News Release 

New experiment design improves reproducibility

Ecologists from the Friedrich Schiller University Jena, Germany, and an international team propose measures to increase the reproducibility of biomedical experiments

Friedrich-Schiller-Universitaet Jena

IMAGE

IMAGE: Prof. Dr Holger Schielzeth from the University of Jena is part of the team that recommends the targeted inclusion of variations in the design of animal experiments in order to... view more 

Credit: (Photo: Jan-Peter Kasper/FSU)

Jena, Germany (02.06.2020) For some scientific disciplines, such as medical or drug research, experiments with live animals are still indispensable. Scientists are aware of their responsibilities in this sensitive area and strive to keep the number of experiments as low as possible. Extensive standardisation processes are supposed to increase the efficiency of the experiments, thus reducing the number needed. However, biological complexity, and in particular a dependence on the context of the individual experiments, often make it difficult to reproduce and generalise the results. In the current research journal "Nature Reviews Neuroscience", an international team led by the University of Bern recommends ways of reducing the number of experiments.

Results often depend on context

"The reproducibility of results is a crucial element of science. Results are reproducible if research results obtained from an initial study can be confirmed in independent replicate studies," explains ecologist Prof. Holger Schielzeth from the University of Jena in Germany, one the study's co-authors. "A fundamental problem of biological research is that the results are often very dependent on context. We therefore propose integrating one of these influencing factors - namely biological variability - into the design of the experiment in order to produce more generally valid results."

Standardisation is limiting

Researchers currently standardise the conditions and characteristics of laboratory animals in experiments, such as for the administration of a potential drug, according to strict criteria. In doing this, they want to eliminate all influencing factors that have nothing to do with the immediate objective of the experiment and thus increase the reproducibility of the results. This standardised approach, however, limits the range of conditions to which the results obtained can be generalised. This means that more studies are necessary to confirm the results.

"We therefore recommend the targeted inclusion of contextual variation into the design of experiments, so as to increase the range to which the results can be reliably transferred," says Schielzeth. "This increases the potential for reproducibility and thus reduces the total number of experiments." A "systematic heterogenisation" of animal characteristics and environmental factors could be achieved in a modified version of the randomised block design. This involves pairing up treatments and experimental controls in small blocks, each block being tested in slightly different contexts.

Fewer follow-up studies needed

This arrangement enables researchers to find out whether specific results can be generalised or are to be attributed to influencing factors specific to the experiment. Scientists would be able to address biological variations within a study and, for example, take different sexes, age categories or housing conditions of the animals into consideration. This would give them more reliable findings from a single experiment. Further research on this new method should lead to better guidelines for future experiments.

"We are aware that this experiment design can lead to an increase in the number of animals used for experiments during an initial study," says Schielzeth. "However, much fewer follow-up studies are needed to verify the result, which leads to a significant reduction in the number of animals overall." The team therefore calls on research institutions and regulatory authorities to introduce systematic heterogenisation as a standard model for experiments.

###

Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.