Current Concepts | Invited Review
Wechsler et al12 tested the power of the placebo effect in
a study of patients with asthma in which pulmonary function tests (PFTs) provided an objective measure of response
to treatment. Patients were randomized to active albuterol
inhaler, sham inhaler, acupuncture or no treatment with
crossover among groups. All patients eventually received all
three treatments. Only the albuterol inhaler patients showed
improved PFTs, but the albuterol inhaler, sham inhaler and
acupuncture groups all reported subjective improvement, and
all three groups felt significantly better than the no-treatment
The placebo effect in surgery and other spine interventions
Certain patients with spine pain undergo interventional
treatments or surgery. It is generally assumed that the predominance of any beneficial reduction in pain was due to
the specific effects of the intervention or surgery. However,
it might be surprising to learn how much of the reduction in
pain after active intervention could be placebo effect rather
than the specific benefits of the intervention. 6, 13
To quantify the magnitude of the placebo effect and to truly
show efficacy of an intervention, a trial would need three arms:
active intervention, sham intervention and no treatment.
With respect to spine, studies comparing vertebroplasty and
radiofrequency neurotomy to sham exist, but I was not able
to find studies for other types of spine surgeries or procedures
that used sham controls.
Several investigators have reported systematic reviews
and meta-analyses to look at active interventions compared
to sham. 6, 13 Gu et al13 performed a systematic review and
meta-analysis of sham surgery to determine the effect size of
sham surgeries on pain, disability and quality of life measures.
They defined surgery as “…any intervention that altered the
anatomy including procedures with incisions as well as implants left in place for more than one day.” Eight trials included
subjects who had spine surgery. In studies with dichotomous
outcomes, the proportion of subjects with improvement in
pain, disability or quality of life measures after sham surgery
was an almost astonishing 42.8%. Improvement was still present at six months. Complication rates were low. Interestingly,
27.1% had improvement in some objective measures.
Jonas and associates reported on 15 RCTs including seven
that looked at back pain treatment with vertebroplasty or
radiofrequency interventions compared to sham. 6 Their
primary outcome was based on trials that reported results as
continuous variables. They concluded that nonspecific effects
accounted for as much as 78% of the improvements in pain. In
a subgroup analysis of studies that only reported dichotomous
outcomes, there was a significant positive effect compared to
sham. More recently, Jonas et al completed a separate meta-analysis of surgeries versus sham and analyzed a subgroup of
spine surgeries versus sham (Jonas, W. Personal Communication). They found the differences in outcome between sham
surgery and specific surgery was not significant, and that
nonspecific effects accounted for 73% of the pain reduction.
While recognizing that vertebroplasty might not be representative of other spine surgeries, every systematic review that
looked at surgery versus sham for orthopedic pain conditions
has found essentially the same thing—a large nonspecific effect. 7, 8 For example, in a two-year follow-up of sham versus
true meniscus surgery, both groups had about a 37% success
These data on spine interventions should not be interpreted to mean that all spine interventions have a very large
placebo effect, only that some do. Certainly, most spine surgeries and procedures are beneficial because of their specific
effects. However, it is important to remember that most spine
surgeries and other spine interventions have not been compared to sham. Instead they are compared to one another or
to continued rehabilitation, possibly an unfair control group.
The message is that it is important to keep in mind that the
placebo effect can be quite large and can be clinically and
scientifically important both in patient care and research.
Also noteworthy are exceptions such as the studies of disc
biacuplasty and sacroiliac joint neurotomy compared to sham
which demonstrated benefit of the active intervention and
only a small placebo effect.15, 16
Some might argue that it is not ethical to use sham surgery
when trying to determine efficacy of a surgery or procedure.
The ethical arguments have been discussed and most authors
opine there is a role for sham surgery in research. 5-9 Perhaps
we would then have a higher level of evidence on which to
make clinical decisions, and in the future, we might avoid
interventions with little or no efficacy and whose benefits are
significantly related to their placebo effect.
Nocebo effect and other negative effects
Not all placebos have a positive effect. Some placebo pills or
sham interventions can result in patients feeling worse, which
is termed nocebo effect. 4 For example, after an enumeration
of potential side effects of a treatment, patients might be more
susceptible to perceive side effects due to the unconscious
anticipation of these adverse effects. In fact, it has been
estimated that 4% to 26% of subjects randomly assigned to
placebo experience side effects sufficiently severe to cause
them to discontinue the placebo.
Perhaps less well appreciated and rarely mentioned is the
potential negative effect of being randomized to the “less dramatic” arm of an RCT such as physical therapy or TAU rather
than the exciting surgical or invasive intervention arm. This
could potentially lead to a negative expectation of improving
compared to the positive expectations of the active arm, which
could make the active arm appear more efficacious. The same
effect can be observed clinically when the patient is offered
a simple treatment—a pill, exercise or continued TAU rather
than a new treatment that he or she read or heard about. Perhaps this effect is a variant of nocebo, but for a different reason.
It could be called a nexcebo effect for negative expectations.