The most valid
way to determine the efficacy of a treatment for fatigue is
to randomly assign patients to receive either the treatment
or some alternate intervention for comparison. Random assignment
minimizes the risk of systematic bias by distributing variables
that may influence outcome between the study groups.
of treatments is a prerequisite for a controlled trial. It
does not, however, eliminate the possibility that the outcomes
observed are related to placebo effects. To establish the
efficacy of an intervention distinct from a nonspecific placebo
effect, the intervention must be compared directly with placebo.
If feasible, a study design that incorporates randomization,
double-blinding of treatments, and a placebo control is most
likely to provide evidence for or against efficacy.
Of course, a placebo
control is not possible with most non-pharmacological interventions,
such as exercise, cognitive therapies, or education about
sleep hygiene. In these studies, patients are usually randomly
assigned to receive the study intervention or a control selected
to reduce some sources of bias. For example, contact with
the investigator is a potential source of bias (i.e. positive
outcomes may not be related to the intervention per se, but
rather to nonspecific positive effects elicited through contact
with the investigator) and a useful control group may include
some degree of investigator contact without access to the
study intervention. In more sophisticated trials, more than
one control group of this type can be created in an effort
to further clarify the intervention effect. For example, a
study of a cognitive therapy (such as hypnosis) may incorporate
a control group that receives some educational materials related
to the problem of fatigue and another control group that is
allowed to make an appointment with the investigator but will
receive no intervention (so-called waiting list control).