Yoga for anxiety: A systematic review of the research evidence

Abstract

Between March and June 2004, a systematic review was carried out of the research evidence on the effectiveness of yoga for the treatment of anxiety and anxiety disorders. Eight studies were reviewed. They reported positive results, although there were many methodological inadequacies. Owing to the diversity of conditions treated and poor quality of most of the studies, it is not possible to say that yoga is effective in treating anxiety or anxiety disorders in general. However, there are encouraging results, particularly with obsessive compulsive disorder. Further well conducted research is necessary which may be most productive if focused on specific anxiety disorders.

Full text

REVIEW
Yoga for anxiety: a systematic review of the research
evidence
G Kirkwood, H Rampes, V Tuffrey, J Richardson, K Pilkington
...............................................................................................................................
Br J Sports Med 2005;39:884–891. doi: 10.1136/bjsm.2005.018069
Between March and June 2004, a systematic review was
carried out of the research evidence on the effectiveness of
yoga for the treatment of anxiety and anxiety disorders.
Eight studies were reviewed. They reported positive results,
although there were many methodological inadequacies.
Owing to the diversity of conditions treated and poor
quality of most of the studies, it is not possible to say that
yoga is effective in treating anxiety or anxiety disorders in
general. However, there are encouraging results,
particularly with obsessive compulsive disorder. Further
well conducted research is necessary which may be most
productive if focused on specific anxiety disorders.
...........................................................................
See end of article for
authors’ affiliations
.......................
Correspondence to:
K Pilkington, Research
Council for
Complementary Medicine
(RCCM)/University of
Westminster, c/o School
of Integrated Health,
University of Westminster,
115 New Cavendish
Street, London W1W
6UW, UK; K.Pilkington@
westminster.ac.uk
Accepted 20 June 2005
.......................
A
nxiety disorders are among the most
prevalent mental health problems found
in the community in the United Kingdom
according to the survey carried out by the Office
for National Statistics (ONS) in the year 2000.
1
Conditions such as mixed anxiety and depressive
disorder, generalised anxiety disorder, phobias,
obsessive compulsive disorder, and panic dis-
order make up over 86% of neurotic disorders
found. Excessive anxiety is a key component or
symptom in all of these conditions.
Yoga is defined as a practice consisting of three
components: gentle stretching; exercises for
breath control; and meditation as a mind-body
intervention.
2
The version used mainly in the
West is hatha yoga, which consists of an
integration of asana (postures), pranayama
(breathing exercise), and meditation.
3
Although
yoga has its origins in Indian culture and
religion, it can be practised secularly.
2
No
systematic reviews have been published on the
benefits of yoga in anxiety or anxiety disorders.
This is despite the fact that a recent analysis of
publication trends has shown an increase in
publication frequency and growing use of ran-
domised controlled trials to study yoga as a
therapeutic intervention.
4
The only systematic
review that looks specifically at yoga as an
intervention for any condition is that on yoga
in epilepsy, which was inconclusive because of a
lack of studies.
5
The effect of exercise on anxiety has, however,
been reviewed. There is some evidence of an
anxiolytic effect,
6
with aerobic exercise possibly
more beneficial than non-aerobic exercise.
78
There is also some evidence that exercise may
be particularly beneficial in people with more
severe anxiety.
79
None of these reviews, how-
ever, appear to have included yoga as a form of
exercise.
There are a number of studies that look at the
effects of yoga on anxiety levels in non-clinical
samples. Berger and Owen
10
compared the effects
of swimming, fencing, body conditioning, and
yoga classes and found that only the yoga
treatment group recorded a significant short
term reduction in state anxiety. Ray et al
11
reported that yoga reduced anxiety but only
among male students. Netz and Lidor
12
showed
that participants in yoga as well as swimming
and the Feldenkrais method recorded lower
anxiety levels than a control group. However, in
a study of elderly people, Blumenthal et al
13 14
found that yoga participants fared worse than
those in an aerobic exercise group and no better
than the other treatment regimens on anxiety
measures. It is difficult to predict on the basis of
the findings of these studies the effect of yoga on
people with anxiety or a specific anxiety disorder,
and therefore it is important to identify the
evidence that is currently available.
METHODS
Aim and objectives
The aim of this study was to evaluate the
evidence from a range of sources of the effec-
tiveness of yoga for the treatment of anxiety and
anxiety disorders.
Summary of the search strategy
A comprehensive search for clinical research was
carried out. Searches were conducted on major
biomedical and specialist databases and web-
sites. Citations were sought from relevant
reviews and various appropriate specialised
books. Relevant websites were also included in
the search, including those of specialist yoga and
mental health organisations.
Databases
The following databases were searched between
March and June 2004:
N
General databases: ClNAHL, Cochrane Central
Register of Controlled Trials (CENTRAL),
Cochrane Database of Systematic Reviews,
Database of Abstracts of Reviews of Effects,
EMBASE, Medline (and PubMed), PsycINFO
N
Specialist CAM databases: AMED, CISCOM
N
Specialist yoga websites: International Asso-
ciation of Yoga Therapists,
15
Yoga Biomedical
Trust
16
N
Specialist mental health websit es: MIND,
17
Mental Health Foundation
18
884
www.bjsportmed.com

A search of the Cochrane Collaboration Depression,
Anxiety and Neurosis (CCDAN) Controlled Trials Register
was also conducted in December 2004.
Search terms
The following terms for yoga and anxiety were used in the
search of AMED, CINAHL, EMBASE, Medline and PsycINFO
(OVID): (exp yoga OR yoga.mp OR yogic.mp OR asanas.mp
OR dhyana.mp OR pranayama.mp OR meditation OR
meditat*) AND (exp anxiety OR exp anxiety disorders OR
anx$.mp).
