Research, Medicine, and "Placebos"

Full text

The American Journal of Bioethics-Neuroscience
REFERENCES
Howick, J. 2009. Questioning the methodologic superiority of
‘placebo’ over ‘active’ controlled trials. American Journal of Bioethics–
Neuroscience 9(9): 34–48.
ICH. 2000. ICH (International Conference on Harmonization)Harmo-
nized Tripartitie Guidelines Choice of Control Group and Related Is-
sues in Clinical Trials. E10, Centre for Biologics Evaluation and Re-
search. Available at http://www.ich.org/cache/compo/475-272-
1.html#E1
Moerman, D. E. 2000. Cultural variations in the placebo effect: Ul-
cers, anxiety, and blood pressure. Medical Anthropology Quarterly 14:
51–72.
Poole, C. 1987. Beyond the confidence interval. American Journal of
Public Health 77: 195–199.
Rennie, S. 2009 The FDA and Helsinki. Hastings Center Report 39(3):
49.
Rothman, K. J., Greenland, S., and T. L. Lash, eds. 2008. Modern
Epidemiology, Third Edition. Philadelphia, PA: Wolters Kluwer/
Lippincott Williams & Wilkins.
St ¨
urmer,T., Rothman, K. J., and Avorn,J. 2008. Pharmacoepidemiol-
ogy and in silico drug evaluation: Is there common ground? Journal
of Clinical Epidemiology 61: 205–206.
Sun, D. C., and Stempien, S. J. 1971. The Veterans administration
cooperative study on gastric ulcer. 3. Site and size of the ulcer as
determinants of outcome. Gastroenterology 61(Suppl 2): 576–584.
World Medical Association. 2008. Declaration of Helsinki: Ethical
Principles for Medical Research Involving Human Subjects. Available at
http://www.wma.net/e/policy/63.htm
Research, Medicine, and “Placebos”
Daniel E. Moerman, University of Michigan–Dearborn
People, being more complex than parachutes or socket
wrenches, require more subtle things be done in order to un-
derstand them. “Placebo controlled trials” are one of them.
Jeremy Howick in his recent article comparing ‘active’ and
‘placebo’ controlled trials, seems even to this anthropolo-
gist to be remarkably distanced from the medical treatment
that he is concerned about (Howick 2009). The notion that
before the mid-19th century, most medical treatments were
“either no better than placebo, or worse,” clearly underval-
ues both medicine and placebo treatment, and suggests that
the Arabs, the Romans, the Greeks, or the Neanderthals for
that matter, could provide no useful medical care. He con-
tinues by noting that some contemporary treatments were
useless or harmful concluding that history teaches us “that
we cannot always assume that our existing treatments are
effective (more effective than placebo)” (34). In this sen-
tence he suggests that “placebo” doesn’t “do anything.” If
this were true, we would not need any sort of controlled
trial to determine medical effectiveness!
Howick (2009) proceeds to note that we do not need
a placebo-controlled trial to prove the effectiveness of a
parachute. This, however, is a delusional way to address
medical effectiveness. Howick fails to recognize the very
reason that we need controlled trials, namely that there
are biases and experiences in medicine that transcend his
simplistic mechanical metaphors. Some years ago, I pro-
posed a study with socket wrenches (Brit: socket spanner)
whereupon we would sneak into various mechanic shops,
Address correspondence to Daniel E. Moerman, William E. Stirton Professor Emeritus of Anthropology, University of
Michigan—Dearborn, 4901 Evergreen Road, Dearborn, MI 48128. E-mail: dmoerman@umich.edu
and randomly replace the proper mechanisms of the vari-
ous socket wrenches with sham ratchets (Moerman, 2002).
I noted that, were we to discover the next working day that
these wrenches worked as well tightening nuts as any other,
we would have good reason to be surprised, and would
probably want to report this in the press. But such surprises
happen in medicine all the time, so often that there is re-
ally little to report. A colleague, a primatologist, advises me
that, when infant chimpanzees (and other primates) become
anxious, upset and fuss and cry, their mothers pick them
up, and rock or otherwise comfort them, sometimes with
the breast, sometimes with some quiet grooming. Noth-
ing “characteristic” here, in Howick’s terms (2009, 34). But
in these simple acts, I would argue, lie the foundations of
medical care. Being comforted when you cry requires no
drugs, no placebos, nothing but a caring “person,” a mother
or child who can soothe the unhappy infant. And, more of-
ten than not, that’s all that’s required. I suppose there is
no point in mentioning that this approach would not be of
any particular use on an automobile engine, or a parachute.
“In the mechanistic tradition that still underlies much of
modern biomedicine, believing in the power of a placebo
to erase pain is as irrational as filling the gas tank of your
car with Earl Grey tea” (Morris 1997, 187). And this is pre-
cisely why we need placebo controlled trials when we are
going to make claims for the specific, active, or characteristic
quality of a particular pharmaceutical of the form “Xanax
(alprazolam) cures anxiety.”
64 ajob-Neuroscience September, Volume 9, Number 9, 2009

