Sei sulla pagina 1di 12

Volume 19, Number 1 The Journal of Clinical Ethics 58


Placebo Use in Clinical Practice:

Report of the American Medical Association
Council on Ethical and Judicial Affairs

Nathan A. Bostick, Robert Sade, Mark A. Levine,

and Dudley M. Stewart, Jr.

INTRODUCTION prescribed by the physician.2 Although there is

some debate as to the origins of the placebo ef-
The use of placebos in research has received fect,3 much has been learned in recent years
much more attention than has their use in clini- regarding its anatomical and physiological foun-
cal practice. This report is intended to guide dations.4
physicians clinical use of placebos in ways that
respect patients autonomy by allowing them to ETHICALLY APPROPRIATE PLACEBO USE
participate actively in the medical decision-
making process. Physicians administer placebos because pla-
For the purposes of this report, a placebo is cebos might relieve the symptoms that cause dis-
defined as a substance that the physician be- tress to their patients.5 Historically, physicians
lieves has no known specific pharmacological used placebos without patients knowledge, at
activity against the condition being treated. Pla- a time when they had great latitude in provid-
cebos can be therapeutically beneficial to some ing treatment without a patients consent if they
patients when they give rise to the so-called believed the intervention to be medically indi-
placebo effect.1 In general, this refers to a change cated.6 Accordingly, placebos often were used
in the patients condition that is attributable to to relieve pain or other complaints that appeared
the symbolic aspects of the overall care, rather to have no objective medical explanation.7 Such
than the medicinal qualities of the substance use of placebos could convey benefits derived
from the placebo effect or from the symbolic af-
Nathan A. Bostick, MA, MPP, is a Senior Research Associ- firmation of physicians willingness to help their
ate at the American Medical Association, Chicago, patients.8 The deception associated with placebo use,
Robert Sade, MD, a Professor of Surgery at the Medical however, is now widely viewed as problematic
College of South Carolina, Charleston. because it directly conflicts with contemporary
Mark A. Levine, MD, is an Associate Professor of Medicine notions of patient autonomy and the practice of
and Preventive Medicine/Biometrics at the University of Colo- shared decision making.9 Today, if physicians
rado, Denver. attempt to deceive patients by representing pla-
Dudley M. Stewart, Jr., MD, is a Clinical Professor of Psy- cebos as pharmacologically active medications,
chiatry at the Louisiana State University School of Medicine they risk undermining their patients trust.10
in New Orleans. 2008 by The Journal of Clinical Ethics. All Loss of trust is a serious consequence because
rights reserved. it is a foundational component of the patient-
Volume 19, Number 1 The Journal of Clinical Ethics 59

physician relationship.11 If trust is undermined, not pharmacologically active, namely, a pla-

patients may be less satisfied with their physi- cebo.19 By obtaining the patients cooperation
cians and therefore less likely to consult them in this manner, the physician need neither iden-
when making health-related decisions.12 More- tify which medication is the placebo nor seek
over, patients may not adhere to treatment rec- specific consent immediately before its admin-
ommendations when trust in their physician has istration.20 This example of shared decision
been compromised, thereby adversely affecting making demonstrates an approach that respects
patients overall health outcomes.13 the autonomy of patients and fosters trust within
Deceptive use of placebos poses other po- the patient-physician relationship. Moreover,
tential harms to patients, as well. For example, the authorized use of placebos is not expected
this use of placebos may mask and potentially to significantly diminish their clinical effective-
delay the treatment of medical conditions.14 ness as research suggests that little variation in
Furthermore, some patients may encounter ad- clinical outcomes is observed between patients
verse side-effects resulting from placebo use, an who are informed that they are to be treated with
occurrence known as the nocebo phenomenon.15 placebos and patients who are administered pla-
Ultimately, the deceptive use of placebos is cebos in a deceptive manner.21
not ethically acceptable because it may harm When physicians are faced with significant
patients to a greater degree than it helps them. clinical or diagnostic uncertainty, the authorized
This is particularly true in cases when place- use of placebos may prove particularly valuable
bos are utilized to serve the convenience of the for conducting single-patient controlled stud-
physician rather than to promote the welfare of ies, known as N-of-1 trials.22 In these trials, a
the patient. Perhaps the most pernicious use of disease-specific intervention and a placebo are
placebos is for mollifying a patient who is de- alternated through several treatment cycles; the
manding, displays a difficult personality, or has duration of each cycle depends on the nature of
a complex problem that has become frustrating the disease. 23 Such studies can be single-
to the physician. Placebos should never be used blinded, in which only the patient is unaware
in this way because it is fundamentally incon- of which drug is being administered, or they
sistent with physicians professional obligations can be double-blinded, in which the assignment
to promote patients welfare and respect the of treatment is managed by a third person, such
autonomy of patients. as a pharmacist, so neither patient nor physi-
In some instances, it may be most appropri- cian knows which medication is in use. In ei-
ate to forego the use of placebos altogether. Sev- ther case, the patient keeps a detailed journal
eral studies have described placebo-like effects of the waxing and waning of symptoms. At the
that lead to better health outcomes when phy- studys conclusion, the physician can differen-
sicians are able to comfort and reassure pati- tiate between benefits attributable to the phar-
ents presenting with symptoms that do not ap- macologically active drug and to the placebo.24
pear to have a clear medical basis.16 This seems Throughout this process, the patients progress
to work best when physicians establish partner- should be monitored and the placebo discon-
ships with patients that are built on respect and tinued if the active agent is found clearly to be
trust, and encourage adherence to treatment more effective.25
In other instances, physicians may utilize CONCLUSION
placebos within their clinical practice without
relying on the act of deception. In these cases, Placebos are substances that the physician
physicians should make decisions regarding the believes have no specific pharmacological ac-
use of placebos in partnership with their pati- tivity against the condition being treated. They
ents.18 For example, a physician could explain may be used in clinical practice to determine a
to a patient that a more certain diagnosis or bet- diagnosis or appropriate treatment in the face
ter understanding of his or her condition could of clinical uncertainty. Physicians must avoid
be achieved by evaluating the effects of differ- deception when administering placebos by in-
ent types of medication, including one that is forming the patient that a placebo may be used.
Volume 19, Number 1 The Journal of Clinical Ethics 60