Search strategies were adapted for each of the databases
searched. The searches of CISCOM and CCDAN were carried
out by the information specialists responsible for these
databases. Efforts were made to identify unpublished and
ongoing research using relevant databases such as the
National Research Register (UK) and Clinicaltrials.gov (US).
Filtering
This was carried out by two independent researchers, and
relevant research was categorised by study type according to
a flow chart system developed for this project.
The basic categories used were: randomised controlled
trials; controlled clinical trials without randomisation;
uncontrolled clinical trials and case series; case reports/
studies; qualitative research; surveys; other research studies.
Selection criteria
Types of study
All clinical trials, controlled and uncontrolled, were identi-
fied, but only controlled trials, whether randomised or not,
were included in this review. Dissertation abstracts were
excluded because of lack of detail of methodology and
outcome measures. Attempts were also made to locate
relevant qualitative studies.
No language restrictions were imposed at the search and
filtering stage.
Types of participants
All studies were included in which participants were
described as suffering from anxiety. This primarily took the
form of a diagnosis of an anxiety disorder by whatever
diagnostic method was deemed appropriate by the papers’
authors. In addition, studies in which participants were
suffering from anxiety as determined by the authors via
measurement scales and those in which participants were
about to undergo a procedure that was anticipated to be
anxiety provoking were also included.
Types of intervention
Yoga of various styles. Studies where the intervention was
solely meditation based were not included.
Types of outcome measures
Anxiety rating scales and scales to measure specific symp-
toms of anxiety disorders.
Data collection and appraisal
Data was extracted systematically using a specially designed
data extraction form. Data extracted included details of
selection criteria and procedure, the participants, the inter-
vention and any comparison or control intervention, aspects
of the methodology, and outcome measures and results.
Clinical trials were appraised using a standardised apprai-
sal framework specifically developed for this project and
based on criteria recommended by the Centre for Reviews
and Dissemination.
19
Study quality was measured by recording details of method
of randomisation, concealment of allocation, blinding of
assessors, and loss to follow up. Also recorded were statistical
power, presence or absence of intention to treat analysis,
reporting of baseline characteristics, and outcome measures
reported.
For each study, two researchers conducted data extraction
and appraisal independently, and any disagreements or
discrepancies were resolved by discussion. Where consensus
could not be obtained, a third reviewer was available for
consultation.
Data analysis
The primary outcome measure was taken to be that stated by
the study or the first anxiety measure mentioned if none was
identified. Additional post hoc (calculated as part of this
review) calculations, including the mean difference and
standardised mean difference (effect size), were performed
where possible to estimate the effect and to allow compar-
isons to be made between the studies.
20
Clinical commentaries
A clinician with relevant training and experience was asked
to comment on each study focusing on clinical relevance and
practical issues. Commentary frameworks were specifically
developed for this project, and these incorporate a number of
closed and open questions with space for further comments.
Table 1 provides summaries of these commentaries.
RESULTS
Systematic reviews
There were no systematic reviews found specifically on the
topic of yoga for anxiety or anxiety disorders, although yoga
is included in a systematic review of complementary and self
help treatment for anxiety disorders.
21
Randomised and non-randomised controlled trials
Eight studies were found. All were controlled trials; six were
randomised
22–27
and two were non-randomised.
28 29
In five
studies the inclusion criterion was a diagnosis of an anxiety
disorder (anxiety neurosis, obsessive compulsive disorder,
and psychoneurosis).
22–24 28 29
In two studies participants were
suffering from anxiety, namely examination anxiety and
snake phobia, as determined by the authors from measure-
ment scales,
25 27
and in the remaining study, participants
were about to undergo a procedure, which was anticipated to
be anxiety provoking, namely an examination.
26
Other studies
No relevant qualitative research studies were located.
Excluded studies
Four studies (five papers) were excluded in which the
participants were ostensibly healthy, and anxiety was
measured as one of a battery of outcome measures.
10–14
A
brief description of these has already been given in the
introduction. Four studies in which the primary inclusion
criterion was a physical illness were excluded.
30–33
Two studies
were excluded because of the lack of a control group.
34 35
Language of studies included
All studies were in English, six were from India and one each
from the United States and Canada.
Summary of each study
Table 1 presents a breakdown of the study methodology plus
the main results as identified in the original study plus post
hoc calculations.
In a hospital based study in the United States, Shannahoff-
Khalsa et al
22
used yoga as a treatment intervention for
obsessive compulsive disorder (OCD). Study participants
Yoga for anxiety 885
www.bjsportmed.com

Table 1 Summary of studies
Study Inclusion criteria Exposure* Outcome measure(s)
Results from study for primary
outcome measure Comments
Shannahoff-Khalsa
et al
22
(n = 22)
OCD.
DSM-III-R diagnosis plus
minimum of 15 on Y-BOCS
for the adults.
Yoga treatment: kundalini yoga (includes
specific OCD technique as well as mantra
meditation) (n= 12) (7).
Control: relaxation response and
mindfulness meditation (n = 10) (7).
Both were one hour weekly treatments
with instructor plus daily practice.
Trial duration: 3 months
Primary: Y-BOCS at baseline
and after three months.
Others: SCL-90-R (OC and GSI),
POMS, PSS, and PIL
Significant between group mean
difference (p=0.047).
Intention to treat analysis: significant
reduction in mean for yoga group
(p = 0.023) but not for control group
(p = 0.058).
Mean reduction: yoga group = 9.43
(SD 7.21) (238.4%); control
group = 2.86 (SD 3.13) (213.9%).
Mean difference = 6.57, 95% CI
(0.10 to 13.0) p = 0.047 (p = 0.058`).
Standardised mean difference = 1.10,
95% CI (20.02 to 2.22)
Methodology: randomisation:
adequate.
Allocation concealment: adequate.
Blinding of assessors: unknown.
