PREVENTIVE MEDICINE 26, 607611 (1997)
ARTICLE NO. PM960124
REVIEW
The Nocebo Phenomenon: Concept, Evidence, and Implications for
Public Health
Robert A. Hahn, Ph.D., M.P.H.1
Division of Prevention Research and Analytic Methods, Epidemiology Program Office, Mailstop D-01, Centers for Disease Control and
Prevention, Atlanta, Georgia 30333
The nocebo hypothesis proposes that expectations
of sickness and the affective states associated with
such expectations cause sickness in the expectant. The
nocebo phenomenon is a little-recognized facet of cul-
ture that may be responsible for a substantial variety
of pathology throughout the world. However, the ex-
tent of the phenomenon is not yet known, and evi-
dence is piecemeal and ambiguous. This paper reviews
the concept of nocebo and its association with the pla-
cebo phenomenon, gives examples of evidence for the
nocebo phenomenon, and suggests public health im-
plications.
Key Words: nocebo; placebo; expectation; etiology.
INTRODUCTION
The nocebo hypothesis proposes that expectations of
sickness and the affective states associated with such
expectations cause sickness in the expectant. Result-
ant pathology may be subjective as well as objective
conditions. Some nocebo effects may be transient; oth-
ers may be chronic or fatal. An extreme form of the
nocebo phenomenon was described in Cannons classic
paper [7] as voodoo death. Because expectations are
largely learned from the cultural environment, nocebo
effects are likely to vary from place to place.
The nocebo phenomenon, first named by Kennedy
[17] and then elaborated by Kissel and Barrucand [19],
has not been systematically assessed. In this review, I
formulate a working definition of the nocebo phenom-
enon that relates nocebos and placebos, present a
range of examples of nocebo phenomena, and draw sev-
eral implications for public health.
A WORKING DEFINITION OF THE NOCEBO
PHENOMENON
The nocebo effect is the causation of sickness (or
death) by expectations of sickness (or death) and by
1
Fax: (404) 639-4463.
associated emotional states. Two forms of the nocebo
effect should be recognized: In the specific form, the
subject expects a particular negative outcome and that
outcome consequently occurs; for example, a surgical
patient expects to die on the operating table and does
dienot from the surgery itself, but from the expecta-
tion and associated affect [7,35]. In the generic form,
subjects have vague negative expectationsfor ex-
ample, they are diffusely pessimisticand their expec-
tations are realized in terms of symptoms, sickness, or
deathnone of which was specifically expected. Again,
expectation plays a causal role.
The nocebo phenomenon considered in this review is
distinct from placebo side effects (Fig. 1). Placebo side
effects occur when expectations of healing produce
sickness, i.e., a positive expectation has a negative out-
come. For example, a rash that occurs following admin-
istration of a placebo remedy may be a placebo side
effect. Diverse placebo side effects have been docu-
mented; one review reports an incidence of 19% in the
subjects of pharmacologic studies [29]. In the nocebo
phenomenon, however, the subject expects sickness to
be the outcome, i.e., the expectation is a negative one.
Nocebos may also have side effects, i.e., when negative
expectations produce positive outcomes or outcomes
other than those expected.
When Kennedy [17] and Kissel and Barrucand [19]
first referred to the nocebo phenomenon, they did not
distinguish placebo side effects from the effects of nega-
tive expectations. However, reference to voodoo death,
for example, as an instance of the placebo phenomenon
is etymologically inappropriate. Kennedy and Kissel
and Barrucand distinguished placebos from nocebos
only in terms of positive and negative outcomes, not
also in terms of expectations. Kennedys examples are
all placebo side effects, and Kissel and Barrucand did
not separate examples of placebo side effects from an
example of nocebo in the sense proposed here: 80% of
hospitalized patients given sugar water and told that it
was an emetic subsequently vomited. What distin-
607
0091-7435/97
608 ROBERT A. HAHN
FIG. 1. The Placebo Thesis: relations between expectation and
outcome. Reproduced, by permission of the publisher, from Ref. [14].
guishes nocebos is that the subject has negative expec-
tations and experiences a negative outcome. Schweiger
and Parducci [31] refer to nocebos as negative place-
bos.