RECOMMENDATIONS The American Medical Association Coun-

cil on Ethical and Judicial Affairs gratefully ac-
The Council on Ethical and Judicial Affairs knowledges the following individuals and or-
recommends the following. ganizations for their efforts to review previous
A placebo is a substance provided to a drafts of this report: The American Academy of
patient that the physician believes has no spe- Pain Medicine Council on Ethics; Paul
cific pharmacological effect upon the condi- Appelbaum, MD, the American Psychiatric As-
tion being treated. In the clinical setting, the sociation Council on Psychiatry and Law; Perry
use of a placebo without the patients knowl- Fine, MD, Professor of Anesthesiology, Univer-
edge may undermine trust, compromise the sity of Utah; Richard Milone, MD, Past Chair,
patient-physician relationship, and result in the American Psychiatric Association Ethics
medical harm to the patient. Committee; Laura Roberts, MD, Chair, the
Physicians may use placebos for diagno- American Psychiatric Association Task Force to
sis or treatment only if the patient is informed Update the Ethics Annotation.
of and agrees to its use. A placebo may still
be effective if the patient knows it will be used NOTES
but cannot identify it and does not know the
precise timing of its use. A physician should 1. H. Brody, Placebo, in Encyclopedia of
enlist the patients cooperation by explaining Bioethics, ed. W.T. Reich (New York: Simon &
that a better understanding of the medical Schuster Macmillan, 1995), 1951-3.
condition could be achieved by evaluating the 2. Ibid.; H. Brody, The placebo response:
effects of different medications, including the Recent research and implications for family
placebo. The physician need neither identify medicine, Journal of Family Practice 49, no. 7
the placebo nor seek specific consent before (2000): 649-54.
its administration. In this way, the physician 3. A. Hrbjartsson and P.C. Gtzsche, An
respects the patients autonomy and fosters a analysis of clinical trials comparing placebo
trusting relationship, while the patient still may with no treatment, New England Journal of
benefit from the placebo effect. Medicine 344 (2001): 1594-602; J. Glausiusz, Is
A placebo must not be given merely to the placebo effect a myth? Discover, 14 Sep-
mollify a difficult patient, because doing so tember 2001.
serves the convenience of the physician more 4. K. Irizarry and J. Licinio, An explana-
than it promotes the patients welfare. Physi- tion for the placebo effect? Science 307, no.
cians can avoid using a placebo, yet produce 5714 (2005): 1411-2.
a placebo-like effect through the skillful use of 5. K. Thoms, The placebo in general prac-
reassurance and encouragement. In this way, tice, Lancet 334 (1994): 1066-7.
the physician builds respect and trust, pro- 6. I. Kleinman, P. Brown, and L. Librach,
motes the patient-physician relationship, and Placebo Pain Medication: Ethical and Practi-
improves health outcomes.26 cal Considerations, Archives of Family Medi-
cine 3 (1994): 453-7.
ACKNOWLEDGMENTS 7. Thoms, see note 6 above; P. Lichtenberg,
U. Heresco-Levy, and U. Nitzan, The ethics of
The Council on Ethical and Judicial Affairs the placebo in clinical practice, Journal of
at the time this report was written included Medical Ethics 30 (2004): 551-4.
Robert Sade, MD (Chair); Mark A. Levine, MD 8. H. Spiro, Clinical reflections on the pla-
(Vice-Chair); Regina Benjamin, MD, MBA; cebo phenomenon, in The Placebo Effect, An
Sharon Douglas, MD; Hillary Fairbrother, MD, Interdisciplinary Exploration, ed. A. Harrington
MPH; H. Rex Greene, MD; William Martinez, (Cambridge: Harvard University Press, 1997),
MS; John McMahon, Sr., MD, FACS, PhD; 37-55.
Priscilla Ray, MD; and Dudley M. Stewart, Jr., 9. Kleinman, see note 6 above; CEJA Opin-
MD. ion E-10.01, Fundamental Elements of the Pa-
Volume 19, Number 1 The Journal of Clinical Ethics 61

tient-Physician Relationship, www.ama-assn. ropharmacological dissection of placebo anal-