Attrition rate: yoga, 41.7% (includes
the one adolescent); control, 30%
Sharma et al
28
1
(n = 71)
Anxiety neurosis.
Diagnosed with Feighner’s
diagnostic criteria
Yoga treatment: Yoga (kapalabhati and
ujjayi pranayama).
One week training then twice daily practice
for 10 min each time (n = 41) (33).
Control: placebo capsule, once a day
(n = 30) (24).
Trial duration: 12 weeks
Primary: HAS at three weekly
intervals.
Others: SUD, various
physiological measures and
PSLES
Significant between group mean difference
(p,0.005) at three weeks indicating greater
improvement in yoga group than in control
group (p,0.005).
Significant improvement for yoga group
between 3 and 6 weeks (p,0.01) but
not for control group.
Mean reduction: yoga group = 6.91
(SD 4.26) (223.7%); control group = 3.71
(SD 2.18) (percentage change incalculable
because of unreliable baseline figure).
Mean difference = 3.20 95% CI (1.30 to
5.10) p = 0.001.
Standardised mean difference = 0.89,
95% CI (0.34 to 1.44)
Methodology: randomisation: non-
randomised.
Allocation concealment: N/A.
Blinding of assessors: unknown.
Attrition rate: yoga, 19.5% at 3
weeks, 34.1% at 6 weeks; control,
20% at 3 weeks, 46.7% at 6 weeks;
both increasing later.
Clinical: anxiety neurosis is not a
diagnosis used now. Follow up
length seems appropriate
Sahasi et al
23
(n = 91)
Anxiety neurosis.
DSM-III diagnosis
Yoga treatment: yoga (consisting of
sukhasna, talasna, pranayama,
nishpand bhav, savasna).
Daily practice, 5 days a week instructed,
2 days home practice, 40 min duration
(n = 38) (30).
Control: diazepam (no dose or frequency
given) (n = 53) (18).
Trial duration: 3 months
Primary: IPAT at baseline
and after 3 months.
Others: SSI, LoC, and KcIT
Mean changes significant for yoga group
(p,0.05) but not for diazepam group
(p.0.05).
Mean reduction: yoga group = 3.39
(SD 6.81); control group = 20.36
(SD 8.58).
Mean difference = 3.75, 95% CI (20.76
to 8.26) p = 0.101.
Standardised mean difference = 0.49,
95% CI (20.10 to 1.08)
Methodology: randomisation:
inadequate.
Allocation concealment:
inadequate.
Blinding of assessors: unknown.
Attrition rate: yoga, 21.1%; control,
66.0%.
Clinical: anxiety neurosis is not a
diagnosis used now
Vahia et al
24
(n = 27)
Psychoneurosis.
Diagnosed (method
not stated)
Yoga treatment: five steps: asana, pranayama,
pratyahara, dharana, and dhyana (n = 15).
Control: relaxation with postures resembling
Asanas and breathing practices resembling
Pranayama plus writing (n = 12).
Both groups were given placebo tablets.
Both practised daily, 1 hour, 6 days a week.
Trial duration: 4 weeks minimum
Primary: TAS (weekly, including
before and after treatment).
Others: Rorschach and MMPI
Significant difference between group means
after (p, 0.001) but not before treatment
(p = 0.17).
Reduction in mean for yoga group but not
control.
Concludes significance at 5% level.
Mean reduction: yoga group = 6.66
(226.1%); control group = 20.50 (+1.7%).
Mean difference = 7.16
Methodology: randomisation:
unknown.
Allocation concealment: unknown.
Blinding of assessors: adequate.
Attrition rate: unknown.
Clinical: psychoneurosis is no longer
used as a diagnosis
Vahia et al
29
(n = 39)
Psychoneurosis or
psychosomatic disorder.
Diagnosed (method not
stated)
Yoga treatment: as Vahia et al
28
(n = 21).
Control: anxiolytic and antidepressant drugs
(amitriptyline and chlordiazepoxide), variable
dosage schedule (n = 18).
Trial duration: 6 weeks
Primary: TAS (initial (pre),
intermediate and final (post)
evaluations).
Others: HDRS and BSASI
Significantly greater pre-post reduction in
scores for yoga group than for control group
(p,0.05)
Methodology: randomisation: non-
randomised.
Allocation concealment: N/A.
Blinding of assessors: adequate.
Attrition rate: unknown
886 Kirkwood, Rampes, Tuffrey, et al
www.bjsportmed.com

Study Inclusion criteria Exposure* Outcome measure(s)
Results from study for primary
outcome measure Comments
Broota and
Sanghvi
25
(n = 30)
Examination anxiety (high
score on STAS) plus history
of examination anxiety
Yoga treatment: Broota relaxation technique
(exercises adapted from yoga combined with
autosuggestion) (n = 10).
Control 1: Jacobson’s progressive relaxation
technique (n = 10); control 2: talking only
(n = 10).
All were practised on consecutive days,
20 min duration each.
Trial duration: 3 days
Primary: ACL at baseline (pre)
and after 3 days (post).
Others: self evaluation ladder
scale (anxiety measure)
Significant difference between the three
groups on the pre-post changes (p,0.01).
Mean of the pre-post changes for Broota
group significantly greater than Jacobson
group (p,0.05) and talking control group
(p,0.01)
Methodology: randomisation:
unknown.
Allocation concealment: unknown.
Blinding of assessors: not blinded
(self assessed).
Attrition rate: ‘‘many’’ dropouts for
the control group.
Clinical: role of author – Broota – in
study not clear
Malathi and
Damodaran
26
(n = 50)**
Examination anxiety
anticipated because of
impending examination
Yoga treatment: yoga based intervention
(various asanas, plus prayer, exercises,
visualisation, and meditation) (n = 25)
Control: work such as reading and writing
(n = 25).
