Background: Farm-level antimicrobial stewardship in swine production requires indication-specific knowledge of treatment patterns and the herd-level features associated with them. Methods: We analyzed questionnaire-based data collected in 2015 from 13 Hungarian swine farms covering 15,725 sows and their progeny. The survey captured production indicators, pathogen occurrence, vaccination, resistance-testing practices, drug costs, and disease-group-specific antimicrobial use. As a separate, non-mergeable descriptive temporal comparator, we also considered independent digital farm-monitoring data from three large-scale swine herds from 2022 to 2024. Results: The most frequently reported pathogens were Mycoplasma hyopneumoniae (13/13 farms), Lawsonia intracellularis (12/13), Escherichia coli (12/13), swine influenza virus (11/13), and Streptococcus suis (10/13). S. suis ranked as the leading damaging pathogen on 69% of farms. Among farms with antibiotic cost data (9/13), antibiotics accounted for a mean of 31.8% of veterinary drug expenditures. Among farms with treatment-by-indication data (8/13), the highest relative frequency of reported treatment events was linked to porcine respiratory disease complex, where doxycycline represented 38% of reported PRDC treatment events. Colistin dominated E. coli-associated diarrhea control, whereas beta-lactams were central for S. suis-related disease. In the 2022–2024 comparator dataset, enteric and respiratory disorders and arthritis remained the main recorded health problems, but corrected antimicrobial use was markedly lower in the later dataset. Conclusions: Antimicrobial use showed clear disease-group-specific patterns, supporting syndrome-focused stewardship rather than generic reduction targets.
Internal biosecurity may be associated with antimicrobial use (AMU) in pig production, but observational associations, predictions and explanations of fitted models answer different questions. We analysed 1110 farm–month–age-group units from 18 Hungarian pig farms; an original post hoc restriction retained 939 units from 13 farms. The continuous endpoint was the natural logarithm of one plus a defined AMU index formed by summing substance-specific active ingredient mass-to-recorded-weight ratios. The study-specific biosecurity instrument was used for within-farm monitoring and was not externally validated. We retained a 2 × 2 matrix of unfiltered/filtered and composite/separate-variable models, with categorical animal-group separation, age-group adjustment and farm-cluster uncertainty. In the unfiltered separate-variable model, recorded separation code 2 versus code 0 was associated with a higher log AMU index (β = 0.205; CR1-t 95% CI 0.013–0.397; p = 0.037), although the wild-cluster p-value was 0.116 and sensitivity estimates varied with adjustment. Code 3 binary contrasts were too sparse for precise effect quantification. Four ML families were evaluated using nested farm-grouped and leave-one-farm-out validation. For the inner-selected biosecurity-plus-age procedure, the unfiltered pooled R2 was −0.101 and −0.187, respectively. SHAP retained an exploratory role in explaining held-out tree model predictions. The results identify model-dependent observational patterns but do not establish a causal biosecurity effect or reliable prediction for new farms.