Page 1
C O M M E N T A R Y
ThePlaceboEffect:“RelativelyLarge”and“Robust”
EnoughtoSurviveAnotherAssault
?
Bruce E. Wampold and Zac E. Imel
University of Wisconsin–Madison
?
Takuya Minami
University of Utah
The evidence related to the placebo effect is discussed, and it is empha-
sized that the descriptors “relatively large” and “robust” are appropriate in
the context in which they were used. Basic science and clinical trials,
when interpreted properly, have revealed that the placebo effect is indeed
a real phenomenon. J. Hunsley and R. Westmacott (this issue) as well as
A. Hróbjartsson and P. C. Gøtzsche (this issue) are concerned that B. E.
Wampold,T.Minami,S.C.Tierney,T.W.Baskin,andK.S.Bhati(2005)over-
statedtheclinicaleffectsofplacebowhenitwasneverB.E.Wampoldetal.’s
(2005)intentiontomakeinferencesaboutclinicalutility;however,itisshown
that the placebo effect exceeds many accepted medical interventions.
© 2007 Wiley Periodicals, Inc. J Clin Psychol 63: 401–403, 2007.
Keywords: placebo effect; psychotherapy; effect size
The primary purpose of Wampold, Minami, Tierney, Baskin, and Bhati’s (2005) article
was to examine clinical trials involving placebos to further the understanding of psycho-
therapy. With regard to the trials reviewed by Hróbjartsson and Gøtzsche (2001), the
reanalysis revealed that (a) placebo effects are found where they theoretically should be
(and importantly, not found where they should not be), (b) differences in disorder ame-
nability to placebo are related to the size of placebo effects, and (c) study design is related
to the degree to which treatment effects were larger than placebo effects. Unlike phar-
macology, it is not possible in psychotherapy research to neatly separate mechanisms
(i.e., chemical vs. psychological) and compare relative benefits (Wampold, 2001). We
Correspondence concerning this article should be addressed to: Bruce E. Wampold, 321 Education Building,
1000 Bascom Mall, University of Wisconsin, Madison, WI 53705; e-mail: wampold@education.wisc.edu
JOURNAL OF CLINICAL PSYCHOLOGY, Vol. 63(4), 401–403 (2007)
Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/jclp.20350
© 2007 Wiley Periodicals, Inc.
Page 2
maintain that the aforementioned findings, originally reported in Wampold et al. (2005),
provide important information in this regard, demonstrating an interesting similarity
between placebo mechanisms in medicine and psychotherapy, when the evidence is parsed
appropriately. Unfortunately, Hróbjartsson and Gøtzsche (this issue) and now Hunsley
and Westmacott (this issue) ignored this discussion and avoided any discussion of how
and when placebos are expected to work.
Hróbjartsson and Gøtzsche (this issue) and Hunsley and Westmacott (this issue)
remained focused on the adjectives that Wampold et al. (2005) attributed to the size of the
placebo effect in a set of studies which were, for the most part, not designed to test
hypotheses related to this effect. In particular, they focused on the words “relatively
large” and “robust,” which, when taken out of context, might offend some when applied
to effect sizes of the magnitude detected by Hróbjartsson and Gøtzsche (2001, 2004) and
by Wampold et al. (2005).
“Relatively large,” the words that Hunsley and Westmacott (this issue) found dis-
agreeable and which were used only once in reference to the results for continuous vari-
ables, occurred in the context of placebo effects relative to treatment effects. Wampold
et al.’s (2005) analysis showed that for disorders amenable to placebo and for which the
design was adequate, the placebo effect exceeded the effect of the treatment to which it
was being compared by 20%! Some might think that “relatively large” is an understate-
ment to describe placebo effects that are as good as or better than the medical treatments
to which they were compared. Hróbjartsson and Gøtzsche (this issue) found the word
“robust” to be “powerful spin” on placebo effects, but no one who has read the literature
on placebo effects would be offended by using the word “robust” to describe the consis-
tent scientific research that has documented the effects of placebos and has begun to
clarifybothitspsychologicalandbiochemicalmechanisms(seeWampold,Imel,&Minami,
this issue).
Hunsley and Westmacott’s (this issue) comments ignored or misinterpreted most of
the evidence and arguments presented by Wampold et al. (this issue; Wampold et al.,
2005). Hunsley and Westmacott made the claim that moderator variables were not exam-
ined; the entire basis of the reanalysis of Hróbjartsson and Gøtzsche (2001) was to show
that amenability to placebo action and adequacy of research design had moderated the
effects. They then stated that ratings of adequacy of design were “irrelevant as a moder-
ator of placebo effects.” Not so, when you consider placebo effects vis-à-vis treatment
effects, which was the point of our analysis. For example, for continuous variables and
treatments amenable to placebos, for well-designed studies the ratio of placebo effects to
treatment effects for adequate and inadequate designs was 1.20:1 and .28:1, respectively.
