Content uploaded by Assen Alladin
Author content
All content in this area was uploaded by Assen Alladin on May 12, 2015
Content may be subject to copyright.
This article was downloaded by: [184.64.123.170]
On: 11 May 2015, At: 17:26
Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954
Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
International Journal of Clinical
and Experimental Hypnosis
Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/nhyp20
The Efficacy of Hypnotherapy in
the Treatment of Psychosomatic
Disorders: Meta-analytical
Evidence
Erich Flammer
a
& Assen Alladin
b
a
Constance University , Konstanz, Germany
b
University of Calgary , Calgary, Alberta, Canada
Published online: 29 Oct 2010.
To cite this article: Erich Flammer & Assen Alladin (2007) The Efficacy of
Hypnotherapy in the Treatment of Psychosomatic Disorders: Meta-analytical Evidence ,
International Journal of Clinical and Experimental Hypnosis, 55:3, 251-274, DOI:
10.1080/00207140701338696
To link to this article: http://dx.doi.org/10.1080/00207140701338696
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information
(the “Content”) contained in the publications on our platform. However, Taylor
& Francis, our agents, and our licensors make no representations or warranties
whatsoever as to the accuracy, completeness, or suitability for any purpose
of the Content. Any opinions and views expressed in this publication are the
opinions and views of the authors, and are not the views of or endorsed by
Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and
Francis shall not be liable for any losses, actions, claims, proceedings, demands,
costs, expenses, damages, and other liabilities whatsoever or howsoever caused
arising directly or indirectly in connection with, in relation to or arising out of the
use of the Content.
This article may be used for research, teaching, and private study purposes.
Any substantial or systematic reproduction, redistribution, reselling, loan, sub-
Intl. Journal of Clinical and Experimental Hypnosis, 55(3): 251–274, 2007
Copyright © International Journal of Clinical and Experimental Hypnosis
ISSN: 0020-7144 print / 1744-5183 online
DOI: 10.1080/00207140701338696
251
NHYP0020-71441744-5183Intl. Journal of Clinical and Experimental Hypnosis, Vol. 55, No. 3, Apr 2007: pp. 0–0Intl. Journal of Clinical and Experimental Hypnosis
THE EFFICACY OF HYPNOTHERAPY
IN THE TREATMENT OF
PSYCHOSOMATIC DISORDERS:
Meta-Analytical Evidence
Hypnotherapy and Psychosomatic DisordersERICH FLAMMER AND ASSEN ALLADIN
ERICH FLAMMER
1
Constance University, Konstanz, Germany
ASSEN ALLADIN
University of Calgary, Calgary, Alberta, Canada
Abstract: Hypnotherapy is claimed to be effective in treatment of
psychosomatic disorders. A meta-analysis was conducted with 21
randomized, controlled clinical studies to evaluate efficacy of hypno-
sis in psychosomatic disorders. Studies compared patients exclu-
sively treated with hypnotherapy to untreated controls. Studies
providing adjunctive standard medical care in either treatment condi-
tion were also admitted. Hypnotherapy was categorized into classic
(n = 9), mixed form (n = 5), and modern (n = 3). Results showed the
weighted mean effect size for 21 studies was d
+
= .61 (p = .0000).
ANOVA revealed significant differences between classic, mixed, and
modern hypnosis. Regression of outcome on treatment dose failed to
show a significant relationship. Numerical values for correlation
between suggestibility and outcome were only reported in three stud-
ies (mean r = .31). The meta-analysis clearly indicates hypnotherapy
is highly effective in treatment of psychosomatic disorders.
Hypnosis has been claimed a useful therapeutic tool in psychoso-
matics (Pinnell & Covino, 2000), and there are numerous reports
about its application to a large variety of psychosomatic disorders
(e.g., Anbar, 2001; Banerjee, Srivastav, & Palan, 1993; Cedercreutz,
Lahteenmaki, & Tulikoura, 1976; Citron, 1968; Clarke, & Reynolds,
1991; Domangue, Margolis, Lieberman, & Kaji, 1985; Elton, 1993; Ewer,
& Stewart, 1986; Houghton, Heyman, & Whorwell, 1996; Rucklidge &
Saunders, 2002; Simon & Lewis, 2000; Tschugguel & Berga, 2003;
Younus, Simpson, Collins, & Wang, 2003). This claim is supported
by laboratory research demonstrating the influence of hypnosis on
Manuscript submitted February 14, 2006; final revision accepted August 26, 2006.
1
Address correspondence to Erich Flammer, Constance University, Department of
Psychology, Universitätsstraße 10, D-78464 Konstanz, Germany. E-mail: ErichFlammer
@aol.com
Downloaded by [184.64.123.170] at 17:26 11 May 2015
252 ERICH FLAMMER AND ASSEN ALLADIN
physiological systems mediating or being the target of psychosomatic
diseases. It has been shown that hypnosis exerts an influence on
immunological functioning. Kiecolt-Glaser, Marucha, Atkinson, and
Glaser (2001) analyzed the influence of hypnosis on immune function
in the presence of acute stress. In the hypnosis condition, the decre-
ment of immune function quantified by proliferative response to mito-
gens, percentages of CD3+ and CD4+ T-lymphocytes, and interleukin
1 production was significantly less compared to controls. Also, Gruzelier,
Smith, Nagy, and Henderson (2001) demonstrated that hypnosis can
buffer the effect of stress on immune function. Medical students at
exam time showed significantly less decline in natural killer (NK) cells
and CD8 cells as well as a significant increase in cortisol compared to
controls. There exists a substantial body of evidence for the influence
of hypnotic suggestions on dermal hypersensitivity reactions. Black
and colleagues demonstrated the possibility of reducing immediate-
type hypersensitivity responses (Black, 1963a) as well as a shift in
dose-response curve of hypersensitivity skin reactions following hyp-
notic suggestions (Black, 1963b). Fry, Mason, and Pearson (1964)
showed significantly smaller dermal reaction after the prick-test chal-
lenge in response to hypnosis in asthmatic patients. In another study
on dermal reaction to a prick-test challenge, Zachariae, Bjerring, and
Arendt-Nielsen (1989) were able to significantly reduce erythema area
in subjects receiving hypnotic suggestions as compared to the control
group. This study was replicated with asthmatic patients by Laidlaw,
Richardson, Booth, and Large (1994), who demonstrated that erythema
produced by a prick-test challenge was significantly reduced with
hypnosis compared to no hypnosis. Zachariae, Jorgensen, Egekvist,
and Bjerring (2001) also studied the influence of hypnotically induced
emotions on immediate-type hypersensitivity reactions to a prick-test.
In high hypnotizable subjects, the increase in erythema area was sig-
nificantly less while they experienced induced happiness or anger
compared to when they experienced induced sadness.
Regarding the cardiovascular system, Williamson et al. (2001) showed
that the manipulation of the sense of effort under hypnosis during
constant-load dynamic exercise was significantly associated not only
with changes in heart rate and mean blood pressure but also with changes
in regional blood flow in cortical regions, which are postulated to have
key roles in the central modulation of cardiovascular responses. More
recently, Jambrik, Sebastiani, Picano, Ghewlarducci, and Santarcangelo
(2005) provided evidence for the ability of hypnosis to prevent stress-
related reduction of peripheral vascular endothelial function. They found
the expected drop of postischemic flow-mediated vasodilatation (FMD)
under acute stress was prevented in high hypnotizable subjects. These
subjects also exhibited a significant increment of the basal arterial diameter
after hypnotic induction compared to the prehypnotic period. Moreover,
Downloaded by [184.64.123.170] at 17:26 11 May 2015
HYPNOTHERAPY AND PSYCHOSOMATIC DISORDERS 253
the increased heterogeneity of ventricular repolarization, a postulated
process in ventricular arrhythmia and sudden cardiac death (Hemingway,
Malik, & Marmot, 2001), under mental stress was significantly attenuated
by hypnotic relaxation (Taggart et al., 2005).
The influence of hypnosis on the respiratory and visceral systems
has also been studied. Ewer and Stewart (1986) found high hypnotiz-
able asthmatic patients to show significant improvement in bronchial
hyperresponsiveness and peak expiratory flow rate following hypno-
sis. In several studies, Prior, Colgan, and Whorwell (1990) demon-
strated that hypnosis was able to normalize abnormal visceral
sensations. Houghton, Calvert, Jackson, Cooper, and Whorwell (2002)
studied the effect of hypnotically induced emotions on visceral sensi-
tivity of the gut in patients with irritable bowel syndrome (IBS). While
hypnotic relaxation significantly increased rectal distension volume
required to cause discomfort compared to the nonhypnotic condition,
hypnotically induced anger reduced this threshold compared to the
nonhypnotic condition, hypnotic relaxation, and happiness.