Nocebos are causal in the same way that commonly
recognized pathogens are, e.g., cigarette smoke of lung
cancer, the tubercular bacillus of tuberculosis [15,38].
That is, nocebos increase the likelihood that the sick-
ness they refer to will occur, and this effect is not the
result of confounding, i.e., the empirical association of
the hypothesized nocebo with another cause of the con-
dition. None of these exposures is a necessary or a suf-
ficient cause of the given outcome.
EVIDENCE OF NOCEBO PHENOMENA
This review of evidence is divided according to the
source or manner of acquisition of expectations. It be-
gins with (1) the effects of inner, mental worlds and
moves to (2) the effects of nosological categories and
self-scrutiny, (3) sociogenic illness, or mass hysteria,
and (4) the deliberate induction of sickness or symp-
toms.
Inner, Mental World
Mood, affect, and some psychiatric conditions are of-
ten associated with negative expectations [1]. For ex-
ample, hopelessness is a prominent component of di-
verse forms of depression. Somatoform disorders such
as hypochondriasis and conversion disorder may also
be associated with expectations of pathology. Some
anxiety disorders, too, may be associated with expec-
tations of pathology. Panic disorder, for example, may
involve a sense of impending doom and a fear of
death [1].
Although several studies indicate an association of
negative expectations and affect with psychiatric con-
ditions and pathological outcomes [46,13,24,28,36,37],
only the study by Anda et al. [2] uses epidemiologic
methods to control for the confounding effects of other
risk factors. Anda et al. examined the effects of depres-
sion on ischemic heart disease (IHD) incidence and
mortality among a sample of U.S. adults. They exam-
ined persons who were free from heart disease at the
outset of the study and excluded subjects whose initial
depressed affect might have been the consequence of
chronic disease. Depression was assessed from the
General Well-Being Schedule [11]. Anda and col-
leagues found that persons with depressive affect were
1.6 times more likely to have nonfatal IHD and 1.5
times more likely to have fatal IHD than persons who
did not have depressive affect, independent of other
known risk factors for IHD. These researchers also ex-
amined the effects of hopelessness on heart disease in-
cidence and mortality, and found a dose responsea
critical criterion in the inference of causality. Greater
hopelessness was associated with greater incidence
and mortality.
Considering that an 11.1% prevalence of depressed
affect was assessed in the study cohorta sample of
U.S. adultsit can be estimated (as the population
attributable risk) that approximately 26,000 deaths a
year (i.e., more than 5% of U.S. IHD mortality and
more than 1% of all U.S. deaths) may be attributable to
depression, independent of other risk factors. Mortality
associated with depressive expectations is an example
of the generic form of the nocebo phenomenon. The
other examples in this review are of specific nocebo
phenomena.
Nosological Categories and Self-Scrutiny
In one specific form, cardiac neurosis or cardiopho-
bia, patients are persistently fearful of heart attacks or
other cardiac symptoms, and report chest pain, de-
scribed by physicians as nonspecific. Although these
patients may not manifest recognized cardiac symp-
toms, there is evidence that belief that one is suscep-
tible to heart attacks is itself a risk factor for coronary
death. Eaker examined women, 45 to 64 years of age, in
the Framingham Study for the 20-year incidence of
myocardial infarction and coronary death [12]. Women
who believed they were more likely than others to suf-
fer a heart attack were 3.7 times as likely to die of
coronary conditions as were women who believed they
were less likely to die of such symptoms, independent
of commonly recognized risk factors for coronary death
(e.g., smoking, systolic blood pressure, and the ratio of
total to high-density lipoprotein cholesterol).