org/ama1/pub/upload/mm/369/ceja_6a07.pdf gesia: expectation activated opioid systems ver-
10. Ibid. sus conditioning activated specific systems,
11. CEJA, Fundamental Elements of the Journal of Neuroscience 19 (1999): 484-94; L.C.
Patient-Physician Relationship, see note 9 Park and L. Covi, Non-blind placebo trial: An
above. exploration of neurotic patients responses to
12. T. Beauchamp and J. Childress, Prin- placebo when its inert content is disclosed,
ciples of Biomedical Ethics, 5th ed. (New York: Archives of General Psychiatry 12 (1965): 336-
Oxford University Press, 2001). 45; D. Holmes and D. Bennett, Experiments to
13. D. Levy, White doctors and black pati- answer questions raised by the use of decep-
ents: influence of race on the doctor-patient re- tion in psychological research: I. Role playing
lationship, Pediatrics 75 (1986): 639-43; A. Kao as an alternative to deception. II. Effectiveness
et al., The relationship between method of phy- of debriefing after a deception. III. Effect of in-
sician payment and patient trust, Journal of the formed consent on deception, Journal of Per-
American Medical Association 280 (1998): 1708- sonality and Social Psychology 29 (1974): 358-
14; D. Mechanic, The functions and limitations 67.
of trust in the provision of medical care, Jour- 22. L. Irwig, P. Glasziou, and L. March, Eth-
nal of Health Politics, Policy and Law 23 (1998): ics of N-of-1 Trials, Lancet 345 (1995): 469.
661-86. 23. C. Nikles, A. Clavarino, and C. Del Mar,
14. P. Hbert et al., Bioethics for clinicians: Using n-of-1 trials as a clinical tool to improve
Truth telling, Canadian Medical Association prescribing, British Journal of General Practice
Journal 156 (1997): 225-8. 55, no. 512 (2005): 175-80.
15. A.J. Barsky et al., Nonspecific medica- 24. J.D. Price and J.G. Evance, N-of-1 ran-
tion side effects and the nocebo phenomenon, domize controlled trials (N-of-1 trials) singu-
Journal of the American Medical Association larly useful in geriatric medicine, Age and
287 (2002): 622-7. Aging 31 (2002): 227-32.
16. Thoms, see note 5 above; K. Thomas, 25. Spiro, see note 8 above.
General practice consultations: Is there any 26. Adopted by the American Medical As-
point in being positive? British Medical Jour- sociation House of Delegates November 2006.
nal 294 (1987): 1200-2.
17. Brody, The Placebo Response, see note
2 above; L. Egbert et al., Reduction of post-op-
erative pain by encouragement and instruction
of patients, New England Journal of Medicine
270 (1964): 825-7.
18. CEJA Opinion E-10.01, Fundamental
Elements of the Patient-Physician Relationship,
see note 9 above; CEJA Opinion E-8.08, In-
formed Consent,
19. M. Barry, Involving Patients in Medi-
cal Decisions: How Can Physicians Do Better?
Journal of the American Medical Association
282 (1999): 2356-7.
20. Beauchamp and Childress, see note 12
above, pp. 84-6.
21. M. Amanzion and F. Benedetti, Neu-
Volume 19, Number 1 The Journal of Clinical Ethics 62

Placebos: Current Clinical Realities

Rachel Sherman and John Hickner

In general, the placebo effect disappears when the patient knows he/she is receiving a placebo. This is in itself
interesting, but since we believe in telling patients the truth, lying to patients in order to get the placebo effect
presents an ethical conundrum.

The use of placebos would be nullified if the patient knew. To use them when the patient does not know is unethical.

As a physician trained in an environment which emphasized the importance of trust, and the validation of
the individuals symptoms despite unexplainable physiology, I am opposed to the use of placebos
even if research were to show benefit. The end does not justify the means.

INTRODUCTION Believe in the Mind-Body Connection,2 is the

first significant U.S. study on placebo use in
The power of the placebo effect was recog- clinical practice since 1979,3 other than a study
nized in ancient times, as the following quota- of medicine interns from a single residency pro-
tion from Socrates, according to Plato, illus- gram published in 1999.4 Nearly half of the
trates: [The cure for the headache] was a kind Chicago-based academic internists who re-
of leaf, which required to be accompanied by a sponded to our survey reported using placebos
charm, and if a person would repeat the charm in clinical practice, and nearly all believe in
at the same time that he used the cure, he would their therapeutic potential. Yet, the use of pla-
be made whole; but without the charm the leaf cebos outside of clinical trials is a source of ethi-
would be of no avail.1 cal tension,5 as the three physician comments
Placebos remain clinically relevant and from our survey data (noted above the introduc-
philosophically interesting in the twenty-first tion) also illustrate.
century. While modern researchers often inves- In light of the ethical tension between har-
tigate placebos and the placebo effect within the nessing the therapeutic power of the placebo
context of experimental studies, there is little and respecting patients autonomy, and in the
data on the current use of placebos in clinical context of twenty-first century neuroscience hot
practice. Our January 2008 study, Academic in pursuit of the biological substrate of the mind-
Physicians Use Placebos in Clinical Practice and body connection,6 the publication of Placebo
Use in Clinical Practice: Report of the Ameri-
Rachel Sherman, is a Fourth-Year Medical Student at the can Medical Association Council on Ethical and
University of Chicago Pritzker School of Medicine, Judicial Affairs, in this issue of JCE is timely. It will, no doubt, stimulate further discourse on
John Hickner, MD, MSc, is Professor and Vice Chair of the the appropriate role of placebos in modern day
Department of Family Medicine at the University of Chicago medical practice. The AMA statement provides
Pritzker School of Medicine, jhickner@medicine.bsd. an initial set of guidelines regarding the ethical 2008, The Journal of Clinical Ethics. All rights use of placebos and attempts to reconcile pla-
reserved. cebo use with modern principles of patient au-
Volume 19, Number 1 The Journal of Clinical Ethics 63