While the available clinical reports, clinical studies, and laboratory
studies hint at the usefulness of hypnosis in psychosomatics, a system-
atic review of the efficacy of hypnotherapy in treating psychosomatic
illness has not yet been conducted. In this article, randomized, con-
trolled clinical studies will be systematically evaluated and meta-analytic
evidence for the efficacy of hypnotherapy in psychosomatic disorders
will be presented.
Meta-analytic methods allow estimation of the average outcome of a
treatment across a large number of studies with the focus not on statis-
tical significance of the single outcome but on the size of treatment
effect. This study on the efficacy of hypnosis will consider only ran-
domized, controlled trials that compare a patient group exclusively
treated by hypnosis with an untreated patient group. Because dura-
tions of follow-up periods are expected to be heterogeneous, for com-
putation of the effect sizes per study only the first measurement after
completion of treatment (“posttreatment”) was taken into account. To
ensure a neutral and comprehensible evaluation of the efficacy of hyp-
notherapy, all variables of a study for which sufficient information was
provided to calculate effect sizes were used for the computation of the
efficacy; that is, no selection of variables was made.
METHOD
Randomized, controlled trials ensure high relevance for a meta-
analytic study (Benson & Hartz, 2000; Concato, Shah, & Horwitz, 2000;
Matt & Navarro, 1997; Shadish & Ragsdale, 1996) and, therefore, in this
analysis, only randomized, controlled clinical studies that compared a
treatment and a waiting control group have been included.
Downloaded by [184.64.123.170] at 17:26 11 May 2015
254 ERICH FLAMMER AND ASSEN ALLADIN
Search Strategies
Relevant literature has been found by searching the databases
PsycLIT, Medline, and Dissertation Abstracts for the period 1887–2005
for clinical studies on the efficacy of hypnotherapy using the search
keys “Hypn*,” “Hypnotics,” “Psychother*” as well as a combination of
these by the operators “and,” “or,” “not.” Limits were set on “humans.”
Moreover, the literature reviews by Wadden and Anderton (1982),
Rhue, Lynn, and Kirsch (1993), and Kirsch, Montgomery, and
Sapirstein (1995) were used. In addition, other relevant studies were
identified by consideration of cited literature from papers already
examined (“footnote chasing”).
Criteria for Inclusion
To be eligible for the present meta-analysis, the studies had to meet
the following criteria:
(a) Inclusion of clinical studies only, that is, studies with psychotherapeu-
tic or medical indication for hypnotic intervention. Studies were
included if the efficacy of hypnotherapy was assessed in the treatment
of either patients with disorders that could be coded according to Inter-
national Classification of Diseases-10th ed. (ICD-10) or with patients
undergoing medical procedures (e.g., in dentistry, surgery, or cancer
treatment). However, for a study to be included in the analysis, it was
not required for the study to provide an explicit diagnosis based on
ICD-10 criteria. Studies that used hypnosis for treatment of warts were
also included. Studies that were intended to merely increase perfor-
mance without psychotherapeutic indication (e.g., improvement of ath-
letic or academic performances) were excluded.
(b) The exclusive use of hypnotic techniques, that is, a treatment condition
that applied only hypnotic interventions (hypnosis-only condition).
The combination of hypnotherapy with another form of psychotherapy
was excluded. In case of medical intervention, the studies that utilized
hypnosis in combination with standard medical care (e.g., medical care
in cancer patients) were included.
(c) The use of between-groups comparisons, that is, comparing a hypnosis-
only condition with a waiting control. The waiting control group was
excluded from any form of explicit psychotherapeutic intervention.
Studies that used hypnosis for supporting medical interventions (e.g.,
medical care for burn patients) and standard medical care, as well as
the control conditions, were also included into the meta-analysis.
(d) Randomized assignments of treatment and control conditions.
The above criteria were met by 91 studies. Out of this pool, 22 studies
that assessed the effectiveness of hypnotherapy in psychosomatic disor-
ders were selected. A disorder was defined as a psychosomatic disorder
(e.g., dyspepsia, IBS) if it met criteria for somatoform disorder (F45)
according to the ICD-10 classification. In addition, disorders were
Downloaded by [184.64.123.170] at 17:26 11 May 2015
HYPNOTHERAPY AND PSYCHOSOMATIC DISORDERS 255
included where the presence of psychological or behavioral factors have
played a major part in the etiology of physical disorders (ICD-10, F54)
(e.g., asthma, duodenal ulcer, dermatitis), or presumably affected a phys-
ical condition (e.g., headache, chronic pain, pain associated with chronic
disease, or enuresis). Studies that evaluated the efficacy of hypnosis with
cancer, viral infections (HIV-infection), and warts were excluded.
Coding of the Studies
All 22 studies were coded with regard to sample size (for treatment
group and for control group), patient characteristics, setting of treat-
ment, types of disorder treated, and the kinds of comparison (pre/
posttreatment comparison or comparison between treatment group
and control group). With respect to the hypnotic interventions utilized,
the studies were assigned into three categories: classical hypnosis, mod-
ern hypnosis, and mixed form of hypnosis. Classical hypnosis consisted of
direct suggestions (for relaxation, for alleviation of symptoms, and for
inducing imaginations or visualization), relaxation, and posthypnotic
suggestions. Hypnotic interventions that used indirect suggestions (for
relaxation, etc.), metaphors, and age regression were categorized as
modern hypnosis. Studies that primarily used classical interventions
but also included modern elements were also assigned to the category
of classical hypnosis. On the other hand, if predominantly modern
forms of hypnotic intervention were used but with some classical ele-
ments in them, these studies were assigned to the category of modern
hypnosis. In one study (Llaneza-Ramos, 1989), the investigator
reported the use of “Ericksonian hypnotherapy” with chronic head-
aches but gave no further information about treatment. This interven-
tion was coded as modern hypnosis. Studies that combined classical
and modern hypnosis without any predominance of classical or mod-
ern interventions were rated as mixed form of hypnosis.
COMPUTATIONS AND STATISTICAL ANALYSIS
To counteract a distorted estimation of the efficacy of hypnosis by
subjective selection of variables, no selection of dependent variables
for the individual study was made. Instead it was assumed that the
choice of dependent variables made by the investigators represented
an appropriate operationalization of the constructs examined. Doing
so allows the computation of the average effect size for the individual
studies to be replicated by other investigators as well.
For every dependent variable from a study, an effect size was com-
puted from the reported test statistic either relative to a control group
or in relation to pre-/posttreatment differences. As the measure for the
effect size, the standardized mean difference d was used (J. Cohen,
1988). For between-groups comparisons, the effect size was defined as
Downloaded by [184.64.123.170] at 17:26 11 May 2015
256 ERICH FLAMMER AND ASSEN ALLADIN
with = mean of treatment group at first posttreatment
measurement, = mean of control group at first posttreatment
measurement, and = pooled standard deviation from treatment
and control group. For pre-/posttreatment comparisons the effect size
was computed as above with = mean of treatment group at first post-
treatment measurement, = mean of treatment group at pretreat-
ment measurement and = pooled standard deviation from
pretreatment and posttreatment measurements. When insufficient data
were provided for a direct computation of effect sizes, they were calcu-
lated using the procedures described by Rosenthal (1984, pp. 20–31).
When the results were only reported to be not significant, the effect
sizes were assumed to be zero. For results that were only reported to
be significant, a p value of .05 was assumed and converted in an effect
size estimate (Rosenthal, pp. 26, 33). Because of the heterogeneity of
the available follow-up data, only those values that were recorded at
the first point (time) of measurement after termination of treatment
were used for the computation of effect sizes.
To calculate a mean effect size, different units of analysis can be
used. Effect sizes can be averaged either across individual dependent
variables, or across averaged effect sizes from individual trials. Aver-
aging across individual dependent variables results in a substantial
violation of the assumption of statistical independence, which is cru-
cial for standard inferential statistical tests (Gleser & Olkin, 1994). To
solve the problem of intercorrelations, only a single average effect size
was determined for every study, so that the effect sizes that were
included in the meta-analysis could be assumed to be independent
(Hunter & Schmidt, 1990). For the computation of the average study
effect size, one standardized mean difference was determined per
variable.
