Covelli must be commended for his conscientiousness regarding the number of animals necessary in a protocol and for setting up an IACUC subcommittee to review the available literature on animal number justification.

The number of animals that is appropriate for a specific protocol depends on complicated variables, including the type of study proposed. Thus, federal policies and regulations on animal number justification are deliberately ambiguous, to allow enough flexibility to cover various study types and models in animal use protocols. For instance, justification for teaching protocols will differ from that of breeding studies. Likewise, statistical justification is appropriate for certain studies but not others. Typically, pilot studies to formulate hypotheses or comparative studies to determine whether two or more values are equivalent will require statistical animal number justification, whereas teaching or breeding protocols will not, but the number of animals used must be justified regardless.

Covelli's subcommittee rightly concluded that the most definitive statement about animal number justification is that offered by the Guide1, which is incorporated into the PHS Policy2, indicating a 'need to justify the number of animals required and to use statistical methods whenever possible'. However, the subcommittee should have cautioned that the number of animals used should be minimized and that, no matter the type of study, a rationale for the number of animals used must be provided. In my opinion, Covelli was not right in claiming that the IACUC can accept 'fairly loose' criteria for justifying the number of animals to be used. I feel that justification of animal numbers used in a study must address alternatives and the 3Rs (replacement, refinement and reduction)3 as well as the type of study. The justification for the number of animals needed should not be based on how many experiments laboratory personnel can perform per week, month or year; neither can the cost of animals used be the determinant for preferring a particular species or model over another. Rather, the species and models that are most appropriate for the study should be used.

Accepting 'fairly loose' criteria for justifying the number of animals used in a study could be interpreted by researchers to mean any statement would be acceptable, without careful thought as to whether the number of animals to be used is justified. Such an interpretation would defeat Covelli's desire for a reasonable assurance that the smallest number of animals required for scientifically valid results were used. Statements such as 'in our experience, this is the minimum number of animals needed for statistical significance' do not provide enough justification. Suppose the animal numbers used 'from experience' were large owing to a lack of procedural refinement on the part of the investigator(s). In accepting this rationale, the IACUC would not be ensuring that the proper justification had been provided.

Return to Protocol Review