Abstract
The performance of a disk diffusion test using broth from positive blood cultures as inoculum (direct disk diffusion [dDD]) was evaluated for a collection of 20 challenge isolates of Enterobacteriaceae, Acinetobacter baumannii, and Pseudomonas aeruginosa Isolates seeded into human blood were inoculated into Bactec Plus Aerobic/F, VersaTREK Redox 1, and BacT/Alert FA Plus bottles and incubated in the respective automated blood culture systems. Disk diffusion results were compared to reference disk diffusion results. Categorical agreement (CA) values for dDD, after removal of random errors due to natural MIC variation, were 87.8%, 88.4%, and 92.2% for the BacT/Alert, Bactec, and VersaTREK systems, respectively. No very major errors (VME) were observed, and major error (ME) rates were 3.0%, 2.3%, and 1.7%, respectively. Incubation of the dDD test samples for 6 h compared to incubation for 16 to 18 h resulted in 19.9% of tests having too light of growth to allow reading of zones of inhibition. Among the evaluable dDD tests, CA values were 58.9%, 76.6%, and 73.2% for the isolates seeded into the BacT/Alert, Bactec, and VersaTREK systems, respectively. VME rates for isolates seeded into these systems were 2.2%, 1.8%, and 3.0%, respectively, and ME rates were 25.4%, 6.1%, and 2.8%, respectively, at the 6-h reading. The best performance of dDD was found for blood cultures with bacterial concentrations in the range of 7.6 × 10(7) to 5.0 × 10(8) CFU/ml; CA values ranged from 94.7 to 96.2% for these concentrations after 18 h of incubation and from 76.9 to 84.1% after 6 h of incubation. These preliminary data demonstrate the potential accuracy of dDD testing by the clinical laboratory.