Because estimates of effect sizes from larger studies are more pre-
cise than estimates from smaller studies (Hedges & Olkin, 1985), study
level average effect sizes should be weighted to obtain unbiased mean
effect size estimates (Matt & Navarro, 1997). Weights are the inverse
estimated variances of the average effect size of a single trial (Hedges &
Olkin, 1985). The mean effect size d
+
was calculated as the average of
variance weighted trial level effect sizes (Hedges & Olkin, p. 111).
To determine whether studies yielding different effect sizes can be
reasonably described as sharing a common population effect size,
homogeneity was tested using a chi-square test (Hedges & Olkin, 1985,
p. 123). In case of heterogeneity of variance of the effect size, estimates
were decomposed into a variance component because of the variability
of the underlying population parameters and sampling error (Hedges &
Olkin, p. 194). If effect sizes exhibited variability beyond expectation
d
XX
SD
TC
pooled
=
−
X
T
X
C
S
D
pooled
X
T
X
C
SD
pooled
Downloaded by [184.64.123.170] at 17:26 11 May 2015
HYPNOTHERAPY AND PSYCHOSOMATIC DISORDERS 257
due to sampling error, a cluster analysis (Hedges & Olkin, pp. 265–283)
was conducted to identify homogeneous clusters. As a measure of
treatment success, the binomial effect size display (BESD) was com-
puted (Rosenthal & Rubin, 1982). The BESD (BESD = .50 ± r/2) refers
to the effect size r, that is, the point biserial correlation r(pb) between
treatment condition (hypnosis or waiting control) and therapy out-
come, which can easily be obtained from the standardized mean differ-
ence d by conversion (Rosenthal, 1984, p. 25). The BESD represents the
estimated difference between treatment group and control group with
regard to success rates. For example, a d of .50 (i.e., an r of .24) results
in binomial effect sizes of 38% and 62%. This means that without treat-
ment 38% of patients experience an alleviation of symptoms, but after
treatment 62% of patients can expect symptom alleviation.
Due to publication bias (studies yielding significant results may be
more likely to be published), reporting bias (no or insufficient informa-
tion about nonsignificant results provided by the investigators of a
study), or retrieval bias (inadequate key terms or limitation to some
specified languages), the available studies in a meta-analysis may not
be representative of all studies addressing the research question
(Greenhouse & Iyengar, 1994). To address these possible sources of
bias, a fail-safe N (Orwin, 1983) was determined. The fail-safe N indi-
cates the number of further studies with an average effect size of d = 0
that have to be included in the analysis in order to lower the mean
effect size to a predefined critical value. Further, a possible presence of
publication bias was assessed by a funnel plot (Light & Pillemer, 1984).
In general, effect sizes from larger studies are more precise (i.e., show
less variability) than those from smaller studies (Hedges & Olkin,
1985). This suggests that when effect sizes are plotted as a function of
sample size, the distribution should take the shape of an inverted fun-
nel (Mullen, 1989), provided no publication bias is present.
Statistical analyses were carried out using the SAS-JMP IN com-
puter package release 5.1 and the Meta-Analysis Programs computer
package version 5.3 by Schwartzer.
RESULTS
The characteristics of the 22 studies included in the meta-analysis
are summarized in Table 1.
In the first evaluation, the hypnotherapeutic methods employed in
the 22 studies were examined (see Table 1). The classical approach to
hypnosis was predominantly used in 45.5% (n = 10) of the trials,
while only 13.6% (n = 3) of the trials consisted of modern hypnosis.
The mixed form of hypnosis was used in 22.7% (n = 5) of the trials.
For 18.2% (n = 4) of the trials, the categorization of the hypnotherapy
could not be made.
Downloaded by [184.64.123.170] at 17:26 11 May 2015
258 ERICH FLAMMER AND ASSEN ALLADIN
Table 1
Characteristics of the 22 Studies Used in the Meta-Analysis
Number of studies including waiting control: 22
Total number of patients 1091
Dropouts (%) mean for 21 studies 12.2
Number of treated patients:
Hypnosis 420
Control 423
Dropouts (%):
Hypnosis (M for 20 studies) 5.9
Control (for 10 studies) 4.6
Age of patients:
Children /adolescents 2
Adults 13
Mixed 2
No specification 5
Sex of patients:
Female 1
Male 3
Mixed 16
No specification 2
Patient group:
Inpatients 1
Outpatients 18
Mixed 1
No specification 2
Setting:
Individual sessions 10
Group sessions 1
Mixed 1
No specification 10
Number of sessions (M for 20 trials) 6.3
Follow-up:
Trials with follow-up 14
Trials without follow-up 8
No specification 0
M (weeks) for 22 trials (with/without follow-up) 17.45
M for 14 trials (with follow-up) 29.54
Type of treatment:
Classical hypnosis 10
Modern hypnosis 3
Mixed 5
Indecisive 4
Suggestibility:
Suggestibility tested 12
Validated test of suggestibility used 11
Downloaded by [184.64.123.170] at 17:26 11 May 2015
HYPNOTHERAPY AND PSYCHOSOMATIC DISORDERS 259
Table 2 provides detailed information about the types of hypnotic
interventions used. Classical hypnosis was used in 86.4% (n = 19) of
the trials, while 36.4% (n = 8) of the studies employed modern methods
of hypnosis. Seventeen studies (77.3%) targeted solely on the psycho-
somatic symptoms and two studies (9.1%) focused on maladaptive
cognitions or irrational ideas, in addition to targeting somatic symp-
toms. Two studies (9.1%), in addition to using symptom-focused inter-
ventions, reported using hypnotherapy for facilitating expression of
emotions, gaining insight, and dealing with underlying causes of the
somatic symptoms. Therefore, most of the studies included in the
meta-analysis involved classical hypnosis.
Table 3 describes the individual studies with regard to their average
effect sizes, disorders, etc. Values from d = .20 to d = .50 are rated as
low, values from d = .50 to d = .80 as medium; and values of d > .80 are
regarded as large effect sizes (J. Cohen, 1988).
To provide a more accurate estimation of the mean effect size, one
study (Hill, 1981), yielding an extremely high between-groups effect
size of d = 7.07, was excluded from all analyses concerning between-
groups effect sizes. Computation of the weighted mean effect size for
21 studies resulted in d
+
= .61 (SE = .076, Z = 7.88; p = .0000). Figure 1
shows the stem-leaf diagram for the distribution of the weighted effect
sizes.
The unweighted mean effect size amounts to d = .83 (SE = .1465).
The BESD for d
+
= .61 are .36 and .65. This means that without treat-
ment 36% of the patients experience an alleviation of their symptoms,
after treatment, however, 65% of the patients can expect alleviation.
Table 2
Types of Hypnotherapeutic Interventions Used in the 22 Trials of the Meta-Analysis
Intervention Number of trials % of trials
Direct suggestions 8 36.36
Indirect suggestions 5 22.73
Symptom-orientated suggestions 20 90.91
Posthypnotic suggestions 3 13.64
Self-hypnosis 15 68.18
Relaxation 16 72.73
Imagery 8 36.36
Visualization 4 18.18
Displacement 2 9.09
Dissociation 2 9.09
Ego-strengthening 3 13.64
Age regression 2 9.09
Metaphors 3 13.64
Symbolization 1 4.55
Downloaded by [184.64.123.170] at 17:26 11 May 2015
260 ERICH FLAMMER AND ASSEN ALLADIN
Table 3
Details on the 22 Trials Used for the Meta-Analysis
Author(s) Disorder N*In-/
Outpatient
Sex Age group Setting
Hypnosis
Type
# of
Sessions
Effect Size
Between
Groups
Pre/Post
Attias et al., 1990 tinnitus 24 Out male adults Individual modern 4 .71 .71
Borkovec et al.,
1973
insomnia 19 Out female unknown Individual classical 3 .61
Calvert et al.,
2002
func. Dyspepsia 61 Out mixed unknown individual unknown 12 .73
Colgan et al., 1988 duodenal ulcer 30 Unknown mixed unknown group classical 7 .89
Edwards et al.,
1985
enuresis 24 Out male child./teens unknown classical 6 .00 1.08
Ewer & Stewart,
1986
asthma 39 Out mixed adults unknown classical 6 .16 .26
Galovski et al.,
1998
IBS 12 Out mixed adults unknown mixed 12 1.71
Gay et al., 2002 osteoarthritis 23 Out mixed adults unknown modern 8 1.36 1.04
Hill, 1981 stress 100 Out mixed adults unknown classical 3 7.07 2.62
Hoppe, 1984 chronic pain 29 Out mixed adults Individual mixed 8 .74 0.00
Kohen, 1995 asthma 14 Mixed mixed child./teens Individual mixed 2 1.17 0.00
Langewitz et al.,
2005
hayfever 66 Out mixed mixed unknown unknown unknown .59 .89
Downloaded by [184.64.123.170] at 17:26 11 May 2015
HYPNOTHERAPY AND PSYCHOSOMATIC DISORDERS 261
Llaneza-Ramos,
1989
chronic
headache
35 Out mixed unknown unknown modern 9 2.62
Maher-Loughnan
et al., 1962
asthma 55 Out mixed mixed unknown classical unknown .65 1.00
Melis et al., 1991 chronic
headache
26 Out mixed adults unknown unknown 4 .83
Moene et al., 2003 convrsn.
