Search

Background: Dependence upon azithromycin in the treatment of enteric fever is increasing, particularly with the emergence of Salmonella typhi strains with extended spectrum β-lactamase activity and the already high prevalence of quinolone resistance. Accurate determination of azithromycin susceptibility is crucial and underlined by recent reports of azithromycin resistance. We investigated concerns of discordance in azithromycin susceptibility estimation between local and reference laboratories. Methods: Retrospective audit of isolates from patients attending a central London hospital with enteric fever (May 2011-April 2019). Estimations of azithromycin and ciprofloxacin MICs by the local and reference laboratories were compared. Genomic data and laboratory practices were reviewed. Results: In isolates with matched clinical and reference laboratory MICs (n=19), there was poor inter-laboratory concordance: 5/19 MICs concordant (weighted κ = 0.190, adjusted for concordance within 1 log2 dilution); susceptibility interpretation concordant in 8/19 (κ=0). All isolates reported locally as resistant were found to be sensitive by the reference laboratory. No azithromycin resistance genes were detected. By contrast, for ciprofloxacin: 13/18 MIC gradient strip results concordant (weighted κ=0.823); susceptibility interpretation concordant in 17/18 (κ=0.85). Of the possible sources of variation identified, we believe that variable interpretation of “trailing edge” MIC estimation was key, mitigated in the reference laboratory by a “second reader” system. Conclusions: There is marked variation in azithromycin MIC gradient strip reporting between a local laboratory and the national reference laboratory, particularly over-reporting of resistance by the local laboratory. We would advise clinical laboratories to review their experience and consider adopting a “second reader” system.