Novel characterisation of dairy herds in Wales: A description of principal herd typologies and antimicrobial use patterns

Antimicrobial resistance (AMR) is one of the top global public health concerns. Reducing and refining antimicrobial use (AMU) in farmed livestock is vital in slowing the development of AMR and preserving the efficacy of antimicrobials (AMs) in both humans and animals. Understanding the risk factors for AMU, however, is crucial to informing sustainable and effective farm AMU reduction and prudent use strategies. As a range of farm-level variables are likely to impact AMU, multidimensional exploratory analyses play a pivotal role in identifying direct and indirect factors influencing variation in AMU typically observed between dairy herds. This study used exploratory approaches to investigate whether typologies of dairy herds could be determined on the basis of farm characteristics, health parameters and management practices, and whether these herd types were differentiated by AMU. This study was conducted on 21 dairy herds in Wales, United Kingdom. Comprehensive surveys were administered and 224 usable variables regarding farm characteristics, herd health parameters and management practices were collated. AM sales data for each herd were used as a proxy for AMU. Multiple correspondence analysis (MCA) and hierarchical clustering on principal components (HCPC) were performed. The top 10 dimensions yielded by MCA explained 65.7 % of the total variance. Two data-driven typologies of dairy herds, produced from the first two cut-points of the HCPC dendrogram, were visualised and described. Here, five partitions of relatively homogeneous herds (herd types) were characterised and contrasted by 73 variable categories. Herd types were primarily constructed by variables focused on drying-off practices (and use of intramammary [IMM] AMs), herd size, stock purchasing and culling rates in addition to those concerning husbandry, disease management, grazing practices and veterinarian contact. Herd types characterised by performing blanket dry cow therapy (BDCT) used a higher mass of dry cow IMM AMs, EMA Category C and B AMs and had higher medium total AMU (mg/PCU) compared to herd types characterised by performing selective dry cow therapy (SDCT). From this study help untangle the myriad of factors influencing AMU at herd level and provide insight into the challenges of good AM stewardship. Strategies for sustainable reductions in AMU should be directed toward specific herd types identified, such as targeted interventions to implement SDCT. Multivariate exploratory approaches of dimensionality reduction and clustering are invaluable in elucidating the risk factors for AMU when utilising high-dimensional datasets. Future prospective studies are needed to validate herd types and confirm causality of findings.