Back

Accuracy of three automated systems for susceptibility testing selected Gram-negative bacilli against five broad-spectrum beta-lactam agents

Abstract number: 1134_04_221

Sader H., Fedler K., Cookson B., Brace C., Fritsche T., Jones R.

Objectives:  

To evaluate the accuracy of three automated systems; MicroScan WalkAway (MSWA), Vitek 2 (VT2) and Vitek Legacy (VTL), for susceptibility (S) testing P. aeruginosa (PSA) and various Enterobacteriaceae (ENT) species against aztreonam (AZT), cefepime (CPM), ceftazidime (CAZ), imipenem (IMP) and piperacillin/tazobactam (P/T).

Methods:  

Recent clinical strains (100 PSA and 20 ENT) from North American hospitals were selected to over-represent isolates with CPM and P/T MIC values within ±1 log 2 dilution of the current NCCLS S and resistant (R) breakpoints for the studied compounds. Categorical results from automated systems were compared to the consensus of 3 reference/standardized methods: broth microdilution, E-test (AB BIODISK, Solna, Sweden) and disk diffusion. Categorial disagreements were classified as: in very major (VM, false-S), major error (MA, false-R) and minor errors (MI; involving the intermediate category).

Results:  

The consensus testing S/R rates (%) among PSA strains were 45/55 for P/T, 46/40 for CAZ, 48/35 for CPM, 69/27 for IMP and 36/35 for AZT. A summary of the categorial disagreements for PSA is shown in the Table. All three systems showed a high, unacceptable rate of VM for P/T (21–32%). For other drugs VM rates ranged from 0 to only 4% (IMP tested on VT2). MA rates were acceptable for comparisons (0–3%) and MI rates were generally elevated (5–33%), reflecting the high proportion of consensus results within the intermediate category or skewed, erroneous results for CPM (VT2 and VTL) and AZT (all systems). Among ENT (20 strains), VM was 5% for AZT and 0% for P/T, CAZ, CPM and IMP (100% S) on all 3 systems, and MA was detected only on MSWA (0–15%). MI rates were higher for P/T [20 (VT2) to 35% (MSWA and VTL)] and CPM [15 (VTL) to 35% (MSWA)].

 Error rates (%)
System/Error typeAZTCPMCAZIMPPIT
MSWA     
VM000121
MA01111
MI33141111-
VTL     
VM301026
MA00100
MI3222125-
VT2     
VM022432
MA10310
MI28231212-

Conclusions:  

The results of this study demonstrates that the automated systems (MSWA, VT2 and VTL) generally failed to accurately detect P/T-R among PSA. The criteria used to select the strains (MIC values close to breakpoints or within R range) increased the sensitivity of detecting significant categorical disagreements, dominated by MI. Re-evaluation of the P/T testing for PSA would be prudent for these systems to minimize adverse therapeutic outcomes worldwide.

Session Details

Date: 01/08/2007
Time: 00:00-00:00
Session name: XXIst ISTH Congress
Subject:
Location: Oxford, UK
Presentation type:
Back to top