Both were practised for 1 hour, 3 times
a week
Trial duration: 3 months
Primary: STAI at pre and post
practice, one month before
exams and on the day of the
examination
Significant pre-post reduction in mean for
yoga group 1 month before examination
and on day of examination (p,0.001 for
both), no such significant reduction for the
control group.
Mean reduction on day of examination:
yoga group = 15.91 (234.0%); control
group = 21.55 (+3.4%).
Mean difference = 17.46
Methodology: randomisation:
unknown.
Allocation concealment: unknown.
Blinding of assessors: not blinded
(self assessed).
Attrition rate: unknown
Norton and
Johnson
27
 (n = 40)
Snake phobia (moderate)
identified by SNAQ responses
Yoga treatment: slightly modified form of
Agni yoga (n = 20).
Control: progressive relaxation (n= 20).
Phase 1: both were practised for 4 sessions
(45 min each) plus home practice. Total
duration: 3 weeks.
Phase 2: one 15 min practice then one off
approach of snake. Variable scheduling
dependent on participant’s convenience
Primary: SNAQ at baseline
(before phase 1) and after
phase 2.
Others: relaxation score
(phase 1), approach score,
pulse rate, and subjective fear
(phase 2)
Significant group by treatment interaction
on post phase 2 SNAQ (p,0.05).
No significant simple main effects on post
phase 2 SNAQ.
Reduction for yoga group = 4.99 (212.7%);
control group = 3.20 (28.6%).
Mean difference = 1.79
Methodology: randomisation:
unknown.
Allocation concealment: unknown.
Blinding of assessors: not blinded
(self assessed).
Attrition rate: unknown
Clinical: clinical relevance
questionable
Note: values in italics were calculated post hoc as part of this review.
*Values in parentheses are minus dropouts.
Values reproduced or calculated where data were sufficient, standardised mean differences calculated as per Hedges formula, bias corrected.
39
`Without the assumption of equal variances.
40
1Results included at three weeks only
Assumed values given were standard deviations rather than standard errors.
**Results included for day of examination only.
Results for cognitive and somatic anxiety groups combined.
DSM-III-R, Diagnostic and statistical manual of mental disorders, third edition, revised; OCD, Obsessive compulsive disorder; SD, standard deviation; CI, confidence interval.
Key for scales: ACL, anxiety check list; BSASI, Bell’s social adjustment scale; HAS, Hamilton anxiety scale; HDRS, Hamilton’s depression rating scale (HRSD, HAMD); IPAT, Institute for personality and ability testing (anxiety scale); KcIT, Knox-
cube imitation test; LoC, locus of control; MMPI, Minnesota multiphasic personality inventory; PIL, purpose in life test; POMS, profile of mood states; PSLES, presumptive stressful life events scale; PSS, perceived stress scale; SCL-90-R OC and
GSI, symptoms checklist-90-revised obsessive compulsive and global severity index; SNAQ, snake attitude questionnaire; SSI, symptom sign inventory; STAI, state-trait anxiety inventory; STAS, Spielberger’s test anxiety scale; SUD, subjective
unit of disturbance; TAS, Taylor’s anxiety scale; Y-BOCS, Yale-Brown obsessive compulsive scale.
Table 1 Continued
Yoga for anxiety 887
www.bjsportmed.com

were all from the community, each with a confirmed DSM-
III-R (Diagnostic and statistical manual of mental disorders, third
edition, revised) diagnosis of OCD. The sample was rando-
mised, with 12 practising a version of kundalini yoga, which
consisted of a number of techniques including mantra
meditation as well as an OCD specific technique, which
involved breathing through only the left nostril (see
Shannahoff-Khalsa et al
36
for full details). The remaining 10
patients practised a control regimen, which was also
meditative, enabling the researchers to test the hypothesis
that ‘‘meditation techniques in general may not be effective’’
and that ‘‘disorder-specific’’ techniques may be required. At
three months, the yoga group showed significantly greater
improvements on the Yale-Brown obsessive compulsive scale
and other scales than the control group. Post hoc analysis
indicates, however, that this may only be true under the
assumption of equal variances (table 1). A reduction of 20–
35% on the Yale-Brown obsessive compulsive scale is deemed
clinically significant by the authors. This was achieved for the
yoga group (38.4% reduction) but not the control group
(13.9% reduction).
Two hospital based Indian studies investigated yoga as a
treatment for anxiety neurosis among psychiatric outpatients
diagnosed by either Feighner’s diagnostic criteria
28
or DSM-
III.
23
In a non-randomised study, Sharma et al
28
found
significantly greater improvement in the yoga group than
the placebo control group on the Hamilton anxiety scale
measures, at least in the first three weeks of the trial. After
three weeks there was a reduction of 23.7% among the yoga
group. This reduction is likely to be of clinical significance;
however, the mean Hamilton anxiety scale score of 22.7 after
the intervention is still at a level that would normally require
treatment.
37
Other results based on the subjective intensity of
anxiety symptoms were more mixed. Sahasi et al
23
compared
yoga with diazepam, an established anxiolytic, in a rando-
mised controlled trial. At the end of the three month trial, the
authors discovered that the yoga group had recorded
significantly lower Institute for Personality and Ability
Testing anxiety scale scores and symptom sign inventory
scores, findings that were not replicated in the diazepam
group. They also found that overall improvement as assessed
by the consultant psychiatrist was greater for the yoga group
(76.7%) than for the diazepam group (50%), although no
statistical significance is attached to this. A post hoc t test on
the between group mean difference for the Institute for
Personality and Ability Testing score was non-significant
(table 1).
