Ch. 10— Information Resources and Computer Systems • 237 
Second, the collection of descriptive data on those 
scientists interested in the proposed database and 
on their disciplinary specialties provides a basis 
for matching the scope of the database to the 
breadth of disciplinary interest. Specifically, the 
scope and design of the database depend on the 
range of purposes for collecting the research and 
testing data, the size of the prospective audience, 
and the needs of the users . It is essential to recog- 
nize that the needs of any user audience are dy- 
namic and subject to change. A feasibility study 
should include an analysis of current trends in user 
application as a basis for inclusion of sufficient flex- 
ibility to permit later modification. 
In identifying the potential user community , the 
following considerations are key: 
• Can a model be developed to estimate with 
a high level of accuracy the number of poten- 
tial users? 
• Can a projection of the number of potential 
donors be made from a similar model? 
• To what extent will the user community sup- 
port assessment of operational charges to de- 
fray costs? 
• What will it cost to collect, systematize, store, 
and retrieve the data for a computerized, on- 
line system? 
Third, critical to the acceptance and success of 
a registry of research and testing data is peer re- 
view by experts, at all levels of database develop- 
ment. These levels include: 
• system design; 
• definition of data elements; 
• establishment of standards for data acceptance; 
• compilation and building of data files; and 
• post -hoc evaluation of the system (i.e., feed- 
back resulting from experience gained by ac- 
tual use of the system). 
The peer-review process assures that experi- 
enced researchers have judged the design, stand- 
ards, and data to be used. The process enhances 
quality control, although it imposes the penalty 
of high costs and slow input of data. 
EXPANDING THE LADB CONCEPT: A COMPUTERIZED REGISTRY 
OF RESEARCH AND TESTING DATA 
The concept behind the LADB could be expanded 
in at least two important dimensions. First, the 
scope could be broadened beyond baseline results 
to include experimental results from research and 
testing. How great an increase in size would this 
be? For every measurement obtained from a group 
of control animals, measurements are obtained 
from an estimated one to nine groups of experi- 
mental animals. This makes a registry of control 
and experimental data from 2 to 10 times the size 
of a registry of baseline data alone. 
Second, the coverage could be enlarged beyond 
principally rodents to all vertebrate species. How 
great an increase would this entail? Several hun- 
dred vertebrate species could be involved. The 
number of species would increase by a factor of 
more than 100. Yet the bulk of the results would 
still be derived from rats and mice, since rats and 
mice account for 12 million to 15 million of the 
17 million to 22 million animals used annually in 
the United States (see ch. 3). Increasing the scope 
from rats and mice to all vertebrates would there- 
fore likely enlarge the size of the data registry by 
a factor of 1.5 (17 million to 22 million animals 
divided by 12 million to 15 million rats and mice). 
The creation and maintenance of a computer- 
ized registry of baseline and experimental results 
from all species of vertebrate animals would rep- 
resent an enterprise 3 to 15 times more complex 
than the unsuccessful Laboratory Animal Data 
Bank. 
The factors that led in the 1970s to the assign- 
ment of the LADB project to the NLM remain valid 
today should a similar project be undertaken. NLM 
has related experience in handling substance-ori- 
ented databases (as detailed in table 10-3), such as 
the TDB and BTECS. NLM also operates much larger 
databases, such as TOXLINE and MEDLINE, that 
are bibliographic rather than substance-oriented. 
