Accounting for external events when analysing antimicrobial dispensing in the UK: An interrupted time series approach

Interrupted time series (ITS) analysis was applied to pooled antimicrobial prescribing data from all veterinary practices participating in a randomised controlled trial (RCT) in the UK to investigate how antimicrobial dispensing may be influenced by both study-related and external events. Dispensing rates in cats and dogs were modelled as average days treatment (DDDVet) per animal per month, adjusted for non-linear temporal variation and pre-existing trends. ITS identified significant changes in dispensing rates associated with timepoints coinciding with recruitment into the trial, and with UK national regulatory and legislative changes relevant to antimicrobial use. Post-hoc comparisons indicated that many of the changes in antimicrobial dispensing rates observed during the RCT period were also seen in practices outside the RCT used as contemporaneous comparators. These findings demonstrate how ITS can highlight and distinguish associations between trial outcomes and both study-related and external events. In addition, ITS can provide a structured method for interpreting trial outcomes with greater confidence by explicitly modelling overlapping events, collinearity, and autocorrelation. This approach may help future veterinary epidemiological studies using real‑world data to adjust for external events, particularly over long durations where contextual influences are unavoidable.