1. The ethics of animal experimentation

This section introduces ethical aspects of animal experimentation including the “3Rs”.

Animals are unable to give informed consent, and in any case thRussel and B02e objective of animal experimentation is to improve the health of humans and domestic animals, not rats and mice. So the ethical framework is different from clinical research In most countries they have legal protection such as that provided by European Directive 2010/63/EU (click here for a pdf) as well as local ethical committees.

European legislation protects all vertebrates and cephalopods from two thirds of the way through gestation or as soon as they can live independently. Guidance on UK legislation is given here.

In 1959 two British scientists, Bill Russell and Rex Burch (above) wrote “The Principles of Humane Experimental Technique” (see here) in which they introduced the “3Rs”. These should be used to assess every experiment.


Wherever possible live animals should be replaced by non-sentient or less sentient alternatives such as cell cultures, invertebrates or mathematical models.

There are many examples where this has been successful.

  • Rabbits and mice were used to assay batches of insulin, but now this can be done chemically. Biological assay used to use large numbers of animals. Many fewer are used today
  • Monoclonal antibodies  were grown as ascites tumours in mice, but this is now done in-vitro.
  • Even in toxicity testing there are a number of tests, such as the Ames test for mutagenesis which, at least partially replace them .


Pain, distress or lasting harm should be minimised.

  • Anaesthesia and analgesia should be used in surgery.
  • Humane end-points should be used when death is the expected outcome
  • Tumours should not be allowed to grow to an excessive size.
  • The animals should be protected from disease
  • An enriched environment  with space  for natural behaviour should be provided
  • Food and water should only be withdrawn for strictly limited periods.
  • Social animals such as mice and rats should be housed with other animals (a legal requirement under Directive 2010/63/EU)

Reduction (the main subject of this web site)

Use the minimum number of animals consistent with achieving the objectives of the study  This involves:

  • A research strategy with clearly defined objectives that can be achieved with the available resources
  • Choice of an animal model  capable of answering the question being investigated
  • Experiments which use neither too many animals so that some are wasted nor too few so that important effects are missed
  • The correct statistical analysis of the results, including summary statistics such as means and standard deviations as well as indicators of uncertainly such as significance levels and confidence intervals.

Harm-benefit analysis

Some sort of harm/benefit analysis should also be used to ensure that trivial work is not done. Of course, the harm is to the animals and the benefit is to humans so this has to be somewhat subjective.

In most cases special justification is required for work involving cats, dogs, horses and non-human primates.

Laboratory animal welfare organisations

There are several organisations concerned specifically with the welfare of laboratory animals. These include:

Center for Alternatives to Animal Testing

FRAME (Fund for the Replacement of Animal in Medical Experiments):

The National Centre for Replacement Refinement and Reduction

UFAW (The Universities Federation for Animal Welfare)

The RSPCA (Royal Society for the Protection of Animals)

Organisations promoting research methodology and evidence-based medicine

SYRCLE: A centre focussing on systematic reviews of publications involving laboratory animals

SABRE Research UK. (Promoting the use of systematic reviews for better health care)

MedicRes (Promoting evidence based medicine, medical decision making and provioding scientific support to medical research)

CAMARADES. A supporting framework for groups involved in the systematic review and meta-analysis of data from animal studies in experimental stroke.

Laboratory animals as models

Laboratory animals are used as models of humans or other species. In about the year 2000 a number of anti-vivisectionists suggested that the use of animals as models was scientifically invalid because animals are different from humans. Were they right? Little has been written about the theory of models. This pdf looks at the way in which models have been used historically, and shows that they have been of critical importance in the develop of modern medicine. Click arrow.

The “Understanding Animal Research web site explains in detail why animal research is necessary, giving modern examples.


There are problems as shown in these Surveys of statistical quality of published papers

1. This survey of 133 papers (McCance, 1995 Aust. Vet. Journal. 72:322) commissioned by the editors  of the Australian Veterinary Journal found that In the opinion of the statistician:                                                     

  • 61%  would have required statistical revision had they been seen before publication 
  • 5%  had such serious errors that the conclusions were not supported by the data      
  • 30%  had deficiencies in design of the studies including failure to randomise, inappropriate group size, heterogeneity of subjects and  possible bias
  • 45%  had deficiencies in the statistical analysis including the use of sub-optimal methods and errors in calculation
  • 33%  had deficiencies in presentation of the results including unexplained omission of data and inappropriate statistical methods

2. If humans or animals lose a large proportion of their blood they go into a state of shock. These studies use animal models to find ways of reducing mortality. But a meta-analysis of 44 such studies (Roberts et al 2002, BMJ 324:474) found that:

  • Only 2 said how animals had been allocated (i.e. whether they had been randomised)
  • None had sufficient power to detect reliably a halving in risk of death, a clinically relevant outcome
  • There was substantial scope for bias in the experimental designs
  • There was substantial heterogeneity in results, due to method of inducing the bleeding
  • As a result the odds ratios were impossible to interpret
  • The authors queried whether these animal experiments made any contribution to human medicine

3. In this study (Perel et al 2007, BMJ 334:197-200) the authors studied six interventions where the results were known in humans and then did a meta-analysis of all the animal studies they could find to see whether they predicted the human outcome. The results were:


In only three cases did the animal studies predict the human outcome. There was clear evidence of publication bias; most academic investigators only publish positive studies so a meta-analysis tends to over estimate the benefits. The authors noted that many papers seemed to be poorly designed.

A survey of a random sample of 271 papers involving live mice, rats or on-human primates  (Kilkenny C, Parsons N, Kadyszewski E et al. Survey of the quality of experimental design, statistical analysis and reporting of research using animals. PLoS One 2009;4:e7824.) found that of the papers studied:

  • 87% did not report random allocation of subjects to treatments
  • 86% did not report “blinding” where it seemed to be appropriate
  • 100% failed to justify the sample sizes used
  • 5%   did not clearly state the purpose of the study
  • 6%   did not indicate how many separate experiments were done
  • 13% did not identify the experimental unit
  • 26% failed to state the sex of the animals
  • 24% reported neither age not weight of animals
  • 4%   did not mention the number of animals used
  • 35% which reported numbers used, these differed in the materials and methods and the results sections
            •   (Click arrow for a pdf)

Over-all conclusion
Most papers reported their results inadequately. None justified the numbers they used, and in many cases the design of the experiments and/or the statistical analysis were inadequate or even incorrect.

More examples (including some quite recent ones) of papers reporting the need for better experimental design are given in the Notes and Comments page.


Test yourself