Sociogenic Illness
Sickness or symptoms may also occur when one per-
son observes or learns of the sickness or symptoms in
others. Knowledge of sickness in others fosters an ex-
pectation that one may also be subject to the same
 NOCEBO: CONCEPT, EVIDENCE, AND IMPLICATIONS
condition. Perhaps the best recognized form of conta-
gion by observation are epidemics referred to as socio-
genic, psychogenic illness, mass hysteria, or, in the 30% more likely to have a child under 6 years of age.
workplace, assembly line hysteria [9].
Sirois [32] reviewed 78 documented outbreaks of
epidemic hysteria reported between 1872 and 1972.
Of these, 44% occurred in schools, 22% in towns, and
10% in factories. Twenty-eight percent involved fewer
than 10 persons, 32% involved 1030 persons, and 19%
more than 30 persons; 5% were of unreported magni-
tude. (Whereas the largest outbreak noted by Sirois
involved approximately 200 persons, an outbreak has
been described that involved 949 persons [23]). Only
females were involved in 74% of the outbreaks, only
males in 4% [32]. Outbreaks occurred more commonly
among persons from lower socioeconomic classes and in
periods of uncertainty and social stress. Convulsions
were reported in 24% of outbreaks, abnormal move-
ments in 18%, and fainting, globus/cough/laryngismus,
and loss of sensation in 11.5% each. Symptomatology
changed over the 100 years surveyed, from more glo-
bus/cough/laryngismus and abnormal movements to
more fainting, nausea, abdominal malaise, and head-
aches.
Colligan and Murphy [8] point out that sociogenic
outbreaks are commonly associated with a source be-
lieved to be related to the symptoms, e.g., a strange
odor or gas, new solvent, or an insect bite. However,
sometimes reported symptoms do not fit biomedical
knowledge of associations between potential toxins or
pathogens and pathophysiology. Persons affected often
have repetitive jobs, are under unusual stress, and/or
have poor relations with superiors. They may be in
poorer general health and have been absent more often
than persons who are not affected. Colligan and Mur-
phy indicate that sociogenic outbreaks in workplace
settings are substantially underreported.
Sirois [33] estimates that sociogenic outbreaks occur
in approximately 1 of every 1,000 schools per year in
the province of Quebec. A review of recent school out-
breaks in diverse countries indicates attack rates (i.e.,
the proportion of persons exposed who experience the
condition) of 648% [3].
The study of Kerckhoff and Back [18] of the 1962
June Bug outbreak in a Montana mill is one of the
few to carefully reconstruct social patterns of the
spread of a sociogenic condition. In the June Bug event,
those affected fainted or complained of pain, nausea, or
disorientation. Sixty-two (6.4%) of 965 workers were
affected, 59 (95.2%) of them women; all those affected
worked in dressmaking departments. Persons affected
were 70% more likely than controls to believe that the
cause of the outbreak was an insect or other physical
object. Persons affected were 62% more likely to have
worked overtime at least 2 or 3 times a week than
those not affected. Persons affected were less likely to
go to a supervisor with a complaint or to be members of
609
the union. They were 2.2 times as likely to be sole
breadwinners, 5.6 times as likely to be divorced, and
The outbreak began among women who were socially
isolated, subsequently spread among women connected
by links of close social relations, and finally diffused
among women less closely connected. The phenomenon
analyzed by Kerckhoff and Back might be described as
mass somatization.
The effects of a persons social environment on sick-
ness or illness behavior need not involve direct per-
sonal contact. An association has been found between
traumatic death or violence in the community environ-
ment and subsequent suicide or suicide-like behavior
[2527]. In this instance, the first victim serves as a
model with whom others may identify. For example,
when newspaper or television stories about a suicide
are released, the rate of suicide may increase in the
following week; the greater the circulation of the news-
paper, the greater the increase [25]. After Marilyn
Monroes suicide in 1962, 197 suicides occurred in the
United States during the following week12% more
than the number expected on the basis of past suicide
patterns [25]. A recent study indicates that teenagers
are more susceptible to televised publicity about sui-
cide and that increases in suicides are greater for girls
than for boys [27].