tonomy, informed consent, and non-deceptive about informed consent. Similar to most of our
therapeutic practices. As a fourth-year medical surveyed physicians, Browns proposal, which
student and a family physician, neither an ex- he adapted from a previous study, does not ex-
pert in medical ethics, we will comment on the plicitly contain the word placebo: These pills
AMA placebo statement primarily in the con- do not contain any drug. We dont know exactly
text of physicians responses to our recent sur- how they work; they may trigger or stimulate
vey. The survey responses provide preliminary the bodys own healing processes. . . .8 Accord-
U.S. data about the current role of placebos in ing to our study results and Browns recommen-
clinical practice. dation, specific use of the word placebo is not a
customary part of current clinical practice.
EXAMINING THE AMA PLACEBO REPORT While our study offers new insight into phy-
IN THE CONTEXT OF CURRENT sicians behaviors and beliefs, our study is lim-
PHYSICIANS PRACTICES AND BELIEFS ited by its largely multiple-choice design, and
further qualitative research is warranted to gain
We surveyed 466 academic internal medi- better insight into the nuances of individual
cine physicians in the Chicago area in the sum- physicians behaviors and attitudes. Still, many
mer of 2006, and 50 percent responded.7 We physician respondents offered comments that
asked physicians about their use of placebos and frequently referred to the tension between pro-
their knowledge, attitudes, and beliefs about the moting the placebo effect and non-deceptive
placebo effect. Physicians defined placebo in a therapeutic practices. As a potential solution to
variety of ways, but the most commonly agreed what many physicians recognized as an ethi-
upon definition was an intervention not con- cal conundrum, several physicians suggested
sidered to have a specific effect through a known that placebo use would be appropriate only if
physiologic mechanism. Of the physicians we the patient agreed to the possibility of receiving
surveyed, 45 percent reported using placebos a placebo. One physician commented, As in a
in clinical practice. Still, 12 percent said that clinical trial, I would provide a placebo [in clini-
placebo use should be categorically prohibited cal practice] only if the patient knew that she/
in clinical practice, while the rest supported the he may receive a placebo during the course of
use of placebos in a variety of circumstances, treatment, but may not know at what point dur-
including when research supports its efficacy ing therapy. I think this would allow preserva-
(46 percent) and if the physician anticipates the tion of the placebo response. The AMA report
placebo will benefit the patient (31 percent). offers a similar recommendation, stating that the
About one in five respondents suggested pla- patient must be informed and agree to receive a
cebo use was appropriate only after a patient placebo for medical diagnosis or treatment, but
was notified about receiving a placebo. Of the patient need not know the identity of the
those who reported placebo use, when asked placebo at the actual time of use.
about their personal practice, 4 percent intro-
duced the treatment using the word placebo. IS THE PLACEBO FOR THE PATIENT
Still, the majority offered information that may OR THE PHYSICIAN OR BOTH?
have accurately described the nature and pur- The AMA report states that placebos must
pose of the proposed treatment. For instance, not be given merely to mollify a difficult pa-
34 percent introduced the placebo as a sub- tient, because doing so serves the convenience
stance that may help you and will not hurt, of the physician more than it promotes the pati-
and 9 percent as medicine with no specific ef- ents welfare. In 1979, Goodwin, Goodwin, and
fect. Another 33 percent individualized their Vogel found that 75 percent of the physicians
response to this question with statements such surveyed had ordered a placebo for a problem
as, This may help you but I am not sure how it patient, a patient that the nursing staff was com-
works. At the time our data were collected, plaining about.9 Although such use of place-
prior to the AMA report, psychiatrist Walter bos appears to be less common today, 15 per-
Brown proposed a way to introduce a placebo cent of our physician respondents did report
to patients that, he believed, avoided concerns using placebos after unjustified demand for
Volume 19, Number 1 The Journal of Clinical Ethics 64

