Influence of local epidemiology on the performance of common colistin drug susceptibility testing methods

Objectives
To determine the influence of local spread of clonal strains and testing of follow-up isolates on categorical (CA) and essential agreement rates (EA) of common colistin (COL) drug susceptibility testing methods with the broth microdilution (BMD) reference method.
Methods
COL MICs were determined for 178 bacterial isolates (Enterobacteriaceae, n = 97; Pseudomonas aeruginosa, n = 81) collected within one year from 64 patients by BMD according to ISO standard 20776–1 (reference method), the SensiTest BMD panel (ST), agar dilution (AD), the VITEK 2 instrument, and gradient diffusion (GD) using antibiotic strips of two and Muller-Hinton agar plates of three manufacturers. CA and EA with BMD were calculated for all isolates and compared to the subset of 68 unique isolates.
Results
CA ranges were 79.4% to 94.1% for the unique isolateq panel and 89.9% to 96.1% for all tested isolates. EA ranges were 64.7% to 86.8% and 67.4% to 91.0%, respectively. In both panels, EA for all GD assays was lower than 90%. Both lower and higher EA values ranging from—18.3% (MTS on BD agar) to + 6.3% (AD, Vitek 2) were observed in the full one-year sample. Acquisition of colistin resistance under therapy was observed for 3 patients.
Conclusions
i) Repeat testing and local spread of clonal strains can positively or negatively affect CA and EA, ii) CA is more robust towards local influences than EA, iii) EA of GD and AD methods for COL with the reference BMD method is insufficient.