Motor vehicle fatalities also follow newspaper stories
of suicide. Phillips [26] calculates that, on average, mo-
tor vehicle fatalities increase 9% above the expected
rate in the week following front-page reporting of sui-
cides in newspaper stories and that, when the newspa-
per has greater than average circulation, the increase
is 19%.
Sickness/Symptoms Induced
Social psychologists have conducted diverse experi-
ments that demonstrate the effects of negative sugges-
tion on the experience of negative symptoms [16,20,
30,31,34]. In one experiment, 47.5% of asthmatics who
were exposed to (normally innocuous) nebulized saline
solution and told they were inhaling irritants or aller-
gens experienced substantially increased airway resis-
tance and changes in airway resistance and thoracic
gas volume [22]. Controls who did not have asthma
were unaffected by exposure to the same stimulus.
Twelve asthmatic subjects developed full-blown at-
tacks that were relieved by the same saline solution
presented therapeutically. (The researchers also refer
to an asthmatic patient in another study whose allergy
to roses was induced by plastic as well as natural roses,
indicating that the effect of the rose did not result en-
tirely from its botanical properties.)
In a follow-up, double-blind experiment, Luparello et
al. [21] randomized asthmatic patients to four condi-
tions: Two groups were given a bronchodilator, the
610 ROBERT A. HAHN
other two a bronchoconstrictor; half of the group given
each substance was told they were being given a bron-
chodilator, the other half that they were being given a
bronchoconstrictor. For each substance administered,
expectations induced by misinformation about the sub-
stance reduced its physiologic effectiveness by 43% (for
the bronchoconstrictor) and 49% (for the bronchodila-
tor).
Another study was designed to evaluate a method for
the diagnosis of psychogenic seizures, reported to ac-
count for as much as 20% of refractory epilepsy [20]. man culture. A societys culture tells its members how
Lancman and colleagues compared the effect of sugges-
tion on the induction of seizure behavior in patients
with psychogenic seizures and others with known epi-
lepsy. Patients were told that a medicine administered
through a skin patch would induce seizures within 30
sec, and that removal of the patch would end the sei-
zure. Of patients with psychogenic seizures, 77% mani-
fested seizures when the patch was applied, with
symptoms such as nonresponsiveness, generalized vio-
lent thrashing, and uncoordinated movements; 19% of
these patients reported auras, and 44% showed postic-
tal confusion and/or sleepiness. None of the patients
with diagnosed epilepsy manifested seizures.
Finally, another study [16], designed to evaluate a
controversial method of food allergy testing, compared
the effect of injecting the food substancesthe test to
be evaluatedwith the effect of injecting saline diluent
without the substance in question on symptoms that
included itching of the nose, watering or burning eyes,
plugged ears, tight or scratchy throat, nausea, dizzi-
ness, sleepiness, and depression. (Patients with a his-
tory of anaphylactic reactions or documented cardiac
irregularity or other severe reactions to their allergies
were excluded.) In this double-blind study, the propor-
tion of patients who experienced symptoms was not
statistically different in patients given test (27%) and
nocebo diluent injections (24%). Neutralizing injec- enhance our ability to minimize the pathological con-
tions, given to eliminate the reactions, were also
equally effective whether they contained the food sub-
stance orin this casethe diluent placebo. An injec-
tion becomes a nocebo (or placebo) not because of its
contents, but because of the pessimistic (or optimistic)
expectations of its consumer.
DISCUSSION
I have reviewed a range of studies indicating that
socially given negative expectations and their emo-
tional associations facilitate their own realization. Be-
liefs can make us sick as well as healthy. The nocebo
phenomenon is a little-recognized facet of culture that
may be responsible for a substantial variety of pathol-
ogy throughout the world. However, the extent of the
phenomenon is not yet known, and evidence is piece-
meal and ambiguous. There is evidence that inner,
mental states affect pathological outcomes, indepen-
dent of other risk factors; that symptoms may spread
in communities by being witnessed; and that symp-
toms may be caused by experimentally induced expec-
tations. Further investigations should explore the
ways in which, like the placebo phenomenon, the ex-
pectations of the nocebo phenomenon translate diverse
cultural beliefs into physiological process.
