Use of normalised resistance analyses to set interpretive criteria for antibiotic disc diffusion data produce by Aeromonas spp

Abstract

Normalised resistance interpretive (NRI) analysis was applied to the development of interpretive criteria for antibiotic disc diffusion zones of mesophilic Aeromonas spp. Minimum quantitative and qualitative properties of data sets from which normalised resistance interpretation could generate acceptable cut-off values were established by examining published data that had been used to set the epidemiological cut-off values for bacteria of human and aquatic interest. Applying these criteria to disc diffusion data generated in a previous study of 129 mesophilic Aeromonas spp. isolated from pet fish, demonstrated that normalised resistance interpretation was capable of generating acceptable cut-off values for seven (erythromycin ≥ 8 mm, gentamicin ≥ 18 mm, moxalactam ≥ 32 mm, oxytetracycline ≥ 21 mm, streptomycin, ≥ 20 mm tetracycline ≥ 28 mm, and trimethoprim/sulfamethoxazole ≥ 26 mm) of the agents tested. This analysis demonstrated, with respect to these agents, that a single set of interpretive criteria could be developed for application to data generated from all the Aeromonads studied. Provisional cut-off values were generated for three other agents (chloramphenicol ≥ 35 mm, florfenicol ≥ 35 mm and furazolidone ≥ 23 mm). The distribution of zone sizes for these agents showed an apparent bimodality in the high-zone end. As a result, the question of whether, for these three agents, a single set of interpretive criteria could be validly set for all Aeromonads could not be resolved. With respect to the five quinolone agents studied, a cut-off value could only be estimated for oxolinic acid. Because of the high frequency of resistance to quinolones in the strain set, this value (≥ 32 mm) should be treated as a provisional estimate. With respect to some agents, strains manifesting a low-level resistant phenotype, zones ≤ 8 mm smaller than the wild-type cut-off, represented a significant proportion of the strains classified as non wild-type. The difficulties that these frequencies of low-level resistance would present for attempts to set and apply universal, laboratory-independent interpretive criteria are discussed.