Methodologic Superiority
Suppose that, 5 years ago, a well controlled study
showed that Xyzzy was an effective treatment for gener-
alized anxiety disorder, that is, 60% of the subjects who
took Xyzzy reported a 2-week decline in the GAD-7 ques-
tionnaire score from 15 to 9 (“severe” to “mild”), while only
50% of those who took a placebo reported such a decline
(with two groups of 200 patients, this difference would be
significant at p =0.044; that is there is less than one chance in
20 that the result could occur just by chance). Now suppose
that this year, the new drug Ababa was tested not against
a placebo control, but against Xyzzy. Suppose Ababa were
superior to Xyzzy in this trial, and 50% of the Ababa pa-
tients were better after 2 weeks (GAD-7 score declined from
15 to 9), but only 40% of the Xyzzy patients had improved
that much (in this trial, the difference is also statistically
significant, at the same probability (p =0.044, as in the pre-
vious trial). Now, Ababa is not better than the placebo was
5 years ago; it is the same. And, of course, we have no no-
tion of what the effectiveness of the ‘non-characteristic’ care
would be now, since we did not test it.
Howick (2009) seems to think (although he does not
say it in so many words) that control group effects should
be constant. But they are not. He misstates my studies of
ulcer treatment saying that, while the healing rate of the
placebo treatment group varied dramatically (which it did),
that the healing rate of the drug group was constant, which
it certainly was not. Control group rates vary less than drug
rates, but the two are strongly correlated (r =0.40, p <
0.0000); that is, as the control-healing rate increases, so does
the healing rate of the drug (in 131 studies from around
the world) (Moerman, 2000). It is a plausible hypothesis
that effective drugs restrict the range of effects of inert
treatment.
Perhaps more to the point, there is good evidence to
show that these healing rates of the placebo treatment can
change in a good functional manner. Walsh and colleagues
(2002) have neatly shown that the effectiveness of drug treat-
ment for depression increased by 50% over the period from
1980 to 2000 regardless of the drug tested; tricyclics and
SSRIs increased from an average of about 40 percent in 1980
to about 60% in 2000, while the effectiveness of placebo
treatment increased from about 20% to about 30%. It also
seems plausible to attribute all these changes to the changed
attitude in medicine, and in the public, about the possibility
of effectively treating depression pharmaceutically over the
past generation. In the authoritative Pharmacological Basis of
Therapeutics, Fourth Edition (Goodman and Gilman 1970),
the authors conclude that the only clearly effective treat-
ment for depression is electro-convulsive therapy: “ECT
and imipramine may be given concurrently, but it is not
clear if this reduces the amount of ECT required” (191).
Clearly, things have changed, and the effectiveness of an-
tidepressant treatment (active or otherwise) has increased
with these changes in the meanings of the medications.
This means that ‘placebo effects’ will be a variable, and
only rarely a fixed value. This means that it is going always
to be the case that to know if a particular agent is “effective”
(that is, in Howick’s terms, “more effective than a placebo”
[2009, 34]) one must necessarily test it against a placebo.
Of course, a placebo is never effective. The one thing
about which we can be absolutely certain in any placebo
controlled trial is that unless someone is utterly incompe-
tent in designing the placebo—an inert substance—the one
thing to which we simply cannot attribute any patient im-
provement is the ‘placebo.’ It is inert. ‘Inert’ means it doesn’t
do anything. But of course people in control groups often get
better, and often at rates much higher than untreated peo-
ple would. If it’s not the ‘placebo,’ what is it? It’s a whole
panoply of symbols and meanings that change and evolve;
a key element is always physician enthusiasm. One simply
can not assume that since X >P, a n d Y >X, that Y >P.
People are more complicated than either socket wrenches
or parachutes. That is why we need placebo-controlled
trials. 
REFERENCES
Goodman, L. S., and Gilman, A. 1970. The Pharmacological Basis of
Therapeutics. New York, NY: The Macmillan Company.
Howick, J. 2009. Questioning the methodologic superiority of
‘placebo’ over ‘active’ controlled trials. American Journal of Bioethics–
Neuroscience 9(9): 34–48.
Moerman, D. E. 2000. Cultural variations in the placebo effect: Ul-
cers, anxiety, and blood pressure. Medical Anthropology Quarterly
14(1): 1–22.
Moerman, D. E. 2002. Medicine, Meaning, and the Placebo Effect. Cam-
bridge, UK: Cambridge University Press.
Morris, D. B. 1997. Placebo and belief: A biocultural model. In A.
Harrington, ed., The Placebo Effect: An Interdisciplinary Exploration.
Cambridge, MA: Harvard University Press.
Walsh, B. T., Seidman, A. N., Syusko, R., and Gould, M. 2002.
Placebo response in studies of major depression: Variable, substan-
tial, and growing. Journal of American Medical Association 287(14):
1840–1847.
September, Volume 9, Number 9, 2009 ajob-Neuroscience 65