Content uploaded by Lori Pbert
Author content
All content in this area was uploaded by Lori Pbert on Oct 04, 2014
Content may be subject to copyright.
936 AmJPsychiatry 149:7,July 1992
Effectiveness of a Meditation-Based Stress Reduction Program
in the Treatment of Anxiety Disorders
Jon Kabat-Zinn, Ph.D., Ann 0. Massion, M.D., Jean Kristeller, Ph.D.,
Linda Gay Peterson, M.D., Kenneth E. Fletcher, Ph.D., Lori Pbert, Ph.D.,
William R. Lenderking, Ph.D., and Saki F. Santorelli, Ed.D.
Objective: This study was designed to determine the effectiveness ofa group stress reduction
program based on mindfulness meditation for patients with anxiety disorders. Method: The
22 study participants were screened with a structured clinical interview and found to meet the
DSM-IH-R criteria for generalized anxiety disorder or panic disorder with or without agora-
phobia. Assessments, including self-ratings and therapists’ ratings, were obtained weekly be-
fore and during the meditation-based stress reduction and relaxation program and monthly
during the 3-month follow-up period. Results: Repeated measures analyses ofvariance docu-
mented significant reductions in anxiety and depression scores after treatment for 20 of the
subjects-changes that were maintained at follow-up. The number of subjects experiencing
panic symptoms was also substantially reduced. A comparison of the study subjects with a
group ofnonstudy participants in the program who met the initial screening criteria for entry
into the study showed that both groups achieved similiar reductions in anxiety scores on the
SCL-90-R and on the Medical Symptom Checklist, suggesting generalizability of the study
findings. Conclusions: A group mindfulness meditation training program can effectively re-
duce symptoms ofanxiety and panic and can help maintain these reductions in patients with
generalized anxiety disorder, panic disorder, or panic disorder with agoraphobia.
(Am JPsychiatry 1992; 149:936-943)
Self-regulatory behavioral strategies, used alone or
as adjuncts to other behavioral or medication regi-
mens, may offer a unique approach to treating anxiety
disorders. Three major self-regulatory strategies-
meditation, relaxation, and biofeedback-are currently
used in clinical practice for the treatment of anxiety.
Research suggests that all three play a role in reducing
both physiological and psychological components of
Presented at the annual meeting of the Society of Behavioral Mcdi-
cine, Chicago, April 18-21, 1990, and at the First International Con-
gress on Behavioral Medicine, Uppsala, Sweden, June 27-July 1,
1990. Received Oct. 18, 1990; revision received Oct. 16, 1991; ac-
cepted Dec. 2, 1991. From the Stress Reduction and Relaxation Pro-
gram, Division of Preventive and Behavioral Medicine, Department
of Medicine, and the Department of Psychiatry, University of Massa-
chusetts Medical School. Address reprint requests to Dr. Kabat-Zinn,
Stress Reduction and Relaxation Program, Department of Medicine,
University of Massachusetts Medical School, 55 Lake Avenue North,
Worcester, MA 01605.
Supported by a biomedical research support grant from the Univer-
sity of Massachusetts Medical School.
The authors thank Julianne Miller, M.D., for assistance with diag-
nostic assessments, Craig Edelbrock, Ph.D., for research design con-
sultation, and William Smith, Anthony Bell, and Kim Irettel for data
collection and management.
Copyright C 1992 American Psychiatric Association.
anxiety in normal populations and that the latter two
techniques are effective in anxious populations, al-
though with variable efficacy (1-6).
The research on meditation techniques has been
largely limited to nonpsychiatnic populations (7). To
our knowledge, there are no studies of the effectiveness
of meditation for patients with anxiety disorders as de-
lineated by DSM-III or DSM-III-R criteria (8). Two
controlled studies (9, 10) used meditation for patients
with anxiety neurosis as defined by DSM-II criteria, but
both lacked standardized diagnostic procedures. There
was one uncontrolled study of patients diagnosed as
having anxiety neurosis (1 1 ). None of these studies used
a structured clinical interview for diagnosis. All of them
investigated variants of one particular type of medita-
tion, namely, transcendental meditation, in which the
practitioner focuses on a mantra-a word or phrase re-
peated silently to achieve a meditative state.