medication, and 6 percent had used a placebo percent), ibuprofen (12 percent), sub-therapeu-
to get the patient to stop complaining. Rather tic doses of medication (7 percent), and herbal
than prescribing a placebo in these types of cir- supplements (5 percent). Only a small minority
cumstances, the AMA encourages physicians to of physicians reported giving what may be con-
produce a placebo-like effect through the skill- sidered pure placebos, such as prepared pla-
ful use of reassurance and encouragement. cebo tablets (2 percent), saline infusions (3 per-
While reassurance and encouragement are in- cent), and sugar or artificial sweetener pills (1
dispensable physician practices and should be percent). These results suggest that the place-
taught as a central part of medical schools train- bos used by physicians in clinical practice are
ing courses on doctor-patient interaction, we be- rarely biologically inert substances, or dummy
lieve there may be situations when a prescrip- pills, which is how they are typically charac-
tion for a placebo may equally serve the conve- terized in research trials and in popular culture.
nience of the physician and provide supplemen- Unfortunately, impure placebos may have
tal therapeutic benefit for the patient. This thera- known potential negative side-effects and, in the
peutic value may be broadly conceived to ap- case of antibiotics, their overuse promotes drug-
ply to both disease-specific symptoms as well resistant infections. We believe further dialogue
as personality-dependent emotional states that regarding the appropriate use of impure pla-
contribute to a patients overall health, and may cebos in clinical practice is needed.
also influence disease-specific health outcomes.
Although the AMA report acknowledges the A placebo serves as a symbol of healing that
potential use of pharmacologically active medi- triggers positive therapeutic expectations in a
cations as placebos, it does not discuss the ethi- patient. We suggest that the definition of place-
cal implications of this practice. According to bos include but not be limited to a substance,
the AMA report, a placebo may be defined as a as defined by the AMA report. We suggest the
substance that has no known specific pharma- definition of placebos also include interventions
cological activity against the condition being or factors that have no known specific clinical
treated. The use of a pharmacologically active efficacy against the condition being treated.
medication for non-indicated conditions raises Physicians practices such as wearing a white
important ethical questions. For the purposes coat, or the physical examination of the patient
of our discussion, we will refer to pharmaco- (independent of diagnostic purposes) may serve
logically active placebos as impure placebos, as placebo treatments for patients. As one phy-
a term used by professor of law Adam Kolber, sician in our study commented, I have always
who, in contrast, called a biologically inert sub- wondered if the office physical exam is as much
stance a pure placebo. Kolber writes, Impure a sophisticated grooming ritual to relieve stress
placebos can be difficult to detect because the rather than obtain diagnostic information. In the
prescribed medication has a pharmacological outpatient setting it is typically normal but both
effect on some illnesses, and doctors may be able physician and patient are fairly attached to its
to provide plausible-sounding medical ration- performance.
ales for prescribing impure placebos.10 In another example of the symbolic value of
Our study found that physicians rarely pre- a placebo, physician David Watts, a gastroen-
scribed pure placebos. Rather, nearly all of terologist, poet, and writer, spoke of his experi-
the physicians who said they had prescribed a ence prescribing medication to patients that he
placebo prescribed impure placebos. Of the also suggested they may not need to take.
48 percent of physicians who reported giving Theres something about sitting down at the
at least one type of treatment in a situation when desk and writing it [the medication] out long-
there was no evidence of clinical efficacy. hand, tearing the prescription from its pad and
Among the treatments given, 33 percent re- handing it to him [the patient], taking it down
ported giving antibiotics for viral or other non- to the pharmacist who brings forth this amber
bacterial diagnoses, others gave vitamins (20 bottle with a childproof cap and 25 small white
Volume 19, Number 1 The Journal of Clinical Ethics 65

excretions of something wonderful. Something ACKNOWLEDGMENTS

about all of that that is just the right amount.11
This expanded definition of a placebo captures We would like to thank Howard Brody, MD,
the broader symbolic aspects of patient care that PhD, for reviewing our article and for his valu-
may trigger the placebo effect. able comments and advice.


