Research Summary - 3
Reduced animal use in a trial evaluating the efficacy of reduced tylosin phosphate treatment duration for preventing liver abscess
| Date/Time: |
9/13/2025
10:45
|
| Author: |
Paul S Morley |
| Clinic: |
VERO Program, Texas A&M University |
| City, State, ZIP: |
CANYON, TX 79015 |
J.D. Young, PhD, PAS
1
;
P. Morris, PhD
3
;
T.C. Bryant, PhD, PAS
4
;
A. O'Connor, BVSc, DVM
5
;
C. Wang, PhD
3
;
B. Weinheimer, MS
6
;
J.T. Richeson, PhD, PAS
2
;
P.S. Morley, DVM, PhD, DACVIM
1
;
1VERO Program, Texas A&M University, Canyon, TX, 79015
2West Texas A&M University, Canyon, TX, 79016
3Iowa State University, Ames, IA
4Five Rivers Cattle Feeding, Johnstown, CO
5Michigan State University, East Lansing, MI
6Private Consultant, Guelph, ON, Canada
Introduction:
Liver abscesses (LAs) represent a significant economic burden in North American feedlot cattle production, with tylosin phosphate supplementation serving as the primary preventive strategy. However, growing concerns about antimicrobial resistance necessitate evidence-based approaches to optimize and potentially reduce antimicrobial usage. Therefore, we conducted a 2-phase study, using a randomized controlled trial designed to supplement a Bayesian network meta-analysis (NMA) to test bioequivalence of different durations of tylosin supplementation for the control of severe LAs. This investigation demonstrates how Bayesian network meta-analysis can enhance research efficiency by leveraging existing data to enhance statistical power and reduce animal use in new trials.
Materials and methods:
The randomized trial enrolled 16,728 cattle across four treatment arms: negative control (no tylosin), continuous supplementation throughout finishing, and tylosin withdrawal at 84 or 56 days pre-harvest (allocated 1:1:2:2). This primary trial was specifically designed to complement existing literature data, demonstrating how Bayesian approaches can dramatically reduce required sample sizes while maintaining statistical rigor. The network meta-analysis synthesized our trial data with 32 previous controlled trials that included tylosin supplementation and untreated control groups in the evaluation of methods for controlling LA; these trials enrolled a total of 72,086 cattle and featured 10 distinct interventions, creating a comprehensive evidence network.
Results:
Results of NMA of all data, including the primary trial, indicated that continuous tylosin administration reduced the average odds of severe liver abscess occurrence when compared to untreated control animals (OR=0.47; 95% CI=0.39-0.57). While the early withdrawal protocols also showed a significant reduction in the average odds of severe abscesses compared to negative controls (OR=0.59; 95%CI=0.40-0.95), early withdrawal was not found to achieve bioequivalence with continuous supplementation. The 95th percentile of the log odds ratio exceeded the pre-specified non-inferiority margin of 0.2, indicating there were meaningful differences in efficacy.
Significance:
This investigation demonstrates how Bayesian network meta-analysis can enhance large animal research by extracting more information from existing studies to guide new trial designs. By integrating prior evidence, researchers can attain equal or better statistical power with significantly fewer animals, leading to cost savings and ethical benefits. This approach enabled robust conclusions about tylosin withdrawal strategies that would have otherwise required prohibitively large sample sizes using traditional methods.
While continuous tylosin supplementation remains the most effective way to prevent liver abscesses, early withdrawal strategies did not offer the same level of protection. However, removing antimicrobial supplementation during the late feeding period could have the greatest impact on reducing exposures overall. The ethical and societal benefits of these reductions should be weighed against the economic cost of any differences in average effectiveness. Importantly, this study establishes a new paradigm for efficient trial design in veterinary medicine, demonstrating how Bayesian analytical methods can be employed in trial design to leverage prior research to answer contemporary questions with less animal use. This approach could therefore offer significant advantages for cost-effective and ethically responsible research in production animal systems.
