Measuring antimicrobial use on dairy farms: A method comparison cohort study

Antimicrobial use on UK dairy farms is measured for surveillance purposes, with veterinary sales data as a proxy for use. Two other methods of recording use have been used commonly on-farm: medicine waste bins and farm medicine records. However, none of these methods has been validated to measure antimicrobial use. The objective of this research was to assess agreement between the 3 most common methods for measuring on-farm antimicrobial use with a predetermined reference method on UK dairy farms. Antimicrobial use was measured prospectively on 26 UK dairy farms using medicine waste bins into which participants placed all discarded medicine packaging for a 12-mo period. At the end of 12 mo, farm medicine records and veterinary sales data were obtained retrospectively for participating farms. The reference method used was based on pre- and poststudy inventories combined with veterinary sales data. We investigated the systematic difference between the mean on-farm antimicrobial use measured by each of the 3 methods and a reference method, using one-way repeated-measures ANOVA models. Reliability and clinical relevance of the agreement between each pair of methods was quantified using the concordance correlation coefficient (CCC) and the Bland-Altman method, respectively. When compared with the reference method, veterinary sales data had excellent reliability for injectable antimicrobials and intramammary antimicrobials [95% confidence interval (CI) of CCC > 0.90] and moderate to excellent reliability for other antimicrobials (95% CI of CCC: 0.68-0.97). Medicine waste bins had good to excellent reliability for injectable (95% CI of CCC: 0.84-0.99), and intramammary products (95% CI of CCC: 0.78-0.94) and no agreement for other forms of antimicrobial. Farm medicine records did not agree for any form of antimicrobial when compared with the reference method. The use of veterinary sales data as a proxy for on-farm antimicrobial use in the UK represented excellent statistical reliability and offered clinically good agreement with the reference method when used to measure injectable antimicrobials. This study applies to the UK context and included a relatively small number of farms. However, these results have research and policy implications, both nationally and internationally, and are essential in accurately quantifying agricultural antimicrobial use to inform both animal and human health.