Ch. 3— Patterns of Animal Use • 65 
1982. So, although it appears from table 3-6 that 
the usage increased, this was in fact not so (as can 
be seen from table 3-4). Data for all institutions 
from the regression equations show no change in 
mice and a small increase in the use of rats. How- 
ever, since the same pattern of increase by institu- 
tional group reporting can be seen from table 3-6, 
there has likely been no increase or decrease in 
use of these two species between 1982 and 1983. 
In table 3-6, the number of reporting institutions 
includes those that reported any number for any 
species, whether these included rats or mice or 
not. Few significant changes occurred as a func- 
tion of institution type for the 2 years surveyed. 
No trend in animal use can be identified be- 
tween 1982 and 1983, and the available data 
provide no justification for predicting either 
increases or decreases in future years. It 
would have been possible to examine the 1981 
APHIS data sheets and determine whether, on the 
basis of 3 years’ data, a trend for the mandated 
species existed, but the 1981 data sheets would 
not indicate trends for rats and mice. The other 
methods of estimating laboratory -animal use do 
not match the reliability of the APHIS data, and 
thus do not lend much credence to the numbers 
reported in the past. 
Future Animal Censuses 
The major limitation with this estimate of an- 
nual laboratory -animal usage was the need to de- 
pend on available data sources, with all the limita- 
tions just described. Although the APHIS data 
sheets were of considerable value, they still do not 
substitute for an appropriately designed stratified 
random sampling of all possible users. Only then 
would all possible institutions be represented. The 
APHIS scheme depends on institutions to request 
certification. Some may be operating and not re- 
porting to APHIS. Still, with considerable further 
effort, a post-hoc stratification could be done based 
on the APHIS data. 
Estimates could be improved by two major ap- 
proaches. The first, and least expensive, would in- 
volve the use of all annual APHIS reporting forms— 
following an imperative redesign of the form— as 
well as thoroughly determining which registered 
institutions in each year did and did not report. 
Then appropriate statistical estimation techniques 
could be used on an institution-type and year- 
specific basis to correct for missing data. The sec- 
ond, and more ambitious, approach would be to 
conduct a stratified random sample study of all 
possible users . The stratification would be by type 
of institution, size of institution, and species of ani- 
mals. From such a sample, appropriate statistical 
techniques could be used to project to the entire 
population of user institutions. 
In 1985, the National Research Council’s Insti- 
tute of Laboratory Animal Resources announced 
plans for another in its series of surveys of experi- 
mental animal usage. The 1986 census will include 
mammals and birds, but omit fish, amphibians, and 
reptiles. 
SUMMARY AND CONCLUSIONS 
A rough analysis of the number of laboratory 
animals used is important to provide some con- 
text in which to discuss alternatives to using ani- 
mals, evaluate progress toward the goal of using 
fewer animals, and judge the effect that alterna- 
tives might have. OTA therefore evaluated exist- 
ing data on the number of laboratory animals used 
each year in the United States. 
The data sources considered included various 
reports and surveys published by the National Re- 
search Council’s Institute of Laboratory Animal 
Resources, various market surveys, and the an- 
nual reports submitted to USDA’s Animal and Plant 
Health Inspection Service. For the latter source, 
the individual annual reports furnished by each 
registered facility for 1982 and 1983 were evalu- 
ated. Generally, it was found that great dispari- 
ties existed among the different sources. No sin- 
gle data source presents an accurate count of the 
number of laboratory animals used in the United 
States since not one includes all potential users. 
In addition, it is impossible to compare data among 
sources due to the inadequacy of information on 
survey and data collection methodologies, defini- 
tions, areas of use, reporting requirements, and 
the inability to justify completeness of the data. 
In a comparative analysis of data sources, it was 
found that the most useful data were the APHIS 
