Could a phase model help to improve translational animal research?

Abstract

Animal models are widely applied in medical research for different purposes. In particular, results from translational experiments may be used for subsequent clinical development. However, transferability of these findings to the human organism is controversial. Among other factors, this may be traced back to a lack of clear differentiation of the evidence (explorative vs. confirmatory) provided by such experimental results. In general, inferential statistics (i.e. p values) should not be interpreted in as confirmatory unless crucial methodological requirements are met. Therefore, we propose a phase model which reflects the well-established process of clinical research, and we discuss its potential to improve decision making in translational research. The model aims to clarify the reliability of results derived from animal models. The phase model proposes subdividing translational, pre-clinical research into pilot, exploration, and confirmation phases. Experiments for which there is no valid estimation of the expected effect size are designated as pilot studies. Based on these data, experiments in subsequent phases may be planned using both appropriate design and statistical methods. Separating the entire process of translational animal research into three phases could contribute to improved transparency of the evidence derived from such experiments.