|
What
Does a "Negative" Result Mean in a Symptom Study?
Establishing "Assay Sensitivity"
 The
interpretation of a "negative" result (i.e. no
significant difference in primary outcome between two treatment
groups of interest) differs in studies of symptoms and structural
disease. Because one can objectively and reproducibly characterize
and quantitate structural disease (e.g. tumor size and histology,
patient dead or alive, etc.) most investigators interpret
a finding of no difference between a new treatment and a
comparator, given adequate sample size, to mean that there
really is no difference. In contrast, a variety of influences
on the way analgesic clinical trials are executed and pain
is assessed sometimes make it difficult for even the most
expert investigators to have their patients distinguish
the effects of a potent analgesic and a placebo. Borrowing
the analogy of a chemical assay, analgesic experts will
not accept the validity of a finding of "no difference"
between two treatments unless another internal control confirms
"assay sensitivity," that the study methods can
show a difference over the range of interest if one is there
(Max and Laska, 1991).
The
graphs in the following eight examples represent the outcomes
of single-dose analgesic studies and are intended to illustrate
the situations in which placebo or standard analgesic control
groups are needed, and when they are less essential. Although
some of the questions may seem elementary, we urge you to
go through all the questions for the eight figures, because
they are essential to grasping some unique features of clinical
trials of symptom treatments. This is now widely applied
in determining the validity of single-dose analgesic trials
(Max
and Laska, 1991; FDA,
1992), and appears plausible for use in the study of
other symptoms and with modifications to clinical trials
in which repeated doses of drugs are given over days or
weeks.
Interpreting
Analgesic Studies: Eight Cases Illustrating the Information
Gained from Placebo and Positive Controls
This
and the figures to follow represent the results of a study
of a single dose of analgesic given to patients with acute
postoperative pain. The sum of hourly pain relief scores
for four hours after a dose of analgesic (often called TOTPAR
for "total pain relief") are shown for Drug X,
a putative analgesic, placebo, and morphine (or another
analgesic known to be effective in the condition being studied).
The error bars represent standard errors of the mean, and
a difference of roughly two standard errors would indicate
a statistically significant difference between treatments.
Question
3.1
Do these data demonstrate that Drug X is an effective analgesic
in this condition?
|