Establishing the efficacy of an anti-coccidial drug in poultry begins with conducting multiple battery cage studies, where the target animals are challenged with single and mixed Eimeria species inoculum under controlled laboratory conditions. One of the primary outcomes in a battery cage study is the intestinal lesion score defined on a discrete ordinal scale of 0 to 4. So far, the statistical analysis of lesion scores has routinely employed the linear mixed model (LMM). This present work proposes to apply the generalized linear mixed model (GLMM) with the cumulative logit link to statistically analyze coccidial lesion scores collected from battery cage studies. Upon applying this new approach on 9 datasets generated by challenging battery-cage-housed broilers with various mixtures of Eimeria species, it is observed that the GLMM fitted adequately to the data, produced variance component estimates that agreed with the experimental setup, and, at the 0.05 significance level, generated statistical results in complete concordance with the LMM approach. Advantages of the proposed GLMM over the LMM are discussed from several standpoints. Parallel to the regulatory requirement of a ≥1-unit reduction in the mean lesion score for clinical relevant efficacy under the LMM, the clinical relevancy criterion under the GLMM could be set as a ≥10-fold increase in the odds of having low lesion scores. That is, the effect of an anti-coccidial drug product would be deemed clinically relevant in battery-cage studies when the odds of having low lesion scores with the medication is 10 times or more than the odds without the medication.
Keywords: Battery cage study; Coccidiosis; Generalized linear mixed model; Lesion score.
Copyright © 2018 Elsevier B.V. All rights reserved.