Animal Welfare and the Statistical Consultant
Engeman, Richard M., Shumake, Stephen A., The American Statistician
There has been an increasing focus worldwide on how animals are used in research experiments. The public and the scientific community generally concur that animals should be used efficiently and treated humanely when they are required in experimentation. Emphasis, in terms of attitudes and policies, has been placed on reducing the number of animals used in experiments, substituting nonanimal models (including mathematical models) for animal experiments where possible, and adjusting experimental methods to decrease animal suffering (for example, in vitro tissue irritation tests to replace eye irritant studies on rabbits). These original recommendations by Russell and Burch (1959) have been accepted as the standard by many oversight bodies. Their philosophy regarding animal research is summarized as the three R's: Refinement (of experimental technique), Reduction (of number of animals used), Replacement (of animals with substitutes). These general principles have been incorporated into animal welfare regulations and progress is being made in many areas of testing and experimentation towards the well-being of research animals (see, for example, Rao and Huff 1990; Rowan 1990).
Passage of the Animal Welfare Act and the amendments revising it (U.S. Congress 1989) have led to substantial impacts on the uses of animals in experiments in the United States to ensure both wise use and minimum stress to research animals. Similar regulations have been instituted in other countries. These laws have a direct effect on the role of consulting statisticians worldwide. The general areas in which the consulting statistician has been most affected are discussed in this article; both U.S. and Australian regulations are used as examples of the directions that the animal welfare laws have taken. Similar regulations have been instituted in many European countries in recent years (O'Donoghue 1992).
1. ANIMAL CARE COMMITTEES
The implementation of institutional animal care committees is the primary method for assuring that the well-being of research animals has been considered before any study is initiated. Both the Public Health Service (Office for Protection from Research Risks 1986) and the U.S. Department of Agriculture (1989) require Institutional Animal Care and Use Committees (IACUC) to be appointed by the directors of research institutions. Similarly, Animal Experimentation and Ethics Committees (AEEC) in Australia have been formed and they are likewise charged with ensuring the humane use and care for animals. All proposed experiments involving animals must be reviewed and approved by IACUC's or AEEC's according to these laws. Committees are usually comprised of members with expertise in veterinary science or laboratory animal medicine, practicing scientists from several disciplines, and persons representing community concerns (National Health and Medical Research Council, Commonwealth Scientific and Industrial Research Organisation, and Australian Agricultural Council 1990; Orlans, Simmonds, and Dodds 1987). They are responsible for ensuring the appropriate use of animals in experiments, which includes the numbers of animals used, how they are used, and whether they will be used to produce valid results that meet the experimental objectives. For this reason, a statistician would be useful as a permanent member of an institution's committee, or at least all research protocols could be separately reviewed by statisticians to provide input into the research design. Protocol review is already a primary responsibility of many staff and consulting statisticians, and this added responsibility would be strengthened by interaction with these committees.
The Australian Medical Research Committee of the National Health and Medical Research Council (1989) provides a guide to AEEC's for minimizing the number of animals used in each research project. One section in this guide, entitled "Ensuring the Best Statistical and Predictive Techniques Are Used," considers the degree of statistical knowledge and experience of the investigator and whether or not the investigator has sought appropriate statistical advice. …