I conclude with two implications of the nocebo phe-
nomenon for public health.
First, the nocebo phenomenon is a side effect of hu-
the world is divided, interconnected, and known; it
specifies what is valued and what is not, what is good,
beautiful, right, wrong, and indifferent; it provides
rules of conduct whereby the societys members know
how to behave and how to judge the behavior of others
[14]. One element of cultures has been referred to as an
ethnomedicine. A societys ethnomedicine tells soci-
etal members what sicknesses there are, how they are
acquired, how manifested, how treated. The nocebo
phenomenon suggests that the categories of an ethno-
medicine may not only describe conditions of sickness,
but may also foster those conditions by establishing
expectations that they may occur. Thus, a cultural sys-
tem commonly thought to serve a healing function may
also have a contrary outcome, fostering the same pa-
thologies intended to be healed. The assessment of the
extent of this noxious facet of ethnomedicines, includ-
ing our own system of biomedicine, is an important
public health challenge.
Second, and more immediately practical, if commu-
nication about pathological conditions may serve not
only to describe, but, in a sense, also to foster sickness
by creating expectations, then we must be cautious in
both public health communications and in clinical
medicine. We need to know more about how health
messages affect their audience. Such knowledge may
sequences of negative messages. The placebo/nocebo
phenomenon suggests that it may be healthier to err on
the side of optimism than on the side of pessimism.
REFERENCES
1. American Psychiatric Association. Diagnostic and statistical
manual of mental disorders, 3rd ed. Washington: Am. Psychiatr.
Assoc., 1980.
2. Anda R, Williamson D, Jones D, Macera C, Eaker E, Glassman
A, Marks J. Depressed affect, hopelessness, and the risk of isch-
emic heart disease in a cohort of U.S. adults. Epidemiology
1993;4(4):28594.
3. Arcidiacono S, Brand JI, Coppenger W, Calder RA. Mass socio-
genic illness in a day-care centerFlorida. MMWR 1990;31(18):
3014.
4. Black DW, Warrack G, Winokur G. Excess mortality among psy-
chiatric patients. JAMA 1985;253(1):5861.
5. Black DW, Warrack G, Winokur G. The Iowa record-linkage
study. II. Excess mortality among patients with functional dis-
orders. Arch Gen Psychiatry 1985;42:828.
 NOCEBO: CONCEPT, EVIDENCE, AND IMPLICATIONS
6. Bruce ML, Leaf PJ, Rozal GPM, et al. Psychiatric status and 9
year mortality data in the New Haven epidemiologic catchment
area study. Am J Psychiatry 1994;151(5):71621.
7. Cannon WB. Voodoo death. Am Anthropol 1942;44(2):16981.
8. Colligan MJ, Murphy LR. Mass psychogenic illness in organiza-
tions: an overview. J Occup Psychol 1979;52:7790.
9. Colligan MJ, Stockton W. The mystery of assembly-line hysteria.
Psychol Today 1976;12:93116.
10. Conti S, Savron G, Bartolucci G, et al. Cardiac neurosis and
psychopathology. Psychother Psychosom 1989;52:8891.
11. Dupuy HJ. A concurrent validational study of the NCHS Gen-
eral Well-Being Schedule. Washington: U.S. Govt Printing Of-
fice, 1977; Vital and Health Statistics, Ser. 2, No. 73, DHEW
Publication No. (HRA)78-1347.
12. Eaker E, Pinsky J, Castelli WP. Myocardial infarction and coro-
nary death among women: psychosocial predictors from a 20-
year follow-up of women in the Framingham Study. Am J Epi-
demiol 1992;135:85464.
