Comparison between broth microdilution and agar disk diffusion methods for antimicrobial susceptibility testing of bovine mastitis pathogens

In order to counter the antibiotic resistance phenomenon, a prudent and rational use of antimicrobials should be driven by an accurate clinical diagnosis and, when possible, by the isolation of the etiological agent followed by susceptibility testing, with the aim to select the most suitable molecule for therapy. Cow mastitis is considered the main cause of antibiotic use in the cattle breeding sector. The purpose of this study was to compare the broth microdilution (BMD) method performed with Sensititre Custom Plates and the agar disk diffusion (ADD) method in determining antimicrobial susceptibility of 215 isolates from bovine mastitis, including contagious pathogens (Staphylococcus aureus, Streptococcus agalactiae) and environmental (Streptococcus uberis, Streptococcus dysgalactiae, Enterococcus spp., Escherichia coli, Serratia marcescens, Klebsiella pneumoniae). We compared results of the following antimicrobials: amoxicillin/clavulanic acid, ampicillin, cefazolin, ceftiofur, enrofloxacin, erythromycin, kanamycin, oxacillin, penicillin, pirlimycin, rifampicin and trimethoprim/sulphonamides. We applied MIC breakpoints and zone diameter breakpoints as recommended by CLSI and EUCAST. MIC and disk diffusion diameters were compared for 1839 microorganism/antimicrobial combination and discrepancies between the two methods were classified as very major discrepancy (VMD), major discrepancy (MD) and minor discrepancy (MiD). The overall agreement between the two methods was found to be 80.7% with a Cohen's kappa coefficient of 0.397, thus indicating a fair concordance. BMD method and ADD method demonstrated a satisfactory agreement (89 to 100%) for S. aureus and S. marcescens and all antimicrobial agents tested. Low agreement was observed for S. uberis and rifampicin (20%), enrofloxacin (49%), penicillin (51%) and pirlimycin (52%), E. coli and ampicillin (20%), S. dysgalactiae and enrofloxacin (44%), S. agalactiae and rifampicin (25%). A possible explanation for the discrepancies detected could be found in the breakpoints used which, sometimes, are not specific for the tissue-matrix of isolation/animal species/pathogen agent. The majority of the discrepancies found were MiD and MD, revealing a higher restrictiveness of the BMD method, while VMD represented only 0.2% of the total observations, a comforting fact since this type of error may result in treatment failure.