Accuracy of susceptibility testing of Pasteurella multocida and Mannheimia haemolytica

The accuracy of antimicrobial susceptibility determinations relies on the bacterial species identification and the test methodology that is being used. In the present study, both aspects were investigated for 141 Pasteurella multocida and 34 Mannheimia haemolytica sensu lato isolates recovered from the upper respiratory tract of healthy calves. This was performed on the one hand by comparing of classical phenotyping with tDNA-PCR genotyping and on the other hand by pairwise comparison of disk diffusion with agar dilution results for seven antimicrobial compounds (ampicillin, ceftiofur, oxytetracycline, gentamicin, florfenicol, enrofloxacin, and the combination trimethoprim-sulfonamides). Phenotyping and genotyping correlated well (>90%). The pairwise comparisons of the susceptibility methods were investigated traditionally by means of error binding rates and sensitivity and specificity test characteristics. Obtained sensitivities (indication for absence of false susceptible results) were often lower than 85%, especially for older antimicrobial agents (oxytetracycline, gentamicin, trimethoprim-sulfonamides) and when M. haemolytica sensu lato was considered. Specificities (indication for absence of false-resistant results) exceeded 90% for almost all antimicrobial-bacterial combinations. The calculated test characteristics (sensitivities and specificities) were subsequently used in a second dataset of Pasteurellaceae from intensively (n = 99) and extensively housed calves (n = 196), to modify the apparent prevalence of antimicrobial resistance based upon disk diffusion results into an estimated true prevalence. It was concluded that the disk diffusion method is reliable in epidemiological studies like surveillance programs if resistance is sparse, whereas it needs to be interpreted with caution in situations where resistance is abundantly present.