compared to final sensitivity results. The decrease in turn-around time is consistent across organisms: DKB results for Enterobacteriales are reported a median of 21.6 hours earlier, Pseudomonas aeruginosa 20.8 hours earlier, and viridans group streptococci 21.9 hours earlier than final sensitivity reports. The portion of isolates that have actionable DKB results, however, is limited by a high portion of “pending” DKB results. In order to increase the portion of reportable and actionable DKB results, we sought to reduce the number of pending DKB results for isolates that were ultimately determined to be resistant by final AST. To accomplish this, we retrospectively compared the DKB zone sizes to the final microbroth susceptibility interpretation of ~800 bacterial isolates from the past 3.5 years and assessed the performance characteristics of DKB results across a range of zone sizes. To maximize the reporting of actionable data when antimicrobial escalation may be indicated, we specifically examined broad spectrum agents. Initial analysis focused on Enterobacteriales due to the number of isolates and distribution of resistance phenotypes. By adjusting the resistant cutoff of cefepime from 6mm to £14 mm, we found an increase in sensitivity of detecting resistant isolates from 23% to 100%, categorical agreement (CA) of 97%, no very major errors (VME), major errors (ME) 2.8%, and no minor errors (mE). For ceftriaxone, by changing the resistant breakpoint from 6mm to £17 mm, we found an increase in sensitivity from 43% to 100%, CA 98%, no VME, ME 1.3%, and mE 0.95%. For piperacillin-tazobactam, a change in the breakpoint from 6 mm to £15 mm resulted in an increase in sensitivity from 8% to 100%, CA 95%, no VME, ME 2.4%, and mE 1.6%. By adjusting the non-standard DKB breakpoints for reporting resistance to broad-spectrum beta-lactams in Enterobacteriales, we have increased the overall sensitivity of detecting resistant isolates to 100% for these agents without requiring extra steps of centrifugation and standardization of inoculum necessary to more completely adopt CLSI breakpoints. Retrospectively over the study period, implementing these changes could have resulted in reporting changes for 73 bacterial isolates from 59 unique patients and the corresponding potential for reduced time to appropriate therapy for patients with bacteremia.