Vahia et al
24 29
conducted two studies as part of a larger nine
year trial of what they term psychophysiological treatment
for psychoneurosis. Psychoneurosis is a term no longer used
as a diagnosis, which originally referred to both anxiety and
depressive disorders. The psychophysiological treatment was
a form of yoga based on the concepts of Patanjali. In the first
of the studies,
24
the yoga treatment was compared with a
pseudo-yoga treatment, designed to act as a control which
could ensure patient blinding. Both groups of participants
were equivalent for baseline anxiety measured by Taylor’s
anxiety scale. After their respective treatments, the genuine
yoga group recorded significantly lower anxiety scores than
the control group. In the second study,
29
yoga treatment was
compared with the anxiolytic and antidepressant drugs
chlordiazepoxide and amitriptyline, this time among patients
with a diagnosis of either psychoneurosis or psychosomatic
disorder. The results suggested that the patients who had
practised yoga recorded a significantly greater reduction in
Taylor’s anxiety scale score than those in the drug group.
Two Indian studies investigated the use of yoga as a
treatment for examination anxiety.
25 26
Broota and Sanghvi
25
carried out a three way comparison between their Broota
relaxation technique, Jacobson’s progressive relaxation tech-
nique, and a control among university students who had a
history of examination anxiety and recorded a high baseline
score on Spielberger’s test anxiety scale. The Broota relaxa-
tion technique consists of a set of four exercises adapted from
yoga, combined with auto-suggestion. After the three day
intervention, the Broota relaxation group recorded a sig-
nificantly greater improvement on the anxiety check list
measure than the Jacobson’s progressive relaxation group
and the control group. There were no significant between
group differences for the self evaluation ladder scale
measure. In a three month trial conducted with medical
students, Malathi and Damodaran
26
found a significant
reduction in anxiety after treatment for the yoga group one
month before examinations and on the actual day of
examinations. There was no such significant reduction for
the control group. Post hoc t tests between the group means
after treatment showed a significant difference one month
before the examination and on the day of the examination
(p,0.001 for both). No such difference existed before
treatment at either time point. On the day of the examina-
tion, the mean state-trait anxiety inventory score for the yoga
group fell by 34.0% from the moderate anxiety range before
treatment to the low anxiety range after treatment, repre-
senting a clinically significant change.
38
Finally, in a study of the management of specific phobias,
Norton and Johnson
27
treated snake anxious, first year
psychology students with either yoga or progressive relaxa-
tion. The aim of the study was to test the hypothesis that
yoga may be more appropriate for cognitive anxiety whereas
progressive relaxation may be more beneficial for somatic
anxiety. The results give some support for this hypothesis.
Out of the eight studies described above, standardised
mean differences were calculated for only three studies
where there were sufficient data and are presented in table 1.
Appraisal of the methodology
Measure of study quality
The reporting of study methodology was poor in most of the
studies, and there were also some methodological inadequa-
cies (table 1). The potential for bias is therefore high.
Although six out of eight studies were randomised, only one
showed an adequate method of randomisation and allocation
concealment.
22
One study used an inadequate serial number-
ing method with a subsequent block move of patients to the
control group because of an inability to perform yoga.
23
In all
the other randomised controlled trials, no detail of the
method of randomisation or concealment of allocation is
given. Given the nature of yoga, blinding of participants
would generally not be considered feasible. Two studies
were described as double blind,
24 29
although participants
were not blinded in one of the studies.
29
It is still possible
to blind the outcome assessor, and this was carried out in
both of these studies,
24 29
but was either not done, was not
possible because of self assessment, or is not mentioned in
the others. There was a large loss to follow up in the studies
by Sahasi et al
23
and Sharma et al
28
(beyond three weeks), and
this was a concern in the paper of Shannahoff-Khalsa et al,
22
which may bias the results. None of these three studies gave
any reasons for the dropouts, and no detail at all was given in
the others.
Other methodology
Only one study reported a power calculation, fulfilled the
minimum sample size even allowing for dropouts, and gave
information on the clinical significance of the results.
22
The
issue of clinical significance has been addressed in this review
where possible, but two of the studies lacked sufficient
data,
23 29
two studies used their own scales,
25 27
and the
888 Kirkwood, Rampes, Tuffrey, et al
www.bjsportmed.com

necessary information on the scale used in one was not
available.
24
In only one of the studies was any intention to
treat analysis carried out and reported.
22
Two of the studies
clearly did not calculate their results on this basis,
23 28
and in
the others there was no mention of this issue or there was
insufficient detail to deduce the basis of the calculations. The
reporting of age, sex, and baseline anxiety distribution
between the groups was inadequate in some studies.
23–29
This is a particularly serious problem in the study by Sahasi
et al
23
because of the block move of participants after
randomisation. Three studies detailed age and sex distribu-
tion.
22 23 28
Baseline anxiety figures for the primary outcome
measure were satisfactorily equivalent between the treatment
groups in three studies,
22 24 26
and in one study equivalence is
stated.
27
The main co-interventions dealt with were con-
current drug treatments and psychotherapy. Three of the
studies gave some detail on this issue, but none of them dealt
with it in any way adequately or checked that any
instructions given were adhered to.
22 23 28
The other studies
gave no detail on this issue. Feighner’s diagnostic criteria, a
forerunner to the inclusion of diagnostic criteria in the DSM,
was used in one study.
28
Two studies did not state which
diagnostic criteria they used.
24 29
Two of the studies had errors
in the labelling in the tables, which necessitated some
assumptions in the interpretation of the data in them.
26 27
In
one study, some of the data reproduced in the tables were
unreliable.
28
Statistical methods
Adequate statistical methods were used in most stu-
dies.
22 25 27–29
In one study where only a pre-post t test for
each group was performed, there were sufficient data to
allow a post hoc, between group comparison on the mean
difference to be performed.
23
The remaining two studies had
only sufficient data to allow a t test on the post-test means,
although baseline equivalence with respect to anxiety
measure ensured that this comparison was meaningful.