Disorder
44 Out mixed adults unknown mixed 10 .70 .35
Pearson, 1994 chronic pain 22 Unknown Unknown adults unknown classical 1 1.16 .98
Raskin et al., 1999 hypertension 24 In male adults unknown classical 4 .66 .46
Senser et al., 2004 atopic
dermatitis
33 Out unknown adults unknown mixed 12 2.00 1.04
Spanos et al., 1993 chronic
headache
57 Out mixed adults mixed classical ? .08 .24
ter Kuile et al.,
1994
chronic
headache
93 Out mixed adults unknown classical 6 .06 .10
Tosi et al., 1989 duodenal ulcer 13 Out mixed unknown unknown unknown 7 0.030.26
*both treatment and control groups.
Downloaded by [184.64.123.170] at 17:26 11 May 2015
262 ERICH FLAMMER AND ASSEN ALLADIN
Computation of the fail-safe N showed that in addition to the 21 stud-
ies included, further 56 studies with an average effect size of d = 0
would have to exist to reduce the average effect size from d = .61 to d =
.20. The funnel plot of the unweighted effect sizes as a function of sam-
ple size suggests the presence of publication bias, as the shape of the
distribution is asymmetric due to the absence of larger studies with
larger effect sizes (see Figure 2).
Apart from calculating a mean effect size for between-groups com-
parisons, a mean effect size for pre-/posttreatment comparisons was
computed for 16 out of the 22 studies providing the necessary statisti-
cal information. Doing so resulted in a mean weighted effect size of d+ =
.70 (SE = .0812) and in a BESD of 35% and 65%. The fail-safe N neces-
sary to reduce d = .70 to d = .20 amounts to 40 studies.
To test the hypothesis that the average effect sizes computed for
individual studies are estimates of a common population parameter, a
test of homogeneity (Hedges & Olkin, 1985) was carried out for the 21
studies. The test yielded an inhomogeneous distribution of effect sizes,
Q
(20)
= 49.74; df = 20; p = .00024, i.e., the effect sizes of the 21 studies
may not be regarded to stem from a common population. Decomposi-
tion of the variance of the effect sizes according to the method
described by Hedges and Olkin shows that 43.95% of the variability is
explained by sampling error. A subsequent cluster analysis (Hedges &
Olkin) identified three clusters at the 5% level of significance (see Table 4).
F
igure 1. Stem-leaf diagram for the 21 studies included.
Stem Leaf
6
0
7
4
22
89
66777777
2
2
2
2
2
1
1
1
1
1
0
0
0
0
0 0011
Count
1
1
1
1
2
2
8
1
4
Downloaded by [184.64.123.170] at 17:26 11 May 2015
HYPNOTHERAPY AND PSYCHOSOMATIC DISORDERS 263
F
igure 2. Unweighted effect sizes (ES) as a function of sample size for 21 studies.
0
20
40
60
80
100
120
N
–0,5
0 0,5 1 1,5 2 2,5 3
ES (mean
= .83)
Table 4
Results of Cluster Analysis for the 21 Studies Included
Cluster Author Disorder
1 Llaneza-Ramos (1989) chronic headache
2 Attias et al. (1990) tinnitus
Borkovec & Fowles (1973) insomnia
Calvert et al. (2002) functional dyspepsia
Colgan et al. (1988) duodenal ulcer
Galovski & Blanchard (1998) irritable bowel syndrome
Gay et al. (2002) osteoarthritic pain
Hoppe (1984) chronic pain
Kohen (1995) asthma
Langewitz et al. (2005) hay fever
Maher-Loughnan et al. (1962) asthma
Melis et al. (1991) chronic headache
Moene et al. (2003) conversion disorder
Pearson (1994) chronic pain
Raskin et al. (1999) hypertension
Senser et al. (2004) atopic dermatitis
3 Edwards & van der Spuy (1985) enuresis
Ewer & Stewart (1986) asthma
Spanos et al. (1993) chronic headache
ter Kuile et al. (1994) chronic headache
Tosi et al. (1989) duodenal ulcer
Downloaded by [184.64.123.170] at 17:26 11 May 2015
264 ERICH FLAMMER AND ASSEN ALLADIN
Tests of homogeneity confirmed Cluster 2, Q = 12.4269; df = 14; p =
.5721, and Cluster 3, Q = .1154; df = 4; p = .9984, to be homogenous.
To identify possible moderator variables, separate weighted analy-
ses of variance (ANOVA) for sex (male, female, mixed), age group
(children/adolescents, adults, mixed), setting (individual sessions,
group sessions, mixed), patient group (inpatients, outpatients, mixed),
and kind of hypnosis (classical, modern, mixed) were performed.
Weighting was derived using the inverse estimated variances of the
average effect sizes of a single trial. Due to the loss of degrees of free-
dom, these factors were not testable in a common ANOVA-model.
Prior to the analyses, a Shapiro-Wilks W test for normal distribution of
effect size estimates and Levene tests of equality of variances for the
ANOVA-models for age group, setting, patient group, and kind of
hypnosis were carried out. The test for normality failed to confirm the
normal distribution assumption (W = .8993; p = .034). Equality of vari-
ances for sex was, due to small sample sizes, not testable. All other
tests for equality of variances were not significant (all p > .10). The
weighted ANOVAs for sex, age group, setting, and patient group
showed no significant effect (all p > .05). The effect for kind of hypnosis
was significant, F = 5.9098; df = 2, 14; p = .0138. The weighted mean
effect sizes were d
+
= .33 (SE = .1068, n = 9), d
+
= .73 (SE = .1617; n = 5),
and d
+
= 1.42 (SE = .2566; n = 3) for classical hypnosis, the mixed form
of hypnosis, and modern hypnosis, respectively. The unweighted
mean effect sizes were d = .47 (SE = .1383) for modern hypnosis, d =
1.26 (SE = .2591) for the mixed form of hypnosis, and d = 1.56 (SE =
.5607) for classical hypnosis. Because under some circumstances non-
normality can affect the robustness of F tests in ANOVA (Miller,
1986), nonparametric Wilcoxon rank-sum tests for sex, setting, age
group, patient group, and kind of hypnosis were conducted to confirm
the results in the presence of nonnormality. Only the test for kind of
hypnosis was significant (χ
2
= 8.3974; df = 2; p = .015). All other tests
were not significant (all p > .05). Testing for homogeneity confirmed
homogeneity of variance for all three types of hypnotic intervention:
classical hypnosis, Q = 10.3356; df = 9; p = .2423, mixed form of hypno-
sis, Q = 2.4388; df = 4; p = .6556, and modern hypnosis, Q = 8.6753; df =
2; p = .0131.
To assess the relationship between the number of treatment sessions
and treatment outcome, a weighted linear regression model for the 21
studies included was fitted. The model contained the number of ses-
sions as regressor, the average effect size of trial as dependent variable,
and the inverse variances of the average effect sizes as weights. The
regression model accounted for only about 16% (R
2
= .1561) of the vari-
ability of the effect sizes and failed to be significant, F = 3.1668; df = 1,
16; p = .093. The coefficient for the number of sessions also failed to be
significant (see Table 5).
