28 COLLECTING AND ENTERING DATA
to be made on a variety of topics, such as who will be eligible for the study, what
information is needed, and how the needed data will be obtained and analyzed (see
van Belle et al. [2004]). If the data have already been obtained and are in some type
of computer records, it may be sensible first to transfer them to EXCEL files, since
most statistical programs will accept EXCEL files.
Different types of medical or public health studies tend to use mainly different types
of data. The laboratory experiment, clinical trial, and casekontrol studies tend to use
data that is similar to what could be found in a medical record or a laboratory report.
This is medical-type data that can be interpreted by a professional. This information
can either be used directly from records or special tests or can be reformated to fit
the needs of the study. Here the study population tends to be patients, and the person
collecting the data typically would be a physician. This type of study will simply be
called a medical study.
The second type of data is obtained from the respondents in surveys or prospective
studies. This could include data such as attitudinal data, data on risky or recommended
lifestyle behaviors, or data on access to medical care. The respondents do not have
to be patients, and the studies may be performed by epidemiologists or other public
health professionals. Here these studies are called simply public health studies.
3.1.1 Decide What Data You NeedWhether a study is a medical study or a public health study, data on characteristics
such as age, gender, marital status, plus other characteristics of the respondent are
usually collected. What is collected will obviously depend on what is being studied.
The next step is to decide what specific data to collect for the study at hand. Here
input from all the investigators is essential. It is often useful to review the available
literature on the topic under study to see what information others have obtained and
how they have taken it. To end up with data that could be used in a meta-analysis,
it is important to include measurements comparable to those that have been used in
other studies. This will also make it easier to compare results of the new study with
previous results. For example, in measuring pain, most investigators use a pain scale
that goes from 0 to 10, where 0 is no pain and 10 is the worst possible pain. If this scale
is used, results can be compared with other studies and physicians are comfortable
interpreting the scale. Decisions need to be made on a variety of topics, such as what
information it is critical to obtain, how this needed data will be obtained, and how the
investigators expect to analyze it (van Belle et al. [2004]).
In collecting data it is sometimes useful to collect more detail than one currently
expects to use. For example, collecting the age of the patients rather than just collect-
ing which of three broad age groups they fall in may make sense. If it turns out that
one age group has very few people in it, this could limit the analysis. On the other
hand, collecting large amounts of data that never gets analyzed is a waste of time.
If tests are going to be carried out on the patients using new techniques or equip-
ment, test runs should be made. The equipment used to take the measurements should
yield accurate results, be acceptable to patients, and yield the desired information.