24 26
DISCUSSION
A number of controlled trials, both randomised and non-
randomised, that tested yoga as an intervention with respect
to anxiety and anxiety disorders are described in this review.
There were eight studies in which participants were suffering
from anxiety or had a diagnosed anxiety disorder, and all
registered positive results in favour of yoga. However, the
overall reporting of study methodology was poor, and there
were many methodological inadequacies in the studies. It is
encouraging though that the best study methodologically
22
provided a positive result.
The smallest standardised mean difference was in the
study by Sahasi et al,
23
which is perhaps not surprising given
the choice of diazepam as a control treatment. The other two
studies that had sufficient data to allow standardised mean
differences to be calculated produced respectable figures
demonstrating a notable effect.
22 28
There is a shortage of information on safety or contra-
indications with respect to yoga for psychiatric disorders.
Little additional information on adverse effects related to the
practice of yoga can be gained from the studies included in
this review. Shannahoff-Khalsa et al
22
reported an absence of
adverse effects, but there is no mention of adverse effects
within the other studies. As well as recommending that
medical advice be sought before yoga is taken up by anybody
with a medical condition, Ernst
2
states that certain postures
are not recommended during pregnancy, that overstretching
joints may lead to physical damage, and that the meditation
aspect may best be avoided by anyone with a history of
psychotic or personality disorder. Becker
41
states that
meditation is impossible in the presence of active psychosis.
No further details or references are given in either source. The
evidence in general on contraindications with regard to
psychosis and possible psychotic side effects of meditation
appears at this stage to be in the form of case reports,
42 43
and
is certainly an area that requires more investigation. With
exercise in general, there is some concern that undertaking a
programme that is too intense for a person can lead to a
worsening of mood.
7
Motivation and compliance may also be issues that require
consideration before recommendation of a programme such
as yoga.
41
Low participation and high attrition rates, 50% lost
within three to six months, have been found for exercise.
7
Dropout rates were significant or high in all three studies that
reported on this.
22 23 28
The ability to pursue a commitment to
a yoga programme may well be an issue for clients with
anxiety disorders, as may the type of yoga used. There are
many different forms of yoga,
44
and some may be more
suitable than others. Careful consideration should be given to
these matters before recommendation of any programme,
and it would be desirable for instructors to have suitable
mental health experience. Nespoor
45
provides some useful
tips on compliance, such as using family support structures
and emphasising the non-competitive nature of yoga.
The issue of publication bias has not been dealt with,
although efforts were made to find unpublished studies and
ongoing research. It is possible, however, that other, less
positive findings remain unreported. A meta-analysis of the
results was not considered appropriate because of insufficient
data and study heterogeneity, mainly in the conditions
treated.
If yoga does produce an anxiolytic effect, the exact causal
mechanism is likely to be complex. Yoga may best be
delivered as a complete intervention, and if different aspects
are delivered separately, such a reductionist approach
may result in loss of efficacy or effectiveness. Having said
that, it is still worth while devising appropriate controls
in an attempt to tease out the yoga specific effects. Similar
to exercise, biochemical or physiological mechanisms as
well as improvements in self esteem and social situation may
What is already known on this topic
N
A systematic review of complementary and self help
treatments for anxiety disorders concluded that, in the
absence of well conducted studies, it is impossible to
say whether yoga is effective
N
There is surprisingly little research on yoga for clinical
anxiety
What this study adds
N
This new review concentrates exclusively on yoga and
is broader in its remit in that it addresses anxiety and
anxiety disorders
N
Owing to the diversity of conditions treated and the
poor quality of most of the studies, it is still not possible
to say that yoga is effective in treating anxiety or
anxiety disorders in general, but there are encouraging
results, particularly with obsessive compulsive disorder
Yoga for anxiety 889
www.bjsportmed.com

well be important,
46
as could distraction from negative
thoughts.
7
Yoga is an attractive therapeutic option because of its
popularity recently demonstrated in the United States,
47
and,
like exercise, may be of particular use where clients reject
ostensibly psychological diagnoses and treatments.
7
The
National Institute for Clinical Excellence (NICE) recommend
that patients with panic disorder and generalised anxiety
disorder are informed about exercise as part of good general
health.
48
If proved efficacious, yoga would be an attractive
option because it is non-pharmacological, has minimal
adverse effects if practised as recommended, and enjoys
international acceptance.
5
In summary, the eight studies reviewed here report positive
findings for the use of yoga in OCD,
22
examination
anxiety,
25 26
snake phobia,
27
anxiety neurosis,
23 28
and psycho-
neurosis,
24 29
although the latter two diagnostic terms are no
longer used. There were, however, many methodological
inadequacies, and only the OCD study
22
could be described as
being methodologically rigorous.
Owing to the diversity of conditions treated and the poor
quality of most of the studies, it is not possible to say that
yoga is effective in treating anxiety or anxiety disorders in
general. However, there are encouraging results, particularly
with OCD. Further well conducted research is necessary
which may be most productive if focused on specific anxiety
disorders.
ACKNOWLEDGEMENTS
We thank the following: Anelia Boshnakova, Electronic Information
Officer, RCCM for advice and support with search strategies and
searches; Hugh McGuire, Trial Search Coordinator, Institute of
Psychiatry, London for searches of the CCDAN database; Dr Ursula
Werneke, Consultant Psychiatrist, Homerton Hospital, London for
helpful comments on the draft review; The Project Advisory Group
and Specialist Advisory Group (mental health) for the NHS Priorities
Project for advice and support to the project.
Authors’ affiliations
.....................
