Ethics of Research involving animals
Reduction
Definition and scope
12.4 Russell and Burch initially defined Reduction as ‘reduction in the numbers of animals used to obtain information of a given amount and precision’. More recently, this definition has been developed to state: ‘the use of fewer animals in each experiment without compromising scientific output and the quality of biomedical research and testing, and without compromising animal welfare’.3 The proviso that Reduction should not compromise animal welfare is necessary because reduction in the number of animals used can sometimes be achieved by performing more procedures on each animal. This could cause an undesirable increase in the suffering of individual animals. In addition to improved research strategy, as outlined above, Russell and Burch suggested two additional ways in which animal use could be reduced: better control of variation and better statistical analysis.
- Reducing variation: choice of appropriate animal species and strains
Many factors need to be considered in choosing the most appropriate animal model for a particular experiment. Thus, for example, the species, strain, sex and age of the animals are all important criteria. The outcome of a project may depend critically on the strain(s) used. A wide range of inbred strains, mutants, outbred stocks and transgenic strains of mice and rats are available. The use of genetically more uniform or inbred stocks, if appropriate to the particular experiment, may reduce variation and therefore allow the use of fewer animals.
- Statistics and experimental design
A lack of understanding of the basic principles of statistical methods can lead toinappropriate analysis of experimental results and to the conduct of experiments which yield results that are not even amenable to proper statistical analysis. A survey of 78 experiments, described in papers published in two leading toxicology journals between 1989 and 1990, showed that over 60 percent had obvious statistical errors. About one third of the experiments involved far more animals than necessary to achieve the stated aims of the research.4 Where statistical analysis is crucial to the outcome of a research project, it is vital that careful consideration is given to the design of experiments to take account of the degree of variation to be expected, the required statistical power, and the method of statistical analysis to be used. Ways of improving experimental design by controlling variability and allowing the use of more-sophisticated statistical methods have been suggested.5 One way of using these methods in practice would be to improve training of scientists; a more practical and reliable option may be to ensure that scientists have the opportunity to consult at an early stage with a statistical expert.
12.5 The two approaches above are of special relevance for ensuring that numbers of animals intended to be used in a specific research project are reduced as far as possible. But Reduction also has another dimension in the sense that it is desirable to reduce the total number of experiments which are undertaken. In this context, data sharing is an important means of avoiding duplication of testing in toxicology as well as pharmaceutical and academic research. In the case of toxicology testing and pharmaceutical research the results of tests are often commercial property, and the need for confidentiality may sometimes lead to duplication of testing. In basic research, duplication6 may occur when researchers are unaware that a particular experiment or test has already been carried out by other researchers. There have been claims and counterclaims about the extent to which studies are duplicated.7 Nevertheless, ensuring that results from research are shared as much as possible is a useful way of reducing the total number of animals involved in research. The principal way in which data are currently shared is by publication of research in peer-reviewed journals. However, not all research actually undertaken is published. Some therefore argue that it would be desirable to ensure greater availability of reports of ‘negative’, or unsuccessful, research results.8 But there are problems in publishing research findings that are not peer reviewed. The peerreview process helps to ensure that only findings from properly conducted research are published, and publication of poorly conducted research may lead to confusion.
12.6 In the UK, in 2002 the inter-Departmental Group on the 3Rs9 was formed, as a successor to the Inter-Departmental Data Sharing Group, which produced and published in 2000 the inter-Departmental Data Sharing Condordat. The Concordat is a voluntary scheme which seeks to ‘promote opportunities for encouraging agencies, industry and other stakeholders to endorse the principle of data sharing and to extend its scope by looking to overcome the practical, legal, commercial and cultural barriers to its effective implementation.’10 Among other things, the Concordat encourages minimisation of data requirements for tests as far as possible, and the reviewing of procedural and legal barriers to data sharing. Under the Concordat, ‘UK regulatory authorities, as lead agencies, [will] press for agreement on behalf of the UK Government for fullest provisions and procedures which enable data sharing when negotiating, updating and transposing relevant European Directives and when taking part in other international harmonisation processes.’ We return to issues raised by the possible duplication of research in different areas in Chapter 15, where we reconsider the national and international context of research (paragraphs 15.68–15.70 and 15.83). We also explore ways in which the avoidance of duplication can be ensured especially in relation to research involving GM animals (paragraphs 15.71-15.75).
Footnotes4 See also Festing MFW (1996). Are animal experiments in toxicological research the "right" size? in Statistics in Toxicology
Morgan BJT (Editor) (Oxford: Clarendon Press), pp3-11; Festing MFW and Lovell DP (1996) Reducing the use of laboratory animals in toxicological research and testing by better experimental design J R Stat Soc 58: 127-140; see also: Editorial:
Statistically significant (2005) Nat Med 11: 1.
5 See Festing MF (1990) Use of genetically heterogeneous rats and mice in toxicological research: a personal perspective Toxicol
Appl Pharmacol 102:197-204; Festing MF (2001) Guidelines for the design and statistical analysis of experiments in papers submitted to ATLA Altern Lab Anim 29:427-46; Festing MF (2002) The design and statistical analysis of animal experiments ILAR J 43: 191-3; Festing MF and Altman DG (2002) Guidelines for the design and statistical analysis of experiments using laboratory animals ILAR J 43 244-58; Shaw R, Festing MF, Peers I and Furlong L (2002) The design and statistical analysis of animal experiments ILAR J 43:191-3; Howard BR (2002) Control of variability ILAR J 43: 194-201.
6 It is important to distinguish between duplication and replication of experiments, see paragraph 15.16.
7 See, for example BUAV (2001) BUAV Submission to the House of Lords Select Committee on Animals in Scientific Procedures,
available at: http://www.buav.org/pdf/BUAV_HOL_Evidence.pdf. Accessed on: 9 May 2005; see also Home Office (2005) Report
by the Animal Procedures Committee (APC) Review of the cost benefit assessment in the use of animals in research:
Government Response by Caroline Flint MP, Parliamentary Under-Secretary of State for the Home Department, available at:
http://www.homeoffice.gov.uk/docs4/jw280305flint_banner_report_by_the_animal_procedures_committee.pdf. Accessed on:
9 May 2005.
8 See paragraphs 35-37 of Animal Procedures Committee (2001) Report on openness, available at:
http://www.apc.gov.uk/reference/openness.pdf. Accessed on: 9 May 2005.