The placebo effect
Placebo is the preferred control in the double-blind
Randomized Controlled Clinical Trial (RCCT).
Although placebo is not supposed to have any
relevant biological activity, it is well known that
it often produces remarkable therapeutic
responses. This phenomenon occurs across the
therapeutic board. It seems that mere knowledge
that the subject is being treated for its condition
often produces a measurable favorable response
(Bok, 1974; Gribbin, 1981). A high placebo
response will tend to mask the response of the
experimental drug. Since placebo is rarely used
outside the clinical research setting, some people
argue that the comparison with placebo tends to
show lower response rates for the drug than would
later be observed in general use. Thus, goes the
argument, the placebo-controlled trial puts the test
drug at a disadvantage. The counter argument is
that what one sees in the clinic is perhaps the
combination of the placebo effect plus the drug’s
biological effect, and therefore, establishing the
residual effect of the drug over its inherent placebo
effect should be the true objective of the trial.
Whatever the case might be, the placebo effect
invariably results in decrease in the signal-to-
noise ratio. Therefore, measures are often taken
to select subjects whose placebo response is low
or nil. One way of accomplishing this is by treating
prospective subjects with placebo for some time
prior to randomization. Patients whose response
during this screening phase is high or very variable
are then disqualified from participating in the trial.
In summary, the selection of subjects to be
enrolled in the trial using a list of entrance criteria
is an important tool that helps to sharpen the signal-
to-noise ratio, thus making the study more power-
ful. It also helps in understanding the extent to
which the study conclusions can be generalized
to a broader population of patients than those stu-
died under the clinical trial’s artificial conditions.
25.8 The statistical model
The statistical model is the mathematical frame-
work in which the statistician operates. It provides
the statistician with the tools to quantify the infor-
mation obtained during the trial and defines rela-
tionships among the various measurements. It
provides a framework for evaluating the properties
of the statistical methods used to analyze the data
and answer the questions the study is designed to
address.What is a statistical model?
A statistical model consists of a set of assumptions
about the nature of the data to be collected in the
trial and about the interrelationships among var-
ious variables. These assumptions must be specific
enough that they could be expressed by a set of
mathematical expressions and equations.
For example: In a placebo-controlled clinical
trial for testing a new analgesic for treatment of
migraine headaches, the key efficacy variable is
the number of subjects whose headache is
eliminated within 1 h of treatment. A statistical
model appropriate for this situation is as
follows:
Letpdenote the probability that a subject treated
with a drug will have their headache disappear 1 h
after treatment, following an episode of migraine
headache. If the responses of different subjects are
independent of each other, this probability can be
expressed asProb:ðno:of responses¼kÞ
¼cpkð 1 pÞNkð 0 kNÞwhereNis the number of subjects treated andcis a
constant representing the number of possible com-
binations ofkelements out ofN. This model is
known as the Binomial Model.
The trialobjective is to determine if thenew drug
is more efficacious than placebo. Within the con-
text of this model, one could declare the drug as
‘more efficacious’ ifpd>pp, wherepdis the prob-
ability of response for a subject treated with the
new drug andppthe probability of response of a
placebo-treated subject.
The data collected during the trial will pro-
vide information aboutpdandpp, enabling the25.8 THE STATISTICAL MODEL 325