G Kirkwood, Research Council for Complementary Medicine, London,
UK
H Rampes, Barnet, Enfield & Haringey Mental Health NHS Trust,
Northwest Community Mental Health Team, Edgware, Middlesex, UK
V Tuffrey, School of Integrated Health, University of Westminster,
London, UK
J Richardson, Faculty of Health and Social Work, University of Plymouth,
Plymouth, UK
K Pilkington, Research Council for Complementary Medicine, London,
UK. School of Integrated Health, University of Westminster, London, UK
The NHS Priorities Project is funded by the Department of Health. The
views and opinions expressed are those of the authors and do not
necessarily reflect those of the Department of Health.
Competing interests: none declared
REFERENCES
1 Office for National Statistics. Psychiatric morbidity among adults living in
private households. London: The Stationery Office, 2000.
2 Ernst E. Therapies: yoga (section 3). In: Ernst E, ed. The desktop guide to
complementary and alternative medicine. An evidence-based approach.
Edinburgh: Mosby, 2001:76–8.
3 Riley D. Hatha yoga and the treatment of illness. Altern Ther Health Med
2004;10:20–1.
4 Khalsa SB S. Yoga as a therapeutic intervention: a bibliometric analysis of
published research studies. Indian J Physiol Pharmacol 2004;48:269–85.
5 Ramaratnam S, Sridharan K. Yoga for epilepsy (Cochrane Review). Cochrane
Library. Issue 3. Chichester: John Wiley & Sons, Ltd, 2004.
6 Petruzzello S, Landers D, Hatfield B, et al. A meta-analysis on the anxiety-
reducing effects of acute and ch ronic exercise: outcomes and mechanisms.
Sports Med 1991;11:143–82.
7 Salmon P. Effects of physical exercise on anxiety, depression, and sensitivity to
stress: a unifying theory. Clin Psychol Rev 2001;21:33–61.
8 Scully D, Kremer J, Meade M, et al. Physical exercise and psychological well
being: a critical review. Br J Sports Med 1998;32:111–20.
9 Dunn AL, Trivedi MH, O’Neal HA. Physical activity dose-response effects on
outcomes of depression and anxiety. Med Sci Sports Exerc
2001;33:S587–97.
10 Berger BG, Owen DR. Stress reduction and mood enhancement in four
exercise modes: swimming, body conditioning, hatha yoga, and fencing.
Res Q Exerc Sport 1988;59:148–59.
11 Ray US, Mukhopadhyaya S, Purkayastha SS, et al. Effect of yogic exercises on
physical and mental health of young fellowship course trainees. Indian J Physiol
Pharmacol 2001;45:37–53.
12 Netz Y, Lidor R. Mood alterations in mindful versus aerobic exercise modes.
J Psychol 2003;137:405–19.
13 Blumenthal JA, Emery CF, Madden DJ, et al. Cardiovascular and behavioural
effects of aerobic exercise training in healthy older men and women.
J Gerontol 1989;44:M147–57.
14 Blumenthal JA, Emery CF, Madden DJ, et al. Long-term effects of exercise on
psychological functioning in older men and women. J Gerontol
1991;46:352–61.
15 International Association of Yoga Therapists. http://iayt.org/ (accessed 10
Nov 2004).
16 Yoga Biomedical Trust. http://www.yogatherapy.org/ (accessed 10 Nov
2004).
17 MIND. http://www.mind.org.uk/ (accessed 10 Nov 2004).
18 Mental Health Foundation. http://www.mentalhealth.org.uk/ (accessed 10
Nov 2004).
19 Centre for Reviews, Dissemination (CRD). Report number 4 (2nd ed).
Undertaking systematic reviews of research on effectiveness. York: CRD,
2001.
20 The Cochrane Collaboration. Cochrane reviewers’ handbook 4.2.2. Updated
March, 2004:78–9.
21 Jorm AF, Christensen H, Griffiths KM, et al. Effectiveness of complementary
and self-help treatments for anxiety disorders. Med J Aust 2004;181:S29–46.
22 Shannahoff-Khalsa DS, Ray LE, Levine S, et al. Randomized controlled trial of
yogic meditation techniques for patients with obsessive-compulsive disorder.
CNS Spectr 1999;4:34–47.
23 Sahasi G, Mohan D, Kacker C. Effectiveness of yogic techniques in the
management of anxiety. Journal of Personality Clinical Studies 1989;5:51–5.
24 Vahia NS, Doongaji DR, Jeste DV, et al. Further experience with the therapy
based upon concepts of Patanjali in the treatment of psychiatric disorders.
Indian J Psychiatry 1973;15:32–7.
25 Broota A, Sanghvi C. Efficacy of two relaxation techniques in examination
anxiety. Journal of Personality Clinical Studies 1994;10:29–35.
26 Malathi A, Damodaran A. Stress due to exams in medical students: role of
yoga. Indian J Physiol Pharmacol 1999;43:218–24.
27 Norton GR, Johnson WE. A comparison of two relaxation procedures for
reducing cognitive and somatic anxiety. J Behav Ther Exp Psychiatry
1983;14:209–14.
28 Sharma I, Azmi SA, Settiwar RM. Evaluation of the effect of pranayama in
anxiety state. Alternative Medicine 1991;3:227–35.
29 Vahia NS, Doongaji DR, Jeste DV, et al. Psychophysiologic therapy based on
the concepts of Patanjali. A new approach to the treatment of neurotic and
psychosomatic disorders. Am J Psychother 1973;27:557–65.
30 Broota A, Varma R, Singh A. Role of relaxation in hypertension. Journal of
Indian Academy of Applied Psychology 1995;21:29–36.
31 Cohen L, Warneke C, Fouladi RT, et al. Psychological adjustment and sleep
quality in a randomized trial of the effects of a Tibetan yoga intervention in
patients with lymphoma. Cancer 2004;100:2253–60.
32 Oken BS, Kishiyama S, Zajdel D, et al. Randomized controlled trial of yoga
and exercise in multiple sclerosis. Neurology 2004;62:2058–64.
