| |
In a parallel
group (also termed "completely randomized") design, each
patient receives a single treatment. In a crossover design,
each patient receives some (incomplete block) or all (complete
block) of the treatments being studied.
Crossover designs have proven
so valuable in the study of pain and other symptoms that
Chapter 6 will discuss
these designs in considerable detail. The brief presentation
below explores the use of these designs in pain studies.
Crossover designs
In diseases with outcomes
that revert to baseline after treatment is stopped, there
are several obvious advantages to crossover designs. Clinical
trials that use subjective outcomes often require large
sample sizes because detection of a drug effect must compete
with so many other causes of variation in outcome: the nature
of the lesion causing the patient's symptoms, his or her
style of describing symptoms, psychological makeup, interaction
with the study personnel, etc. Much of this between-patient
variation can be eliminated by using a crossover design,
in which treatment comparisons are largely or entirely within
the same patient (Louis
et al., 1984; Jones
and Kenward, 1989; Ratkowsky,
1993; Senn, 1993). Because
of this reduction in variance, and because each patient
is used several times, crossover studies often have greater
statistical power than parallel group designs that include
5 or 10 times the number of patients (Louis
et al., 1984). This is an important practical advantage,
particularly when studies are performed in a single center.
This greater power is particularly appealing given recent
findings in pain research that pain mechanisms may vary
from patient to patient. (See
introduction to this chapter.) As methods are developed
to identify these mechanisms in individual patients, crossover
designs will facilitate examination of responses in small
mechanistically-defined subsets.
|