As a matter of scientific inquiry, the power The quotations at the beginning of this ar-
of the placebo effect will continue to be re- ticle are anonymous physicians quotes from the
searched as an isolated variable, both in formal authors 2006 survey data.
research studies and during N-of-1 clinical tri-
als (trials in which a medication is tested in only 1. Plato, Charmides, or Temperance, trans.
one individual). Ultimately, in clinical practice, B. Jowett, Internet Classics Archive, http://
the separation of placebos and the placebo ef-
fect from other forms of therapy is somewhat 2. R. Sherman and J. Hickner, Academic
artificial. In the context of everyday medicine, physicians use placebos in clinical practice and
we believe the symbolic value of placebos and believe in the mind-body connection, Journal
the power of the placebo effect are best served of General Internal Medicine 23 (2008): 7-10.
not as isolated therapeutic tools, but rather as 3. J.S. Goodwin, J.M. Goodwin, and A.V.
integrated aspects of humanistic and holistic Vogel, Knowledge and use of placebos by house
patient care. In the year 2008, amidst great tech- officers and nurses, Annals of Internal Medi-
nological advances of modern medicine, the cine 91, no. 1 (1979): 106-10.
purpose of the ancient charm used by Socrates 4. J.T. Berger, Placebo medication use in
still resonates. For the charm will do more, patient care: A survey of medical interns, West-
Charmides, than only cure the headache. I dare ern Journal of Medicine 170, no. 2 (1999): 93-6.
say that you have heard eminent physicians say 5. S. Bok, Ethical issues in use of placebo
to a patient who comes to them with bad eyes, in medical practice and clinical trials, in The
that they cannot cure his eyes by themselves, science of the placebo: toward an interdiscipli-
but that if his eyes are to be cured, his head nary research agenda, ed. H.A. Guess et al. (Lon-
must be treated; and then again they say that to don: BMJ Books, 2002), 53-74.
think of curing the head alone, and not the rest 6. F. Benedetti et al., Neurobiological
of the body also, is the height of folly. And ar- Mechanisms of the Placebo Effect, Journal of
guing in this way they apply their methods to Neuroscience 25, no. 45 (2005): 10390-402.
the whole body, and try to treat and heal the 7. See note 2 above.
whole and the part together.12 8. W. Brown, Placebo as a treatment for de-
pression, Neuropsychopharmacology 10, no. 4
POSSIBLE ADDENDA TO THE AMA (1994): 265-9; L.C. Park and L. Covi, Non-blind
RECOMMENDATIONS ON PLACEBO placebo trial: an exploration of neurotic pati-
ents responses to placebo when its inert con-
We offer the following as possible additions tent is disclosed, Archives of General Psychia-
to the AMA report. try 12 (1965): 36-45.
1. Placebos, when used, should be a supple- 9. Goodwin, Goodwin, and Vogel, see note
ment and not a substitute for a clinically indi- 3 above.
cated treatment. 10. A.J. Kolber, A Limited Defense of Clini-
2. Placebos may be useful when there is no cal Placebo Deception, Yale Law and Policy
other effective treatment available for a patient. Review 26 (2007).
3. Pure placebos are generally safer than 11. D. Watts, Commentary: Placebo Effect,
impure placebos. Your Health, NPR, 6 August 2003.
4. Only safe impure placebos should be used 12. See note 1 above.
(for example, certain vitamins).
Volume 19, Number 1 The Journal of Clinical Ethics 66

Clinical Placebo Interventions Are Unethical,

Unnecessary, and Unprofessional
Asbjrn Hrbjartsson

INTRODUCTION year.3 Half of the general practitioners (48 per-

cent), and 10 percent of hospital-based physi-
The American Medical Association Coun- cians, said that they had used a placebo inter-
cil on Ethical and Judicial Affairs (CEJA) has vention more than 10 times within the last year.
written a report on the clinical use of placebo The most frequent reason given for placebo in-
interventions.1 The CEJA position, adopted by terventions was to follow the wish of the pa-
the American Medical Association House of tient and avoid a conflict. The most frequent
Delegates, is that placebo interventions are eth- placebo interventions were antibiotics (for ex-
ically acceptable if the patient is informed about ample, for viral infections). Placebo interven-
the nature of the treatment, but that the infor- tions were considered ethically acceptable by
mation need not be such that the placebo inter- 46 percent of the respondents.
vention is clearly identified, nor is it necessary The other survey, which included 466 aca-
to seek specific consent before its administra- demic physicians (with a response rate of 50
tion. In the following I will argue that this posi- percent) at three Chicago-based medical schools,
tion is not tenable. found that 45 percent of the respondents said
that they had prescribed a placebo intervention
WHY AND HOW DO PHYSICIANS USE within the last year, and 8 percent had used a
PLACEBO INTERVENTIONS? placebo more than 10 times in the last year.4
The most common reason for using placebo
Within the last five years, two large surveys given was to calm the patient (18 percent).
of European or American clinicians have de- Other reasons included, for example, unjusti-
scribed how often, why, and how physicians fied demand for medication (15 percent), and
treat patients with placebo interventions.2 Our to get the patient to stop complaining (6 per-
survey of 772 Danish physicians (with a re- cent). In this survey also, the most typical pla-
sponse rate of 65 percent) found that 86 percent cebo intervention was antibiotics. Half of the
of general practitioners, and 54 percent of hos- respondents (46 percent) thought a placebo in-
pital-based physicians, said that they had used tervention was permitted when research sup-
a placebo intervention at least once in the last ported its efficacy. The respondents reported
that in 4 percent of the cases, patients were in-
Asbjrn Hrbjartsson, MD, MPhil, PhD, is Senior Re- formed, it is a placebo.
searcher at the Nordic Cochrane Centre, Rigshospitalet, The CEJA statement mentions the possibil-
Copenhagen, 2008 by The Journal of ity of nocebo effects, and thereby probably
Clinical Ethics. All rights reserved. thinks of placebo interventions as preparations
Volume 19, Number 1 The Journal of Clinical Ethics 67

