152 • Alternatives to Animal Use in Research, Testing, and Education 
more than one administration per day because 
metabolic rates and excretion rates tend to be 
faster in small animals than in humans (24). 
Statistical Considerations 
To obtain valid results, an experiment must be 
designed so that what is measured provides useful 
and sufficiently accurate information. Statistical 
methods allow a scientist to estimate the minimum 
number of test animals from which conclusions 
can be drawn to estimate the reliability of any con- 
clusions. Statistical analysis can help reduce the 
number of animals needed for a particular test 
procedure. 
To allow for the unexpected (including death, 
illness, or error), the number of animals used al- 
ways exceeds the minimum number needed to de- 
tect expected effects reliably. Determining that 
minimum number of animals is more difficult for 
longer tests, both because the passage of time 
makes the probability of something going wrong 
during the experiment increase, and because cer- 
tain problems are more likely to occur as the ani- 
mals age. 
Another factor affecting the number of animals 
needed is the variability in the sensitivity of indi- 
vidual animals to the substance involved. Thus, 
as few as 6 animals might be used for an eye irrita- 
tion test or 10 per dose level for an acute toxicity 
test. In carcinogenicity or teratogenicity testing, 
many of the animals may be unaffected by the test 
substance, and 100 animals may be needed for each 
dose level. 
Most species experience some cancer and other 
diseases during their life. Any measurement of in- 
cidence as it relates to the dose given must be taken 
against this background incidence, which is gauged 
in an (untreated) control group. Control groups 
may also be important if a test substance is being 
carried in a particular vehicle needed to administer 
the test substance, such as in solution with another 
chemical, that is not itself being tested (vehicle con- 
trol group). The sensitivity of the test animals to 
a substance known to be toxic may also be meas- 
ured for comparison (a positive control group). Be- 
cause there are so many variables that can influ- 
ence a test, toxicologists consider it vital that the 
control and test groups be drawn from the same 
pool of animals and be tested concurrently. 
Any experiment suffers from experimental er- 
ror, of which there are three sources: the natural 
variation due to differences among test animals, 
the variation in experimental conditions, and error 
arising in measurement. Determining the amount 
of error is crucial to drawing reliable conclusions 
from experimental results, but it is also important 
to keep the error as low as possible by controlling 
conditions carefully. Differences among test ani- 
mals are controlled by using genetically similar 
and sufficiently large groups for each condition. 
Even minor environmental factors can influence 
toxic response (15,23). Sources of measurement 
error depend on the measurement technique and 
the equipment. 
Use of Standardized Test Methods 
and Guidelines 
Testing methodologies are standardized to con- 
trol experimental variables, thus allowing results 
to be easily compared. Methodologies may be- 
come standardized through round-robin testing 
in many labs, through publication and imitation, 
and through development by recognized organi- 
zations or agencies. Methodologies or guidelines 
are published by the Food and Drug Administra- 
tion, the Environmental Protection Agency (EPA), 
the Organization for Economic Cooperation and 
Development (OECD), the National Cancer Insti- 
tute, the American Society for Testing and Mate- 
rials, the American National Standards Institute, 
the British Standards Institute, the International 
Agency for Research on Cancer, and others (see 
app. A for information on FDA, EPA, and OECD 
guidelines). 
The most important reason to strive for com- 
patibility among guidelines is to avoid the need to 
repeat identical tests to satisfy particular require- 
ments of various governments and agencies. Com- 
patibility can also avoid nontariff trade barriers. 
Any government that would like to change its test- 
ing requirements to further the cause of animal 
welfare needs to consider the effects of its pol- 
icies on testing in other countries. 
