Time alignment of repeated measurements in the analyses of several veterinary clinical trials

Abstract

A summary analysis for a new drug was required after five early phase veterinary clinical trials had been carried out. Such a meta-analysis of clinical trials involving repeated measurements would appear to be straightforward. However, this is not necessarily the case since different problems can arise. The choice of model is important since it may fit the data closely but have unwanted properties. As well, the trials to be regrouped may not have originally been designed to be combined. Hence, this can lead to various additional problems such as deciding on a time origin, analyzing an unbalanced design, and discovering confounded or collinear variables. Most of these problems were encountered in the five veterinary studies. These will be discussed and solutions suggested.