Hunsley and Westmacott’s (this issue) response narrowly focused on the clinical
implications of an effect size of .29. For the most part, the trials reviewed by Hróbjarts-
son and Gøtzsche (2001) were not designed to ascertain the clinical effects of placebos
but rather to establish the efficacy of a particular treatment, a point also made earlier
(Wampold et al., this issue). For example, included in Hróbjartsson and Gøtzsche’s (2004)
article is a study that tested the analgesic effects of saline injections in neonates which, as
Wampold et al. (this issue) noted, would not by any extant theory of placebos be expected
to produce an effect. Moreover, Hunsley and Westmacott recapitulated Hróbjartsson and
Gøtzsche’s (this isse) error by not citing Vase, Riley, and Price (2002), who found that the
placebo effect in studies designed to detect analgesic effects of placebos were several
multiples greater than those found in the trials reviewed by Hróbjartsson and Gøtzsche.
Despite the fact that the purpose of Wampold et al. (2005) was not to make inferences
about the clinical utility of placebos, and no conclusions were made in that regard, it is
instructive to compare the number needed to treat (NNT) of 7 for the placebo effect under
402
Journal of Clinical Psychology, April 2007
Journal of Clinical PsychologyDOI 10.1002/jclp
Page 3
discussion (i.e., 7 patients need to be treated to achieve one success by means of a pla-
cebo) to an NNT of 129 for aspirin as a prophylaxis for heart attacks, an accepted medical
practice based on a clinical trial that was discontinued because it was thought to be
unethical to withhold treatment from the control group (see Rosenthal, 1990). Moreover,
a perusal of the Web site for the University of Toronto’s Centre for Evidence-Based
Medicine that Hunsley and Westmacott mentioned revealed that placebos are more effec-
tive than many medical practices, some of which are costly, including almost all inter-
ventions in cardiology (e.g., beta-blockers, angioplasty), geriatric medicine (e.g., calcium
andAlendronate sodium for osteoporosis), asthma (e.g., Budesonide), and influenza vac-
cine, among others. “Relatively large” seems apt.
As Wampold et al. (this issue) noted, the placebo has had a long history of being the
villain in the development of modern medicine. We care little about decontextualized and
atheoretical diatribes against placebos, and consequently, it was disappointing that the
scientific literature on placebos was ignored by Hróbjartsson and Gøtzsche (this issue) as
well as by Hunsley and Westmacott (this issue). We never intended to suggest that the
results of the meta-analyses of the trials reviewed by Hróbjartsson and Gøtzsche (2004)
would infer that saline injections for neonatal pain were indicated or, for that matter, that
any placebo should be administered clinically. Rather than quibbling whether the effects
from a corpus of studies ill-designed to detect a placebo effect are sufficiently large to be
clinically useful, attention should be devoted to the scientific study of placebos, which
has the potential to augment healing and promote wellness.
References
Hróbjartsson, A., & Gøtzsche, P. C. (2001). Is the placebo powerless? An analysis of clinical trials
comparing placebo with no treatment. New England Journal of Medicine, 344, 1594–1602.
Hróbjartsson, A., & Gøtzsche, P. C. (2004). Is the placebo powerless? Update of a systematic
review with 52 new randomized trials comparing placebo with no treatment. Journal of Inter-
nal Medicine, 256, 91–100.
Hróbjartsson, A., & Gøtzsche, P. C. (2007). Powerful spin on conclusion in Wampold and col-
leagues’ re-analysis of placebo vs. no-treatment trials despite similar results as in original
review. Journal of Clinical Psychology, 63, 373–377.
Hunsley, J., & Westmacott, R. (2007). Interpreting the magnitude of the placebo effect: Mountain
or molehill? Journal of Clinical Psychology, 63, 391–399.
Rosenthal, R. (1990). How are we doing in soft psychology?American Psychologist, 45, 775–777.
Vase, L., Riley, J. L., III, & Price, D. P. (2002).Acomparison of placebo effects in clinical analgesic
trials versus studies of placebo analgesia. Pain, 99, 443–452.
Wampold, B. E. (2001). The great psychotherapy debate. Mahwah, NJ: Erlbaum.
Wampold, B. E., Imel, Z. E., & Minami, T. (2007). The story of placebo effects in medicine:
Evidence in context. Journal of Clinical Psychology, 63, 379–390.
Wampold, B. E., Minami, T., Tierney, S. C., Baskin, T. W., & Bhati, K. S. (2005). The placebo is
powerful: Estimating placebo effects in medicine and psychotherapy from randomized clini-
cal trials. Journal of Clinical Psychology, 6, 835–854.
The Placebo Effect: Final Reply
403
Journal of Clinical PsychologyDOI 10.1002/jclp
Page 4
Download full-text 
