Accuracy of Direct Antimicrobial Susceptibility Testing of Gram-Negative Bacteria from Positive Blood Cultures Using MicroScan System and Value of Using Expert Rules for β-Lactam Agents.

Direct antimicrobial susceptibility testing (AST) of positive blood cultures with Gram-negative bacteria produces results within 24 h, compared to 48 to 96 h with conventional methods. Positive clinical blood cultures were studied, supplemented with contrived blood cultures inoculated with a spectrum of resistant isolates. Bacterial inocula used for direct AST were quantitated. Direct AST was performed using MicroScan NM43 trays inoculated directly from positive blood cultures (100 μL in 25 mL water) and incubated using a WalkAway instrument, with trays read after 16 h. Reference AST was performed the following day from growth on solid medium using the same trays. Agreement of AST results between direct and reference methods, with and without the use of three expert rules for β-lactams, was evaluated using FDA categorical agreement criteria. Of 86 specimens tested (41 clinical specimens and 45 contrived specimens), the mean bacterial load in positive blood cultures was 8.98 log10 CFU/mL. Fifteen isolates contained extended-spectrum β-lactamases, and 27 contained carbapenemases. Of 1,985 pairs of AST categorical results for 25 antimicrobials, 55.0% were susceptible, 4.7% intermediate, and 40.4% resistant by reference testing. Overall categorical agreement was 92.3%, with 5.3% minor errors, 1.9% major errors, and 0.4% very major errors. Agreement was higher for non-β-lactam agents (95.8%) than for β-lactam agents (90.3%; P < 0.0001). Application of expert rules increased agreement for β-lactam agents to 94.6%. The methods used achieved the study goal of producing accurate, cost-effective AST results directly from positive blood cultures using MicroScan trays with a 16-h incubation time without the need for additional testing. Use of three expert β-lactam rules improved accuracy.