13. Friedman HS, Booth-Kewley S. The disease-prone personality:
a meta-analytic view of the construct. Am Psychol 1987;42(6):
53955.
14. Hahn RA. Sickness and healing: an anthropological perspective.
New Haven (CT): Yale Univ. Press, 1995.
15. Harris HW, McClement JH. Pulmonary tuberculosis. In
Hoeprich PD, editor. Infectious diseases. New York: Harper &
Row, 1983:378404.
16. Jewett DL, Fein G, Greenberg MH. A double-blind study of
symptom provocation to determine food sensitivity. N Engl J
Med 1990;323(7):42933.
17. Kennedy WP. The nocebo reaction. Med World 1961;91:2035.
18. Kerckhoff AC, Back KW. The June Bug: a study of hysterical
contagion. New York: Appleton-Century-Crofts, 1968.
19. Kissel P, Barrucand D. Placebos et effet placebo en medecine.
Paris: Masson, 1964.
20. Lancman ME, Asconape JJ, Craven WJ, Howard G, Penry JK.
Predictive value of induction of psychogenic seizures by sugges-
tion. Ann Neurol 1994;35(3):35961.
21. Luparello TJ, Leist N, Lourie CH, Sweet P. The interaction of
psychologic stimuli and pharmacologic agents on airway reac-
tivity in asthmatic subjects. Psychosom Med 1970;32(5):50913.
22. Luparello T, Lyons HA, Bleecker ER, McFadden ER. Influences
of suggestion on airway reactivity in asthmatic subjects. Psycho-
som Med 1968;30:81925.
611
23. Modan B, Tirosh M, Weissenberg E, et al. The Arjenyattah epi-
demic: a mass phenomenon: spread and triggering factors. Lan-
cet 1983;24/31:14724.
24. Newman SC, Bland RC. Mortality in a cohort of patients with
schizophrenia: a record linkage study. Can J Psychiatry 1991;
36:23945.
25. Phillips DP. The influence of suggestion on suicide: substantive
and theoretical implications of the Werther effect. Am Soc Rev
1974;39:34054.
26. Phillips DP. Motor vehicle fatalities increase just after publi-
cized suicide stories. Science 1977;196:14645.
27. Phillips DP, Carstensen LL. Clustering of teenage suicides after
television news stories about suicide. N Engl J Med 1986;315:
6859.
28. Reich P. Psychological predisposition to life-threatening ar-
rhythmias. Annu Rev Med 1985;36:397405.
29. Rosenzweig P, Brohier S, Zipfel A. The placebo effect in healthy
volunteers: influence of experimental conditions on the adverse
events profile during phase I studies. Clin Pharmacol Ther
1993;54(5):57883.
30. Schachter S, Singer JE. Cognitive, social, and physiological de-
terminants of emotional state. Psychol Rev 1962;69(5):37999.
31. Schweiger A, Parducci A. Nocebo: the psychologic induction of
pain. Pavlov J Biol Sci 1981;16:1403.
32. Sirois F. Epidemic hysteria. Acta Psychiatr Scand 1974;51(252):
744.
33. Sirois F. A propos de la frequence des epidemies d9hysterie.
Union Med Can 1975;104:1213.
34. Sternbach RA. The effects of instructional sets on autonomic
responsivity. Psychophysiology 1964;1(1):6772.
35. Weisman AD, Hackett TP. Predilection to death: death and dy-
ing as a psychiatric problem. Psychosom Med 1961;23(3):232
56.
36. Weissman MM, Markowitz JS, Ouellette R, Greenwald S, Kahn
JP. Panic disorder and cardiovascular/cerebrovascular prob-
lems: results from a community survey. Am J Psychiatry 1990;
147:15048.
37. Wells KB, Stewart A, Hays RD, et al. The functioning and well-
being of depressed patients. JAMA 1989;262(7):9149.
38. United States Surgeon General. Reducing the health conse-
quence of smoking. 25 years of progress. Rockville, MD: U.S.
Public Health Service, 1989.