Basic Statistics

APPLICATION OF SAMPLING METHODS IN BIOMEDICAL STUDIES 23

best treatment. Again, the treatments are given in random order. Phase IV trials are

postmarketing surveillance trials that may be done to check for rare side effects (see

Piantadosi [2005]).

Special methods of assignment are sometimes used so that there is an equal number

of patients in each treatment group over time (see Wooding [1994], Friedman and

Furberg [ 19981, and Fleiss [1986] or Fleiss et al. [2003]). One procedure that is often

used is to randomize the assignment of the patients to treatment A or B within a block

of patients. The size of the block can be chosen so that patients who enter the study in

a limited time period are equally likely to get one of the two treatments. Suppose that

a block of eight patients is chosen; then in each block four patients will get treatment

A and four, B. This can be done by entering the random number table, taking the

first four distinct digits between one and eight, and assigning those patients to, say,

treatment A and the remainder to treatment B.

2.4.4 Samples for Prospective Studies

Examination of prospective studies or panel studies shows that several types of sam-
ples tend to occur. Many prospective studies have been done using a chunk sample
of a special cohort of people who have been chosen for their high rate of cooperation
both initially and at follow-up. For example, studies among employees in a stable
industry such as utilities, the armed services, or religious groups such as the Seventh
Day Adventists can be done with minimal loss to follow-up. The employer or group
will keep track of the persons in the study and the investigator can use their records
to find the study subjects.
Some studies have been done using random samples from a general population
within a geographic area. One of the most noted examples is the Framingham Heart
Study, where the sample was a mixture of adults who were randomly sampled and
volunteers who after hearing of the study wished to participate (see Mould [ 19981). It
was started in 1948 to determine risk factors for heart disease. Over 10,000 adults par-
ticipated in the Framingham study. In Los Angeles, a prospective study of depression
was done where the sample was obtained in a random fashion by the Survey Research
Center at UCLA (see Frerichs et al. [1981]). This type of study is more expensive
to do and the loss to follow-up may be higher than when a cohesive and motivated
chunk sample is taken. Nevertheless, this type of sample allows the researcher to
make inferences as to the population being sampled.

2.4.5 Samples for Case/Control Studies

In case/control studies, the investigator starts with the cases after they are diagnosed

or treated. These studies are also called retrospective since the investigator is look-

ing backward in time. This often involves taking a chunk sample at one or more

institutions that have medical records that the investigator can search to find cases

that meet the eligibility criteria for a particular disease. Sometimes the study is per-

formed solely from available records. Otherwise, the investigator must contact the

case, obtain the person’s consent to enter the study, and interview or examine him or