containing no active drug, for example saline there are no expected additional benefits. The
injections or lactose tablets, so-called pure pla- physician clearly states this to the patient, who
cebos. This is in conflict with the findings of still wants the intervention. This is not a case
these surveys, that the typical placebo is a drug, of deception, but one of the physicians profes-
a so-called impure placebo, in many cases an sional integrity being in conflict with the pati-
antibiotic. The ethical implication is that drugs ents wishes. Should a physician follow the wish
have harmful effects. For example, antibiotics of a patient and prescribe a placebo interven-
can result in serious allergic reactions for the tion? (Given that any treatment is a placebo
individual being treated, and other harms; fur- when there is no expected additional benefit be-
ther, the unjustified use of antibiotics may cause yond that of the treatment ritual.) As stated,
the unnecessary development of bacterial resis- potential harmful effects to the individual pa-
tance, creating potential problems for future pa- tient and, in the case of antibiotics, potential
tients who are in true need of antibiotics. harm to future patients speak strongly against
The central section of the CEJA report starts, this practice. Further, to prescribe an interven-
Physicians administer placebos because pla- tion only because the patient wants it implies a
cebos might relieve the symptoms that cause substantial transformation of the patient-pro-
distress to their patients. The statement thereby vider relation. Ideally, a possibly imprudent
indicates that placebos typically are initiated treatment wish of a patient is checked by the
for the benefit of patients. The report does state, physicians professional considerations, and a
in the middle of the text, and without a heading possibly imprudent treatment suggestion by a
of its own, that it is not ethically acceptable to physician is checked by the patients wishes. If
use placebo interventions to serve the conve- a patients wishes overrule professional consid-
nience of the physician rather than to promote erations, the relation between patient and phy-
the well-being of the patient, however, the sician risks being transformed from one of mu-
prominent place given to the presentation of the tual respect and dialogue to one resembling that
ethically sound motive of helping the patient between customer and shopkeeper.
portrays the typical ethical situation as a di- Still, it is no simple task to decide when a
lemma between two valid ethical principles, treatment should be considered a placebo and
that is, between promoting patients well-being when it is part of a defensive treatment strat-
and respecting patients autonomy. In contrast, egy. In the case of antibiotics, one physician may
the surveys find that placebo interventions typi- conclude that a fever is viral and therefore con-
cally are initiated for the convenience of the sider antibiotics a placebo intervention. Another
physician. physician may conclude that the fever is most
likely viral, but the risk of bacterial infection is
CONVENIENCE PRESCRIPTIONS OF not negligible, and will therefore not consider
PLACEBO SHOULD BE AVOIDED antibiotics to be a placebo intervention. Clearly,
from a theoretical point of view, the two situa-
Convenience prescriptions of placebo inter- tions are very different, but from a practical per-
ventions involve two quite different ethical sce- spective, they may merge in the inevitable clini-
narios that are not considered in the CEJA re- cal uncertainty. Whether an intervention is con-
port. The first scenario involves a placebo treat- sidered a placebo or an active treatment depends
ment, for example to get the patient to stop on a physicians gut feeling about the expected
complaining, without informing the patient treatment benefit (beyond any effect of the treat-
that the treatment is a placebo. Such a practice ment ritual). A physicians gut feeling may be
is clearly unethical, as it implies deception that very different from a patients, and this tension
is unbalanced by benevolence, and that carries represents a challenge to the physician-patient
the risk of harmful effects. relationship. It is a challenge, however, that
The second scenario involves disagreement should be met with further dialogue and recon-
about treatment. A patient wants a treatment siderations about the expected benefits and
that the physician thinks is unnecessary because harms, and not by caving in and compromising
Volume 19, Number 1 The Journal of Clinical Ethics 68

professional integrity by prescribing ineffective REPLACE PLACEBO TREATMENTS WITH

and potentially harmful drugs. EMPATHIC CONSULTATION STRATEGIES

LITTLE EVIDENCE THAT PLACEBO An intervention, including a placebo inter-

INTERVENTIONS IN GENERAL HAVE vention, is but one of many components of the
CLINICALLY IMPORTANT EFFECTS patient-provider interaction. It is challenging to
study the various components of patient-pro-
The CEJA report implicitly takes for granted vider relationships and how the parties inter-
that the effects of placebos are clinically relevant act. Still, reviews indicate that the relation be-
and universal. In the 1950s, it was commonly tween a patient and a supportive person can
accepted that the effects of placebo were clini- have important effects on subjective and objec-
cally relevant for many patients who experi- tive outcomes. One Cochrane review on the ef-
enced both subjective and objective improving fect of a supportive person on labor concluded
outcomes in many clinical conditions.5 How- that it reduced the number of Caesarean sec-
ever, this was a misconception based on flawed tions, the number of women that needed anal-
methodology. In most cases comparisons had gesics, and the number of women with unpleas-
been based on changes from baseline in a pla- ant labours.8 There is no need for placebo in-
cebo group, and not on a more reliable compari- terventions. A better alternative is a consulta-
son between a placebo and a no-treatment group. tion strategy involving dialogue, empathy, in-
Our review of 114 randomized clinical tri- formation about relevant facts, and joint deci-
als comparing placebo with no-treatment groups sions about diagnosis and treatments.
found no statistically significant effect on bi- The CEJA mentions, In some instances, it
nary outcomes or on continuous-observer re- may be most appropriate to forego the use of
ported outcomes.6 We found that patients who placebos altogether. It would have been inter-
had received a placebo reported somewhat re- esting to know CEJAs suggestion of why this is
duced subjective symptoms compared with appropriate only in some instances, and not
untreated patients, but it was unclear to what always.
extent this was due to the bias inherent in
unblinded trials (for example, reporting bias) THE CEJA POLICY ON PLACEBO OPENS
and to which extent the effect was real. We also THE GATE TO A DANGEROUSLY
found a statistically significant effect of placebo SLIPPERY AREA
on pain, but the size of the improvement was
modest, corresponding to 6 mm on a 100 mm The report states that clinical placebo inter-
visual analogue scale. This result was repro- ventions may prove particularly valuable for
duced in 2004, when we updated the review conducting single patient controlled studies,
with more than 50 additional trials.7 Even as- known as N-of-1 trials. Such trials are a spe-
suming that an effect of placebo exists beyond cial version of randomized clinical trials. In this
the expected reporting bias, for example on pain, case an individual person with a reasonable
we know little about the situations that gener- stable chronic condition will engage in testing
ate such an effect. Our predicament is that a the effect of two or more treatments. Typically,
few clinical trials show substantial effects, and the patient will be blinded, for example by us-
many trials show no effects, or in some cases ing a placebo, and typically there will be sev-
harmful effects, and there is no clear pattern that eral periods of treatments in a random order.
explains this variation. However, n-of-one trials are rare, and their ethi-
Therefore, based on over 150 trials there is, cal problems are basically the ethical problems
on average, no evidence of clinically relevant of clinical trials. Therefore, CEJAs emphasis on
effects of placebo interventions over a broad n-of-one trials in a report on the use of placebo
range of clinical conditions. Therefore, the main in clinical practice is misplaced.
ethical argument for placebo prescriptions, that Still, the phrase particularly valuable has
such interventions generally will benefit pati- the implication that the report does not restrict
ents, is highly questionable. use of placebo to n-of-one trials. Placebo inter-
Volume 19, Number 1 The Journal of Clinical Ethics 69

