Comparative evaluation of ectoparasiticide efficacy: non-inferiority and superiority assessment

We present a statistical framework for the comparative evaluation of ectoparasiticide efficacy in controlled animal studies. Such a comparative evaluation can have one of two objectives: an assessment of non-inferiority of a test parasiticide to a reference parasiticide, or an assessment of superiority of one treatment over another. We show that the observed efficacy of an ectoparasiticide can be viewed as a point estimate of its conditional "kill" probability. Thus concepts used in the comparative evaluation of human antibiotics, which involve the comparison of cure probabilities, can be applied to the situation of comparing parasiticide efficacy. In particular, we define non-inferiority of a test parasiticide to a reference parasiticide in terms of the "test - reference" difference of their efficacies. We outline the hypothesis testing framework and associated statistical decision rules for declaring either non-inferiority or superiority. Both non-inferiority and superiority can be evaluated statistically using confidence intervals. SAS and R code for data analysis is presented, and the methodology is applied to a data set from a controlled animal study.