Efforts to improve the design of animal experiments gathers momentum

Last week saw the publication of updated guidance on funding applications involving the use of animals, by the UK’s Research Councils. The announcement was picked up by the specialist and general press, with Nature and the Observer both covering the publication. Our Chief Executive Vicky Robinson discusses how the NC3Rs has been driving improvements in experimental design, analysis, and reporting.

The new guidance was inspired by work from the NC3Rs that had previously shown that not all animal experiments are appropriately designed, analysed and reported. Our work and that of others has shown that in some cases too few animals are used to give meaningful and robust data and in others too many animals are used. In either case animals are wasted and this has serious scientific, ethical, and resource implications as well as undermining public confidence in animal research and the mantra from some organisations that all animal studies are done to the highest possible standards. 

In some instances, the use of too few animals has been blamed on the ‘reduction’ agenda of the 3Rs. This is a poor excuse and shows a complete lack of understanding about Russell and Burch’s principles since reduction refers to using the minimum number of animals consistent with the scientific objectives. The reality is that poor experimental design has blotted the copybook of the scientific community for some time. The new guidance from the UK’s funding heavy weights sends a strong signal that this must change.

We have been championing improvements to the design, analysis and reporting of animal research since our ARRIVE guidelines were published in 2010. There is still much more to do and will shortly be launching our online Experimental Design Assistant which will provide researchers with tailored guidance at the time experiments are being planned and designed in a way that a generic text book on experimental design could never achieve.

The EDA will include information on how to devise a hypothesis, how to identify and address sources of variability, and the reasoning behind the choice of statistical methods and analysis. It will provide tools for randomisation and blinding (for example, generation of a sequence for the allocation of animals which is sent directly to a third party) as well as for power calculations for determining group sizes - all information the funders are seeking in their applications.

If animal studies are essential to expanding the knowledge base and improving human health then it should go without saying that the studies should be designed in a way that provides data and outcomes that can be trusted. There can be no excuse for getting it wrong.

Subscribe to our newsletter