Designing animal research
A good research design is crucial for the quality of animal research and the conclusions that can be drawn from it. To protect humans and animals, we must be sure of the reliability and repeatability of animal studies. The Utrecht Animal Welfare Body therefore stimulates quality improvement that guarantees scientific integrity. In addition, it prevents animals from being used for research that does not sufficiently contribute to the increase in knowledge and the health and welfare of humans and animals.
In this article
There is an ongoing replication crisis within the scientific community. This is particularly the case with animal experimental research. This damages confidence in the usefulness of animal testing and can be dangerous for the target group (usually humans, but also animals) for which animals are a model.
Replication
Many animal studies prove to be difficult to replicate because researchers provide insufficient information about the implementation. Partly for this reason, certain studies are not included in systematic reviews. The value of the study can therefore decrease, even if the study has a high initial impact. This can partly be prevented by complying with the ARRIVE guideline. (An important guideline for transparent and complete publication of animal experimental studies.)
Publication bias
There is also a publication bias within scientific research. This means that the only publishing positive or significant results of experiments gives a distorted view of reality. To prevent this from happening in Utrecht, we encourage animal studies to be registered in advance. Each study can then be included in systematic analyzes.
Statistical analysis plan
Animal experiments can easily be misinterpreted because, in the context of reduction, we work with as few laboratory animals as possible. A well thought-out experimental design is therefore of great importance. The Utrecht Animal Welfare Body checks the number of laboratory animals used and the statistical analysis plan extra closely. If needed, we engage a statistician to help assess the analysis plan. It is important to find out how possible influences on your data can be limited and what the chances are of exclusion.
Some risks can be mitigated in advance, for example by correctly defining the experimental unit, randomizing animals into groups, carefully maintaining rooms and cages, and by blinding treatments, observations and analyses. Knowledge of the animal model is indispensable.
Important resources
- The aforementioned points for attention are included in the PREPARE guideline, which details the correct planning and coordination of animal experiments. On the basis of this guideline you can systematically go through your animal experiment to find out whether parts are missing.
- View our Explanation of experimental design and statistics.
- We recommend using the G * Power application to calculate group size in accordance with your analysis plan .
- For Randomization of groups where certain co-variation such as weight, age, etc. have to be balanced, you can use the RandoMice tool.
- A course for PhD students is available to develop competence in setting up an animal experiment. This is the My Animal Research, Experimental Design course.
- You can convert your test setup with analysis plan into a block diagram, so that you can be made aware of the risks of your design. The Experimental Design Assistant can help you with this. The program with which you create the chart will alert you to errors. This program also includes a built-in randomization and blinding tool.
- My Animal Research: Experimental Design is an interactive, blended learning track, with the aim to complete a work protocol involving experimental animals. The learning track is composed of online training modules, face-to-face sessions, group workshops under supervision of an expert, a practice run of the assessment and a final assessment.