Retrospective evaluation of colistin-resistant isolates in automated system by gradient diffusion method and broth microdilution method

Abstract

Objective: Optimizing colistin susceptibility testing has difficulties because of its high molecular weight and high binding capacity to polystyrene which is frequently used in antibiotic susceptibility testing. We aimed to compare the results of isolates, which were detected as colistin-resistant in the automated system, obtained by using broth microdilution (BMD) method which is the gold standard, with gradient diffusion method (GDM).Methods: We investigated 36 Klebsiella pneumoniae, 9 Acinetobacter baumannii and 5 Pseudomonas aeruginosa isolates, identified by the Phoenix (TM) 100 (Becton Dickinson, Sparks, MD, USA) automated system, isolated from various clinical specimens sent to the Central Microbiology Laboratory between August 2016 and April 2017. The susceptibility of the isolates was also tested by GDM and BMD method.Results: When the colistin resistance rates obtained from the gold standard BMD method were compared with the automated method, the categorical agreement (CA) rate of the automated system was 92% for all isolates, 100% for K. pneumoniae, 77.8% for A. baumannii, and 60% for P. aeruginosa. The very major error (VME) rate was 0%, and the major error (ME) rate was 8% for all isolates. When GDM was used for all isolates, CA was found to be 20% for all isolates, 16.7% for K. pneumoniae, 22.2% for P. aeruginosa and 40% for A. baumannii. VME was found to be 80%, and ME was %0 for all isolates.Conclusions: CA, VME and ME rates of Phoenix (TM) 100 for detecting colistin resistance is within acceptable limits according to ISO 20776 standard, but the rates of GDM is not suitable for this purpose.