Abstract

Oftentimes in animal health studies, a treatment group is randomly assigned to a pen of animals, and the pen of animals as a whole is treated (fed the same medicated feed or water) together. In this scenario, the pen of animals is the experimental unit and the individual animal may be an observational unit. In addition to having the pen as the experimental unit, if multiple sites are used and site is treated as a random factor, this adds complexity to the study. To properly design the study, it is necessary to determine the number of animals in a pen, the number of pens per treatment group, and the number of sites in order to detect treatment differences with desired power. The method for sample size determination depends on the statistical distribution of the primary endpoint in the study, the design of the study, and the proposed statistical analysis method. Focusing on the case where individual animal is not the experimental unit, this paper demonstrates methods and discusses issues for sample size determination in animal health studies with continuous or binary primary endpoints, and gives a simple approach for determining an appropriate number of sites in a multi-site model analyzed with a linear mixed model.

Keywords

sample size, power, experimental unit, experimental design, randomized complete block design, binomial distribution, normal distribution

Creative Commons License

Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.

Share

COinS
 
Apr 27th, 12:00 PM

SAMPLE SIZE DETERMINATION IN ANIMAL HEALTH STUDIES

Oftentimes in animal health studies, a treatment group is randomly assigned to a pen of animals, and the pen of animals as a whole is treated (fed the same medicated feed or water) together. In this scenario, the pen of animals is the experimental unit and the individual animal may be an observational unit. In addition to having the pen as the experimental unit, if multiple sites are used and site is treated as a random factor, this adds complexity to the study. To properly design the study, it is necessary to determine the number of animals in a pen, the number of pens per treatment group, and the number of sites in order to detect treatment differences with desired power. The method for sample size determination depends on the statistical distribution of the primary endpoint in the study, the design of the study, and the proposed statistical analysis method. Focusing on the case where individual animal is not the experimental unit, this paper demonstrates methods and discusses issues for sample size determination in animal health studies with continuous or binary primary endpoints, and gives a simple approach for determining an appropriate number of sites in a multi-site model analyzed with a linear mixed model.