Downloaded by [184.64.123.170] at 17:26 11 May 2015
HYPNOTHERAPY AND PSYCHOSOMATIC DISORDERS 265
Finally, the effect of suggestibility was assessed. In 16 out of 22
trials, a test of suggestibility was used. Out of these, 14 tests were vali-
dated. Seven trials provided information about the correlation
between suggestibility and treatment outcome. Three of the trials
found a significant positive correlation, three trials found a positive
correlation, and one trial failed to find a correlation. Three of the trials
reported numerical values for the correlation coefficient r with a mean
of .31.
DISCUSSION
In this study, a meta-analysis on the efficacy of hypnosis with psy-
chosomatic disorders was conducted, which exclusively included ran-
domized, controlled clinical studies that compared groups of patients
receiving hypnotherapy with a waiting control group. The analysis
indicates medium efficacy of hypnosis by a weighted mean effect size
of d
+
= .61 for psychosomatic disorders. However, the mixed and the
modern forms of hypnosis showed superiority over the classical
approach (d
+
= .33, .73, 1.42, for the classical, mixed, and modern
forms, respectively). This finding is similar to the medium effect size
obtained by Flammer and Bongartz (2003) in their meta-analytic study
of the efficacy of hypnosis with various psychological disorders.
Regarding the effect size obtained in this study, it should be men-
tioned that all the dependent variables from each study were included
in the computation of the effect size. If only some selected variables of
a study are included in the computation of the effect size, the effect
size will vary a lot, depending on which variables are used in the com-
putation (e.g., Shapiro & Shapiro, 1982). Moreover, most of the mea-
sures used in the studies included in the analysis were heterogenous
and of unknown reliability and validity. When standardized and uni-
form measures are used in a meta-analysis, there is a high level of
homogeneity. Out of 81 variables used in the studies, only 15 variables
were measured with validated instruments with known reliability.
Using nonvalidated instruments with unknown reliability is likely to
cause a reduction of the total effect size (J. Cohen, 1988; Hunter &
Schmidt, 1990). It is well known that different kinds of measurement
result in different outcomes (e.g., Shadish & Ragsdale, 1996; Shapiro &
Table 5
Results of Regression Analysis for the 21 Studies
Term coefficient SE tp
Intercept .135 .3142 .43 .6728
Duration .0759 .0427 1.78 .093
Downloaded by [184.64.123.170] at 17:26 11 May 2015
266 ERICH FLAMMER AND ASSEN ALLADIN
Shapiro, 1982)—self-ratings and behavioral counts usually produce
relatively favorable outcomes, whereas physiological and psychomet-
ric measures usually produce relatively unfavorable outcomes. The
estimates of the effect sizes reported in this meta-analysis must, there-
fore, be regarded as very conservative. To avoid computation bias and
subjective selection of variables, the present study included all the
variables from each study to calculate the effect size. But, despite the
overinclusiveness of the variables, a medium effect for the efficacy of
hypnotherapy with psychosomatic disorders was found.
As suggested by the funnel plot of effect sizes versus sample size,
the large treatment effects seen in large studies were missing. This may
be due to publication bias but may also be a consequence of the
restricted strategy used for selecting studies. While the search for the
studies was not limited to certain languages, only studies in English,
German, and French were actually eligible, because of the authors’ lim-
ited skills in foreign languages and limited capacities for translation.
With respect to language bias in medical trials, Egger and colleagues
(1997) have argued that negative findings are more often published in
non-English journals, which contributes to retrieval bias. If this kind of
publication bias exists in psychotherapy research, then the limitation
of languages is likely to lead to an underrepresentation of studies with
nonsignificant or negative results.
Surprisingly, the number of treatment sessions did not influence
treatment outcome in any substantial manner. A possible explanation
is that a linear regression model was fitted rather than a dose-response
relationship (Chatterjee & Price, 1991). But a plot of duration versus
effect size showed no obvious pattern of nonlinear relationship and the
models with logarithmic transformation or quadratic terms also failed
to show significance and fitted the data even worse than a simple lin-
ear regression model. Another explanation for the unexpected lack of
association between therapy dose and outcomes may be that a dose-
response relationship might have been strong for some measures or
some disorders but not for others.
With respect to an association between suggestibility and treatment
outcome, only seven studies provided information about a correlation.
Three studies found a positive correlation, three studies reported a sig-
nificant positive correlation, and one study reported no correlation.
Since the numerical values for the correlation coefficient were pro-
vided in only three studies, the mean correlation must be interpreted
with caution. However, when taken together, the data hint at a posi-
tive correlation between suggestibility and treatment success.
As reported above, classical hypnosis was predominantly used in the
studies of this analysis. While 53.6% of the studies were categorized classi-
cal hypnosis, only 14.3% were assigned to modern hypnosis. Thus, the
results of this analysis essentially refer to the practice of classical hypnosis.
Downloaded by [184.64.123.170] at 17:26 11 May 2015
HYPNOTHERAPY AND PSYCHOSOMATIC DISORDERS 267
The 22 studies selected for this meta-analysis invariably used
symptom-oriented suggestions (90.1%) and relaxation (72.73%) in the
hypnotic treatment of the psychosomatic disorders. But psychosomatic
illnesses are not simply caused by maladaptive behaviors; they can
also be caused by pathogenic emotions such as repressed aggression,
pathogenic grief, anxiety, etc. For example, Pennebaker and Watson
(1991) conceptualized psychosomatic symptoms in terms of negative
affectivity, a variant of negative self-concept. K. Cohen, Auld, and
Brooker (1994) have provided evidence for the relationship between
alexithymia and psychosomatic disorders. This finding has been
recently supported by Waller and Scheidt (2006), who found evidence
for a link between somatoform disorders and a diminished capacity to
experience, to differentiate, and to express affects. Moreover, Rodin,
deGroot, and Spivak (1998) have examined the role of dissociation in
somatization disorders. Craig, Boardman, Mills, Daly-Jones, and
Drake (1993) found lack of parental care and severe illness in child-
hood as the best predictors of somatization in adulthood. With respect
to defense styles, Nickel and Egle (2006) provided evidence for an
association between immature defense styles and somatization. Over
and above those findings, a multitude of studies reveal high comorbidity
between psychosomatic disorders and other psychological disorders.
Bridges and Goldberg (1985) point out that patients in primary care,
especially those with depressive or anxiety disorders, usually present
their symptoms via somatization. After reviewing the relevant literature,
Kirmayer and Young (1998) concluded that somatization is a common
style for expressing distress. In patients with unexplained somatic
complaints, Brown, Golding, and Smith (1990) report lifetime prevalence
rates of 54.6% for major depression and 33.6% for generalized anxiety
disorder. In an extensive meta-analysis of 244 studies, Henningssen,
Zimmermann, and Sattel (2003) found moderate but statistically
highly significant associations between IBS, nonulcer dyspepsia, fibro-
myalgia, and chronic fatigue syndrome with depression and anxiety.
As the examination of the clinical trials subjected to this meta-analysis
reveals, it is clearly evident that hypnosis is essentially used for
symptomatic treatment. It would appear the regular hypnotherapy pro-
tocol for psychosomatic illnesses does not include strategies for dealing
with maintaining factors. Similarly, the treatment protocol does not
address cognitions. While 95.5% (n = 21) of the studies addressed the
symptoms, only 9.5% (n = 2) of the studies focused on maladaptive cog-
nitions or irrational ideas. The same applies to addressing emotions and
to gaining insight into the underlying causes of the somatic symptoms.
Only two studies (9.5%) included these aspects in their treatment plan.
So the interventions employed in the clinical studies included in this
analysis might be well suited to alleviate the symptoms, but they might
not have been effective in preventing relapses, because they did not
Downloaded by [184.64.123.170] at 17:26 11 May 2015
268 ERICH FLAMMER AND ASSEN ALLADIN
address triggering or maintaining factors. If this is true, the treatment
outcome would be expected to be less stable over time compared to
those interventions that focused on triggering and maintaining factors.
This study has several limitations. First of all, the definition of psycho-
somatic disorders is somewhat arbitrary, as there is no such distinct cate-
gory in either the ICD-10 or the Diagnostic and Statistical Manual of Mental
Disorders, fourth edition (DSM-IV) classification systems. In a broad
sense, the concept of psychosomatic illness may include somatoform dis-
orders as well as disorders with psychological factors assumed to play a
major role in the etiology, triggering, and maintenance of somatic com-
plaints. Applying the ICD-10 F54 definition could justify inclusion of
studies on diabetes, infectious diseases, or even cancer. Clearly, the defi-
nition of psychosomatic may not only substantially affect the interpretation
of meta-analytic results but also the magnitude of mean effect sizes itself.