vention, according to the report, may be valu- of the physician rather than for the well-being
able (if not particularly so) outside n-of-one tri- of the patient. Randomized trials generally find
als. One likely scenario envisaged by the CEJA no effects, or modest subjective effects, of pla-
could be a clinician in doubt about a treatment. cebo interventions. Placebo interventions could,
He or she then discusses this with the patient, and should, be replaced by empathic consulta-
and prescribes various medications, including tion strategies.
placebos, to try what works best, but without
the formal design of the randomized sequences NOTES
of n-of-one trial. However, such an approach is
unethical. Without the bias-reducing techniques 1. N.A. Bostick et al., Placebo Use in Clini-
of randomization of treatment periods, and cal Practice: Report of the American Medical
blinding procedures, informal experimentation Association Council on Ethical and Judicial Af-
has a very high risk of bias. Poorly conducted fairs, in this issue of JCE.
research is unethical, also when it comes to n- 2. A. Hrbjartsson and M. Norup, The use
of-one trials. From the patients perspective, it of placebo interventions in medical practice
is preferable to either be referred to another a national questionnaire survey of Danish cli-
physician with more knowledge about the clini- nicians, Evaluation & the Health Professions
cal problem or to participate in a properly con- 26 (2003): 153-65; R. Sherman and J. Hickner,
ducted n-of-one trial. Academic physicians use placebos in clinical
CEJA opens the gate to a dangerously slip- practice and believe in the mind-body connec-
pery area when they recommend that the infor- tion, Journal of General Internal Medicine 23
mation provided to patients need not be such (2008): 7-10.
that the placebo intervention is clearly identi- 3. Hrbjartsson and Norup, see note 2 above.
fied nor is it necessary to seek specific consent 4. Sherman and Hickner, see note 2 above.
before its administration. The surveys indicate The study was published electronically in Oc-
that physicians, when using placebo interven- tober 2007, which is considerably later than the
tions, often inform their patients in a purposely drafting of the CEJA report. However, the point
vague manner, by stating this is a substance here is not the specific study cited, but that none
that may help and will not hurt, or its a medi- of the several published studies of the clinical
cation, or this may help you but I am not sure use of placebo was discussed in the report.
how it works.9 This practice is clearly unethi- 5. H.K. Beecher, The powerful placebo,
cal because patients are unaware that they will Journal of the American Medical Association
receive a placebo. 159 (1955): 1602-6.
6. A. Hrbjartsson and P.C. Gtzsche, Is the
CONCLUSION placebo powerless? An analysis of clinical tri-
als comparing placebo treatment with no treat-
The CEJA recommendations are problematic ment, New England Journal of Medicine 344
from a clinical, research, and ethical perspec- (2001): 1594-602.
tive. The recommendations do not address 7. A. Hrbjartsson and P.C. Gtzsche, Is the
present placebo prescription practices, nor build placebo powerless? Update of a systematic re-
on systematic reviews of randomized trials com- view with 52 new randomised trials comparing
paring placebo with no-treatment. A revised placebo with no treatment, Journal of Internal
recommendation could include the following: Medicine 256 (2004): 91-100.
Clinical placebo interventions are unethi- 8. E. Hodnett et al., Continuous support for
cal, unnecessary, and unprofessional. Placebo women during childbirth, Cochrane Database
interventions are potentially harmful. First, pla- of Systematic Reviews issue 3 (2007): article no.
cebo interventions are often drugs that involve CD003766, http://www.cochrane. org/reviews/
a risk of harmful side-effects. Second, placebo en/ab003766.html.
interventions may damage patient-physician 9. Sherman and Hickner, see note 2 above.
trust considerably, because they often involve
deception and prescriptions for the convenience