Contrasting Treatment- and Farm-Level Metrics of Antimicrobial Use Based on Used Daily Dose vs. Defined Daily Dose for the German Antibiotics Minimization Concept.

Abstract

The German Antibiotics Minimization Concept defines a farm-level benchmarking process based on half-yearly treatment frequencies that applies to six animal populations for fattening: calves (≤ 8 months), cattle (> 8 months), piglets (≤ 30 kg), pigs (> 30 kg), broiler chickens, and turkeys. The treatment frequency defined in the Minimization Concept takes into account the number of animals treated, the treatment duration, and the number of active antimicrobial ingredients, for each individual treatment, and is equivalent to a treatment frequency based on the used daily dose (UDD) and the actual weight of the animals at the time of treatment. With data from the German benchmarking system for the seven half-year periods from the second semester 2014 to the end of 2017, we compared UDD-based metrics of antimicrobial use (AMU) at the treatment and the farm level with metrics based on defined daily doses for animals (DDDvet) and standardized animal weights assumed at the time most likely for treatment. We show the extent to which DDDvet-based metrics would introduce errors into the measurement of AMU at the treatment level and consequently at the farm level. To that end, we introduce the average animal daily dose ratio, an aggregate measure that quantifies how choice of antimicrobial substances, deviations of used doses from recommended doses, of recommended doses from DDDvet values, and of treatment weights from standardized weights, affect a farm's AMU metrics. Our results show that in all animal populations considered benchmarking processes would become less successful at targeting high users. This would be especially true for broiler chickens and turkeys where the relative weight gain during fattening is the largest and overdosing appears to be common practice. Therefore, in AMU monitoring systems with the purpose of benchmarking farms UDD-based metrics are preferable to DDDvet-based metrics.