A restrictive definition of psychosomatic disorders would have been to
include in the analysis only studies that treated somatoform disorders
meeting ICD-10 criteria. Such a limitation would facilitate replication and
create a more homogeneous sample, thus allowing a more straightfor-
ward interpretation of the results. In this study, a broader definition of
psychosomatic disorders was chosen in order to obtain a reasonable sam-
ple size from randomized, controlled trials.
Although a wide variety of conditions are considered psychoso-
matic disorders, only a limited range of psychosomatic problems were
included in this evaluation. This limitation was imposed by the avail-
ability of published randomized, controlled trials. Nonrandomized
trials and studies consisting of single-group pretest-posttest designs
were excluded in order to ensure greater relevance to the meta-analysis
and to counteract positive distortion of the evaluation of treatment effi-
cacy by inflated effect size estimates. But relying solely on data from
randomized trials may not be totally justified if the selected studies
provided insufficient information on highly relevant aspects of treat-
ment and outcome, such as severity of symptoms, qualification of the
therapist, or long-term follow-up data.
Another critical point of this study is that publication bias might have
existed. In fact, the fail-safe N of the 56 studies (more than twice the
number of studies included in the analysis) seems to support the valid-
ity of this bias. But a look at the heavily skewed funnel plot of the effect
sizes as a function of sample size shows that this may not be taken for
granted. Instead, the validity of the results should be judged with some
caution, unless further meta-analytic evaluations are conducted.
An additional shortcoming of the present analysis is its lack of a
closer investigation of factors crucial for the treatment outcome. Even
small differences in diagnostic criteria, age, and severity of symptoms
can lead to the formation of very different groups of patients, inevita-
bly resulting in differential treatment effect. Since the studies selected
Downloaded by [184.64.123.170] at 17:26 11 May 2015
HYPNOTHERAPY AND PSYCHOSOMATIC DISORDERS 269
for the analysis did not provide sufficient information, the effects of
these factors were not subjected to examination. Moreover, the effect of
therapist variables, e.g., qualification and experience, on treatment
could not be investigated due to a lack of information. In future inves-
tigations, such moderator factors as accurate diagnosis, measurements,
and therapist variables should be taken into account.
Finally, the analysis did not break down the studies into homoge-
neous subgroups with respect to dose-response relationship. The find-
ing that there was no significant relationship between treatment dose
and treatment outcome appears anomalous. It is possible that the rela-
tionship was compromised by the presence of some underlying, unde-
tected moderator variables.
CONCLUSIONS
The present meta-analysis indicates hypnosis is an effective adjunct in
the treatment of psychosomatic disorders. However, the three types of
hypnotherapy protocol (classical, modern, and mixed) utilized in the stud-
ies provided differential treatment outcome. Modern hypnotherapy
appears to be comparatively superior to classical hypnotherapy. However,
the review of the studies included in the analysis indicated that hypnother-
apy could be made more effective by including components addressing
emotional and underlying factors rather than just focusing on symptoms.
REFERENCES
References marked by an asterisk indicate studies included in the meta-analysis.
Anbar, R. D. (2001). Self-hypnosis for management of chronic dyspnea in pediatric
patients. Pediatrics, 107, E21.
*Attias, J., Gan, I., Shemesh, Z., Shoham, C., Shahar, A., & Sohmer, H. (1990). Efficacy of
self-hypnosis for tinnitus relief. Scandinavian Audiology, 19, 245–249.
Banerjee, S., Srivastav, A., & Palan, B. M. (1993). Hypnosis and self-hypnosis in the
management of nocturnal enuresis: A comparative study with imipramine therapy.
American Journal of Clinical Hypnosis, 36, 113–119.
Benson, K., & Hartz, A. J. (2000). A comparison of observational studies and random-
ized, controlled trials. New England Journal of Medicine, 342, 1878–1886.
Black, S. (1963a). Inhibition of immediate-type hypersensitivity response by direct
suggestion under hypnosis. British Medical Journal, 6, 625–629.
Black, S. (1963b). Shift in dose-response curve of Prausnitz-Küstner reaction by direct
suggestion under hypnosis. British Medical Journal, 13, 990–992.
*Borkovec, T. D., & Fowles, D. C. (1973). Controlled investigation of the effects of progres-
sive and hypnotic relaxation on insomnia. Journal of Abnormal Psychology, 82, 153–158.
Bridges, K. W., & Goldberg, D. P. (1985). Somatic presentation of DSM-III psychiatric
disorders in primary care. Journal of Psychosomatic Research, 29, 563–569.
Brown, F. W., Golding, J. M., & Smith, G. R. (1990). Psychiatric comorbidity in primary
care somatization disorder. Psychosomatic Medicine, 52, 445–451.
Downloaded by [184.64.123.170] at 17:26 11 May 2015
270 ERICH FLAMMER AND ASSEN ALLADIN
*Calvert, E. L., Houghton, L. A., Cooper, P., Morris, J., & Whorwell, P. J. (2002). Long-
term improvement in functional dyspepsia using hypnotherapy. Gastroenterology,
126, 1778–1785.
Cedercreutz, C., Lahteenmaki, R., & Tulikoura, J. (1976). Hypnotic treatment of headache
and vertigo in skull injured patients. International Journal of Clinical & Experimental
Hypnosis, 24, 195–200.
Chatterjee, S., & Price, B. (1991). Regression analysis by example (2nd ed.). New York:
Wiley.
Citron, K. M. (1968). Hypnosis for asthma – A controlled trial. British Medical Journal, 4, 71–76.
Clarke, J. H., & Reynolds, P. J. (1991). Suggestive hypnotherapy for nocturnal bruxism: A
pilot study. American Journal of Clinical Hypnosis, 33, 248–253.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences. Hillsdale, NJ: Erlbaum.
Cohen, K., Auld, F., & Brooker, H. (1994). Is alexithymia related to psychosomatic disor-
der and somatization? Journal of Psychosomatic Research, 38, 119–127.
*Colgan, S. M., Faragher, E. B., & Whorwell, P. J. (1988). Controlled trial of hypnotherapy
in relapse prevention of duodenal ulceration. Lancet, 1(8598), 1299–1300.
Concato, J., Shah, N., & Horwitz, R. J. (2000). Randomized, controlled trials, observa-
tional studies, and the hierarchy of research designs. New England Journal of Medicine,
342, 1887–1892.
Craig, T. K., Boardman, A. P., Mills, K., Daly-Jones, O., & Drake, H. (1993). The South
London somatization study I: Longitudinal course and influence of early life experi-
ences. British Journal of Psychiatry, 163, 570–588.
Domangue, B. B., Margolis, C. G., Lieberman, D., & Kaji, H. (1985). Biochemical correlates of
hypnoanalgesia in arthritic pain patients. Journal of Clinical Psychiatry, 46, 235–238.
*Edwards, S. D., & van der Spuy, H. I. (1985). Hypnotherapy as a treatment for enuresis.
Journal of Child Psychology and Psychiatry and Allied Disciplines, 26, 161–170.
Egger, M., Zellweger-Zähner, T., Schneider, M., Junker, C., Lengeler, C., & Antes, G.
(1997). Language bias in randomised controlled trials published in English and
German. Lancet, 350, 326–329.
Elton, D. (1993). Combined use of hypnosis and EMG biofeedback in the treatment of
stress-induced conditions. Stress Medicine, 9, 25–35.
*Ewer, T. C., & Stewart, D. E. (1986). Improvement in bronchial hyper-responsiveness in
patients with moderate asthma after treatment with a hypnotic technique: A random-
ized controlled trial. British Medical Journal, 293(6555), 1129–1132.
Flammer, E., & Bongartz, W. (2003). On the efficacy of hypnosis: A meta-analytic study.
Contemporary Hypnosis, 20, 179–197.
Fry, L., Mason, A. A., & Pearson, B. (1964). Effect of hypnosis on allergic skin responses
in asthma and hay-fever. British Medical Journal, 1, 1145–1148.
*Galovski, T. E., & Blanchard, E. B. (1998). The treatment of irritable bowel syndrome
with hypnotherapy. Applied Psychophysiology and Biofeedback, 23, 219–232.
