The Potential for Sample Testing at the Pen Level to Inform Prudent Antimicrobial Selection for Bovine Respiratory Disease Treatment: Investigations Using a Feedlot Simulation Tool.

Background: Antimicrobial drugs are used to treat bacterial diseases in livestock production systems, including bovine respiratory disease (BRD) in feedlot cattle. It is recommended that therapeutic antimicrobial use (AMU) in food animals be informed by diagnostic tests to limit the emergence of antimicrobial resistance (AMR) and preserve the effectiveness of available drugs. Recent evidence demonstrates preliminary support for the pen as a prospective target for AMR testing-based interventions in higher-risk cattle. Methods: A previously reported agent-based model (ABM) was modified and then used in this study to investigate the potential for different pen-level sampling and laboratory testing-informed BRD treatment strategies to favorably impact selected antimicrobial stewardship and management outcomes in the western Canadian context. The incorporation of sample testing to guide treatment choice was hypothesized to reduce BRD relapses, subsequent AMU treatments and resultant AMR in sentinel pathogen Mannheimia haemolytica. The ABM was extended to include a discrete event simulation (DES) workflow that models the testing process, including the time at sample collection (0 or 13 days on feed) and the type of AMR diagnostic test (antimicrobial susceptibility testing or long-read metagenomic sequencing). Candidate testing scenarios were simulated for both a test-only control and testing-informed treatment (TI) setting (n = 52 total experiments). Key model outputs were generated for both the pen and feedlot levels and extracted to data repositories. Results: There was no effect of the TI strategy on the stewardship or economic outcomes of interest under baseline ecological and treatment conditions. Changes in the type and number of uses by antimicrobial class were observed when baseline AMR in M. haemolytica was assumed to be higher at feedlot arrival, but there was no corresponding impact on subsequent resistance or morbidity measures. The impacts of sample timing and diagnostic test accuracy on AMR test positivity and other outputs were subsequently explored with a theoretical "extreme" BRD treatment protocol that maximized selection pressure for AMR. Conclusions: The successful implementation of a pen-level sampling and diagnostic strategy would be critically dependent on many interrelated factors, including the BRD treatment protocol, the prevalences of resistance to the treatment classes, the accuracy of available AMR diagnostic tests, and the selected "treatment change" thresholds. This study demonstrates how the hybrid ABM-DES model can be used for future experimentation with interventions proposed to limit AMR risk in the context of BRD management.