Sir

Your News Feature 'Standard model' (Nature 454, 682–685; 2008) highlights problems in using mice as models for human neurodegenerative disease. But these pitfalls may be widespread in animal research (see, for example, P. Perel et al. Br. Med. J. 334, 197; 2007).

Animal studies are not always well designed and null results are rarely published. The standards should be raised to be comparable to those already established in clinical research. Preclinical research projects using animals should be prospectively registered and systematically reviewed. Also, they should be properly designed by using randomization, adequate sample sizes and blinding for evaluation of outcome.

Any rigorous scientific research requires investigators to set their basic hypotheses in the context of what is already published, to avoid unnecessary replication and to justify the new study. However, there is often a failure in animal research to apply these standards. This can expose research volunteers and patients to flawed and inadequate research and put biotech investors at risk of substantial long-term financial loss.

Supporting the introduction of these requirements for animal research would improve the body of evidence available to policy-makers and investors with regard to human health. It would also minimize wastage of laboratory animals and improve patient safety.

See also Animal research: too much faith in models clouds judgement.