*Gay, M.-C., Philippot, P., & Luminet, O. (2002). Differential effectiveness of psychologi-
cal interventions for reducing osteoarthritis pain: A comparison of Erickson hypnosis
and Jacobson relaxation. European Journal of Pain, 6
, 1–16.
Gleser, L. J., & Olkin, I. (1994). Stochastically dependent effect sizes. In H. Cooper & L. V.
Hedges (Eds.), Handbook of research synthesis. New York: Russell Sage Foundation.
Greenhouse, J. B., & Iyengar, S. (1994). Sensitivity analysis and diagnostics. In H. Cooper &
L. V. Hedges (Eds.), Handbook of research synthesis. New York: Russell Sage Foundation.
Gruzelier, J., Smith, F., Nagy, A., & Henderson, D. (2001). Cellular and humoral immu-
nity, mood and exam stress: The influences of self-hypnosis and personality predic-
tors. International Journal of Psychophysiology, 42, 55–71.
Hedges, L. V., & Olkin, I. (1985). Statistical methods for meta-analysis. Orlando: Academic Press.
Hemingway, H., Malik, M., & Marmot, M. (2001). Social and psychosocial influences on
sudden cardiac death, and ventricular arrhythmia and cardiac autonomic function.
European Heart Journal, 22, 1082–1101.
Downloaded by [184.64.123.170] at 17:26 11 May 2015
HYPNOTHERAPY AND PSYCHOSOMATIC DISORDERS 271
Henningsen, P., Zimmermann, T., & Sattel, H. (2003). Medically unexplained physical
symptoms, anxiety, and depression: A meta-analytic review. Psychosomatic Medicine,
65, 528–533.
*Hill, R. W. (1981). Hypnosis: A group treatment for smoking, obesity and the perception of
stress. Unpublished doctoral dissertation. Virginia Commonwealth University,
Richmond, Virginia.
*Hoppe, F. (1984). Hypnotische Schmerzlinderung durch therapeutische Anekdoten:
Eine Untersuchung zur Verarbeitung von Mikro- und Makro-Suggestionen bei chro-
nischen Schmerzpatienten [Hypnotic pain reduction through therapeutic anecdotes:
A study of processing micro- and macro-suggestions in chronic pain patients].
Zeitschrift fuer Klinische Psychologie. Forschung und Praxis, 13, 300–321.
Houghton, L. A., Calvert, E. L., Jackson, N. A., Cooper, P., & Whorwell, P. J. (2002).
Visceral sensation and emotion: A study using hypnosis. Gut, 51, 701–704.
Houghton, L. A., Heyman, D. J., & Whorwell, P. J. (1996). Symptomatology, quality of
life and economic features of irritable bowel syndrome – The effect of hypnotherapy.
Alimentary Pharmacology and Therapeutics, 10, 91–95.
Hunter, J., & Schmidt, F. L. (1990). Methods of meta-analysis. Newbury Park/Beverly Hills,
CA: Sage.
Jambrik, Z., Sebastiani, L., Picano, E., Ghewlarducci, B., & Santarcangelo, E. L. (2005).
Hypnotic modulation of flow-mediated endothelial response to mental stress. Inter-
national Journal of Psychophysiology, 55, 221–227.
Kiecolt-Glaser, J. K., Marucha, P. T., Atkinson, C., & Glaser, R. (2001). Hypnosis as a
modulator of cellular immune dysregulation during acute stress. Journal of Consulting
& Clinical Psychology, 69, 674–682.
Kirmayer, L. J., & Young, A. (1998). Culture and somatization: Clinical, epidemiological,
and ethnographic perspectives. Psychosomatic Medicine, 60, 389–393.
Kirsch, I., Montgomery, G., & Sapirstein, G. (1995). Hypnosis as an adjunct to cognitive-
behavioral psychotherapy: A meta-analysis. Journal of Consulting and Clinical Psychol-
ogy, 63, 214–220.
*Kohen, D. P. (1995). Relaxation/mental imagery (self-hypnosis) for childhood asthma:
Behavioural outcomes in a prospective, controlled study. Australian Journal of Clinical
& Experimental Hypnosis, 24, 12–28.
Laidlaw, T. M., Richardson, D. H., Booth, R. J., & Large, R. G. (1994). Immediate-type
hypersensitivity reactions and hypnosis: Problems in methodology. Journal of Psycho-
somatic Research, 38, 569–580.
*Langewitz, W., Izakovic, J., Wyler, J., Schindler, C., Kiss, A., & Bircher, A. J. (2005).
Effect of self-hypnosis on hay fever symptoms – A randomized controlled interven-
tion study. Psychotherapy and Psychosomatics, 74, 165–172.
Light, R. J., & Pillemer, D. B. (1984). Summing up: The science of reviewing research.
Cambridge, MA: Harvard University Press.
*Llaneza-Ramos, M. L. (1989). Hypnotherapy in the treatment of chronic headaches.
Philippine Journal of Psychology, 22, 17–25.
*Maher-Loughnan, G. P., MacDonald, N., Mason, A. A., & Fry, L. (1962). Controlled trial of
hypnosis in the symptomatic treatment of asthma. British Medical Journal, 2, 371–376.
Matt, G. E., & Navarro, A. M. (1997). What meta-analyses have and have not taught us
about psychotherapy effects: A review and future directions. Clinical Psychology
Review, 17
, 1–32.
*Melis, P. M. L., Rooimans, W., Spierings, E. L. H., & Hoogduin, C. A. (1991). Treatment
of chronic tension-type headache with hypnotherapy: A single-blind time controlled
study. Headache, 31, 686–689.
Miller, R. G. (1986). Beyond ANOVA, basics of applied statistics. New York: Wiley.
*Moene, F. C., Spinhoven, Ph., Hoogduin, C. A. L., & van Dyck, R. (2003). A randomized
controlled clinical trial of a hypnosis-based treatment for patients with conversion dis-
order, motor type. International Journal of Clinical and Experimental Hypnosis, 51, 29–50.
Downloaded by [184.64.123.170] at 17:26 11 May 2015
272 ERICH FLAMMER AND ASSEN ALLADIN
Mullen, B. (1989). Advanced BASIC meta-analysis. Hillsdale, NJ: Erlbaum.
Nickel, R., & Egle, U. T. (2006). Psychological defense styles, childhood adversities and
psychopathology in adulthood. Child Abuse and Neglect, 30, 157–170.
Orwin, R. G. (1983). A fail-safe N for effect size in meta-analysis. Journal of Educational
Statistics, 8, 157–159.
*Pearson, T. A. (1994). Determining pain management treatment modalities and effectiveness.
Unpublished doctoral dissertation. Spalding University, Louisville, Kentucky.
Pennebaker, J. W., & Watson, D. (1991). The psychology of somatic symptoms. In L. J.
Kirmayer & J. M. Robbins (Eds.), Current concepts of somatization: Research and clinical
perspectives (pp. 21–35). Washington, DC: American Psychiatric Press.
Pinnell, C. M., & Covino, N. A. (2000). Empirical findings on the use of hypnosis in medicine:
A critical review. International Journal of Clinical and Experimental Hypnosis, 48, 170–194.
Prior, A., Colgan, S. M., & Whorwell, P. J. (1990). Changes in rectal sensitivity after hyp-
notherapy in patients with irritable bowel syndrome. Gut, 31, 896–898.
*Raskin, R., Raps, C., Luskin, F., Carlson, R., & Cristal, R. (1999). Pilot study of the effect
of self-hypnosis on the medical management of essential hypertension. Stress Medi-
cine, 15, 243–247.
Rhue, J. W., Lynn, S. J., & Kirsch, I. (1993). Handbook of clinical hypnosis. Washington, DC:
American Psychological Association.
Rodin, G., deGroot, J., & Spivak, H. (1998). Trauma, dissociation, and somatization. In
J. D. Bremner & C. R. Marmar (Eds.), Trauma, memory, and dissociation (pp. 117–223).
Washington, DC: American Psychiatric Press.
Rosenthal, R. (1984). Meta-analytic procedures for social research. Beverly Hills, CA: Sage.
Rosenthal, R., & Rubin, D. B. (1982). A simple, general-purpose display of magnitude of
experimental effect. Journal of Educational Psychology, 74, 166–169.
Rucklidge, J. J., & Saunders, D. (2002). The efficacy of hypnosis in the treatment of pruri-
tis in people with HIV/AIDS. International Journal of Clinical and Experimental Hypno-
sis, 50, 149–169.