33 Taneja I, Deepak KK, Poojary G, et al. Yogic versus conventional treatment in
diarrhea-predominant irritable bowel syndrome: a randomized control study.
Appl Psychophysiol Biofeedback 2004;29:19–33.
34 Allen KS, Steinkohl RP. Yoga in a geriatric mental clinic. Activities, Adaptation
& Aging 1987;9:61–8.
35 Bhushan S, Sinha P. Yoganidra and management of anxiety and hostility.
Journal of Indian Psychology 2001;19:44–49.
36 Shannahoff-Khalsa D-S. Kundalini yoga meditation techniques for the
treatment of obsessive-compulsive and OC spectrum disorders. Brief
Treatment Crisis Intervention 2003;3:369–82.
37 Best Treatments - NHS Direct. http://www.besttreatments.co.uk/btuk/
conditions/9962.html (accessed 13 June 2005).
38 Sesti A-M. State Trait Anxiety Inventory (STAI) in medication clinical trials. QoL
Newsletter 2000;25:15–16.
39 Curriculum Evaluation and Management Centre. http://
www.cemcentre.org/ebeuk/research/effectsize/Calculator.htm/ (accessed
11 Apr 2004).
40 Woodward M. Epidemiology: study design and data analysis. London:
Chapman & Hall, 1999:77.
41 Becker I. Uses of yoga in psychiatry and medicine (chapter 3). In: Muskin P,
ed. Complementary and alternative medicine and psychiatry (Review of
psychiatry series, vol 19, no 1; Oldham JM, Riba MB, series eds).
Washington, DC: American Psychiatric Press, 2000:111, 133.
42 Hansen W. Psychoses and meditation. Ugeskr Laeger 1981;143:20–2.
43 Lazarus AA. Psychiatric problems precipitated by transcendental meditation.
Psychol Rep 1976;39:601–2.
44 Hoare S. How do I choose between the different types of yoga? The Mind
guide to yoga. Rev, ed. London: MIND, 2004:6–7.
45 Nespoor K . Twelve years of experience with yoga in psychiatry.
Int J Psychosom 1993;40 :105–7.
46 Fox K. The influence of physical activity on mental well-being. Public Health
Nutr 1999;2:411–18.
890 Kirkwood, Rampes, Tuffrey, et al
www.bjsportmed.com

47 Saper HB, Eisenberg DM, Davis RB, et al. Prevalence and patterns of adult
yoga use in the United States: results of a national survey. Altern Ther Health
Med 2004:44–9.
48 National Institute for Clinical Excellence. Anxiety: management of anxiety
(panic disorder, with or without agoraphobia, and generalised anxiety
disorder) in adults in primary, secondary and community care: quick
reference guide. December 2004.
methods such as yoga, thus avoiding the adverse effects and
possible habituation associated with anxiolytics. It highlights
the weak methods used in studies of non-pharmacological
interventions and the difficulty in performing a meta-analysis
of psychological outcome measures. The results are likely to be
confounded by lack of blinding, various types of intervention,
varying duration of follow ups, and varied outcome measures,
making it difficult to present summary outcome data. The need
for good quality studies is emphasised.
S Ramaratnam
Department of Neurology, Apollo Hospitals, Madras, India;
rsridharan@vsnl.com
..............
COMMENTARY
..............
This review is timely, addressing the important question of
whether anxiety can be treated by non-pharmacological
Clinical Evidence—Call for contributors
Clinical Evidence is a regularly updated evidence-based journal available worldwide both as
a paper version and on the internet. Clinical Evidence needs to recruit a number of new
contributors. Contributors are healthcare professionals or epidemiologists with experience in
evidence-based medicine and the ability to write in a concise and structured way.
Areas for which we are currently seeking contributors:
N
Pregnancy and childbirth
N
Endocrine disorders
N
Palliative care
N
Tropical diseases
We are also looking for contributors for existing topics. For full details on what these topics
are please visit www.clinicalevidence.com/ceweb/contribute/index.jsp
However, we are always looking for others, so do not let this list discourage you.
Being a contributor involves:
N
Selecting from a validated, screened search (performed by in-house Information
Specialists) epidemiologically sound studies for inclusion.
N
Documenting your decisions about which studies to include on an inclusion and exclusion
form, which we keep on file.
N
Writing the text to a highly structured template (about 1500-3000 words), using evidence
from the final studies chosen, within 8-10 weeks of receiving the literature search.
N
Working with Clinical Evidence editors to ensure that the final text meets epidemiological
and style standards.
N
Updating the text every 12 months using any new, sound evidence that becomes available.
The Clinical Evidence in-house team will conduct the searches for contributors; your task is
simply to filter out high quality studies and incorporate them in the existing text.
If you would like to become a contributor for Clinical Evidence or require more information
about what this involves please send your contact details and a copy of your CV, clearly
stating the clinical area you are interested in, to CECommissioning@bmjgroup.com.
Call for peer reviewers
Clinical Evidence also needs to recruit a number of new peer reviewers specifically with an
interest in the clinical areas stated above, and also others related to general practice. Peer
reviewers are healthcare professionals or epidemiologists with experience in evidence-based
medicine. As a peer reviewer you would be asked for your views on the clinical relevance,
validity, and accessibility of specific topics within the journal, and their usefulness to the
intended audience (international generalists and healthcare professionals, possibly with
limited statistical knowledge). Topics are usually 1500-3000 words in length and we would
ask you to review between 2-5 topics per year. The peer review process takes place
throughout the year, and out turnaround time for each review is ideally 10-14 days.
If you are interested in becoming a peer reviewer for Clinical Evidence, please complete the
peer review questionnaire at www.clinicalevidence.com/ceweb/contribute/peerreviewer.jsp
Yoga for anxiety 891
www.bjsportmed.com