*Senser, C., Habermüller, M., & Revenstorf, D. (2004). Hypnotherapie bei atopischer
Dermatitis [Hypnotherapy in atopic dermatitis]. Akt Dermatol, 30, 103–108.
Shadish, W. R., & Ragsdale, K. (1996). Random versus nonrandom assignment in con-
trolled experiments: Do we get the same answer? Journal of Consulting and Clinical
Psychology, 64, 1290–1305.
Shapiro, D. A., & Shapiro, D. (1982). Meta-analysis of comparative therapy outcome
studies: a replication and refinement. Psychological Bulletin, 92, 581–604.
Simon, E. P., & Lewis, D. M. (2000). Medical hypnosis for temporomandibular disorders:
Treatment efficacy and medical utilization outcome. Oral Surgery, Oral Medicine, Oral
Pathology, Oral Radiology, and Endodontics, 90, 54–63.
*Spanos, N. P., Liddy, S. J., Scott, H., Garrard, C., Sine, J., Tirabasso, A., et al. (1993). Hyp-
notic suggestion and placebo for the treatment of chronic headache in a university
volunteer sample. Cognitive Therapy and Research, 17, 191–205.
Taggart, P., Sutton, P., Redfern, C., Batchvarov, V. N., Hnatkova, K., Malik, M., James, U.,
& Joseph, A. (2005). The effect of mental stress on the non-dipolar components of the
T wave: Modulation by hypnosis. Psychosomatic Medicine, 67, 376–383.
*Ter Kuile, M. M., Spinhoven, P., Linssen, A. C., Zitman, F. G., van Dyck, R., & Rooijmans,
H. G. M. (1994). Autogenic training and cognitive self-hypnosis for the treatment of
recurrent headaches in three different subject groups. Pain, 58, 331–340.
*Tosi, D. J., Judah, S. M., & Murphy, M. A. (1989). The effects of a cognitive experien-
tial therapy utilizing hypnosis, cognitive restructuring, and developmental stag-
ing on psychological factors associated with duodenal ulcer disease: A multivariate
experimental study. Journal of Cognitive Psychotherapy: An International Quarterly,
3, 273–290.
Tschugguel, W. B., & Berga, S. L. (2003). Treatment of functional hypothalamic amenor-
rhea with hypnotherapy. Fertility and Sterility, 80, 982–985.
Downloaded by [184.64.123.170] at 17:26 11 May 2015
HYPNOTHERAPY AND PSYCHOSOMATIC DISORDERS 273
Wadden, T. A., & Anderton, C. H. (1982). The clinical use of hypnosis. Psychological Bulle-
tin, 91, 215–243.
Waller, E., & Scheidt, C. E. (2006). Somatoform disorders as disorders of affect regula-
tion: A development perspective. International Review of Psychiatry, 18, 13–24.
Williamson, J. W., McColl, R., Mathews, D., Mitchell, J. H., Raven, P. B., & Morgan, W. P.
(2001). Hypnotic manipulation of effort sense during dynamic exercise: Cardiovascu-
lar responses and brain activation. Journal of Applied Physiology, 90, 1392–1399.
Younus, J., Simpson, I., Collins, A., & Wang, X. (2003). Mind control of menopause.
Women’s Health Issues, 13, 74–78.
Zachariae, R., Bjerring, P., & Arendt-Nielsen, L. (1989). Modulation of type I immediate
and type IV delayed immunoreactivity using direct suggestions and guided imagery
during hypnosis. Allergy, 44, 537–542.
Zachariae, R., Jorgensen, M. M., Egekvist, H., & Bjerring, P. (2001). Skin reactions to his-
tamine of healthy subjects after hypnotically induced emotions of sadness, anger,
and happiness. Allergy, 56, 734–740.
Die Effektivität von Hypnotherapie bei der Behandlung von
psychosomatischen Störungen: Metaanalytische Evidenz
Erich Flammer und Assen Alladin
Zusammenfassung: Es wird behauptet, dass Hypnotherapie effektiv bei der
Behandlung psychosomatischer Störungen sei. Eine Metaanalyse über 21
randomisierte kontrollierte klinische Studien wurde durchgeführt, um die
Effektivität von Hypnose bei psychosomatischen Störungen zu evaluieren.
Die Studien verglichen Patienten, welche ausschließlich mit Hypnotherapie
behandelt wurden, mit unbehandelten Kontrollpersonen. Studien, welche
adjunktive medizinische Standardversorgung beinhalteten, wurden ebenfalls
in die Untersuchen einbezogen. Die Art der Hypnotherapie wurde dabei
klassifiziert als klassisch (n = 9), gemischt (n = 5) oder modern (n = 3). Die
Ergebnisse dieser 21 Studien zeigten eine gewichtete mittlere Effektstärke
d
+
= .61 (p = .0000). Eine Varianzanalyse erbrachte signifikante Unterschiede
zwischen klassischer, gemischter und modernen Hypnose. Eine Regression
des Ergebnisses auf die Behandlungsdosis ergab keinen signifikanten
Zusammenhang. Die numerischen Werte für die Korrelation zwischen
Suggestibilität und Ergebnis wurden lediglich in 3 Studien berichtet
(Mittelwert r = .31). Die Metaanalyse belegt klar, dass Hypnosetherapie bei
der Behandlung psychosomatischer Störungen effektiv ist.
R
ALF SCHMAELZLE
University of Konstanz, Konstanz, Germany
L’efficacité de l’hypnothérapie dans le traitement des troubles
psychosomatiques : données probantes méta-analytiques
Erich Flammer et Assen Alladin
Résumé: L’hypnothérapie agirait efficacement dans le traitement des
troubles psychosomatiques.
Une méta-analyse a été menée sur 21 études cliniques comparatives, à
répartition aléatoire, afin d’évaluer l'efficacité de l'hypnose dans le
traitement des troubles psychosomatiques. Ces études visaient à comparer
Downloaded by [184.64.123.170] at 17:26 11 May 2015
274 ERICH FLAMMER AND ASSEN ALLADIN
des patients traités exclusivement à l’aide d’hypnothépie à des cas témoins,
non traités. Des études incorporant des traitements d’appoint standard, dans
l’un ou l’autre des bras de traitement, ont également été admises.
L’hypnothérapie a été divisée en trois groupes : classique (n = 9), mixte (n =
5) et moderne (n = 3). Les résultats ont montré que la valeur de l’effet moyen
pondéré de ces 21 études était de d
+
= 0,61 (p = 0,0000). Une analyse de
variance a révélé des différences considérables entre l’hypnose classique,
mixte et moderne. Toutefois, on n’a pu établir un lien significatif entre la
régression des résultats et la dose de traitement. Des valeurs numériques
révélant une corrélation entre la suggestibilité et les résultats n’ont été
rapportées que dans trois études (moyenne r = 0,31). Cette méta-analyse
indique clairement la grande efficacité de l’hypnothérapie dans le
traitement des troubles psychosomatiques.
J
OHANNE REYNAULT
C. Tr. (STIBC)
La eficacia de la hipnoterapia en el tratamiento de trastornos psicosomáticos:
Evidencia meta-analítica
Erich Flammer y Assen Alladin
Resumen: Concluyo que la hipnoterapia es eficaz en el tratamiento de
trastornos psicosomáticos. Realicé un meta-análisis con 21 estudios clínicos
aleatorios controlado para evaluar eficacia de la hipnosis en los trastornos
psicosomáticos. Los estudios compararon pacientes exclusivamente tratados
con hipnoterapia y un grupo control. También incluyo estudios que
emplearon la atención médica convencional como un co-adyuvante.
Categoricé a la hinoterapia como clásica (n = 9), mixta (n = 5), o moderna (n = 3).
Los resultados mostraron un tamaño de efecto promedio balanceado de los
21 estudios de d = .61 (p = .0000). Una ANOVA arrojó diferencias significativas
entre la hipnosis moderna, mixta, y clásica. Una regresión de resultado y
dosis de tratamiento no mostó resultados significativos. Sólo en 3 estudios
se dieron valores numéricos sobre la correlación entre la sugestionabilidad y
los resultado (media de r = .31). El meta-análisis claramente indica que la
hipnoterapia es altamente efectiva en el tratamiento de los trastornos
psicosomáticos.
E
TZEL CARDEÑA
University of Lund, Lund, Sweden
Downloaded by [184.64.123.170] at 17:26 11 May 2015