In general, these studies suggested that transcenden-
tal meditation may be as effective as other behavioral
techniques, such as biofeedback on relaxation, in the
treatment of anxiety. Another uncontrolled study (12)
investigated mindfulness meditation as an adjunct to
psychotherapy for patients with a wide range of psychi-
atnic disorders, excluding schizophrenia and other psy-
KABAI-ZINN, MASSION, KRISTELLER, El AL.
AmJPsychiatry 149:7,July 1992 937
choses. In that study, Kutz et al. found that according
to both the patients’ self-assessments and the therapists’
assessments, there was moderate to marked improve-
ment in a variety of psychological symptoms, including
anxiety, from before to after treatment.
The lack of diagnostic assessment according to stand-
andized diagnostic criteria in previous studies and the
widespread practice of studying nonclinical popula-
tions (e.g., college students) limit the applicability of ne-
search findings regarding the clinical effectiveness of
meditation. Moreover, the majority of the studies of the
effects of meditation on anxiety have relied solely on
measures of state-trait anxiety to determine outcome.
Such measures do not adequately assess the presence of
panic attacks on avoidance behavior and may fail to
capture the complexity of clinically significant anxiety
symptoms.
The present pilot study was devised to address some
of the shortcomings of previous research that investi-
gated the relation between meditation training and
anxiety reduction. The study was conducted in con-
junction with a well-established outpatient program for
stress reduction and relaxation that involves intensive
training in mindfulness meditation (13, 14), with em-
phasis on its practical applications in coping with stress
and in enhancing adaptive health behaviors. Like other
forms of meditation such as transcendental meditation,
mindfulness meditation helps practitioners to cultivate
greater concentration and relaxation (15). It differs spe-
cifically from transcendental meditation by training
practitioners to attend to a wide range of changing ob-
jects of attention while maintaining moment-to-mo-
ment awareness (mindfulness), rather than restricting
one’s focus to a single object such as a mantra (16) (see
the Method section for an operational definition). The
choice of mindfulness as the primary meditative ap-
proach was due to its immediate applicability to a great
variety of present-moment experiences. This onienta-
tion lends a quality of “ordinariness” to the interven-
tion that makes it more acceptable and accessible to a
wide range of people with different life stnessons and
different medical disorders (17).
The stress reduction and relaxation program serves a
broad spectrum of patients with both physical and psy-
chological disturbances (1 8). Previous studies have
shown that participation in the program results in re-
ductions in both physical and psychological symptoms
of patients in many diagnostic categories. Chronic pain
patients participating in the program reported mark-
edly reduced levels of state anxiety (as measured with
the Symptom Checklist-90-Revised) during the inter-
vention period-levels that were maintained oven a 4-
year follow-up period (17, 19, 20). Similar changes were
reported oven a 2-year follow-up period by patients
with stress-related medical disorders (Kabat-Zinn, un-
published manuscript).
The specific objectives of the present investigation
were 1) to conduct a prospective outcome study, with a
repeated measures design, to test the efficacy of treating
patients diagnosed with anxiety disorders according to
DSM-III-R criteria in a well-established, meditation-
based outpatient stress reduction program and 2) to ex-
amine whether variables at intake were predictive of
outcome at follow-up.
METHOD
Potential subjects were selected from among all pa-
tients referred to the stress reduction and relaxation
program in two consecutive cycles (spring and fall of
1988). The Symptom Checklist-90-Revised (SCL-90-R)
(21) and the Medical Symptom Checklist (17) were ad-
ministered to all patients referred to the program, as
part of the intake evaluation. Those who scored above
the 70th percentile on the anxiety subscale of the SCL-
90-R and reported more than 10 anxiety-related symp-
toms (out of 37 possible symptoms) on the Medical
Symptom Checklist were invited to take part in a for-
mal screening interview to assess their appropriateness
for inclusion in the study. A referral diagnosis of panic
attacks or anxiety also qualified an individual to be in-
vited to participate in the screening procedure for the
study. Patients who met the study criteria and who
agreed to participate were then interviewed by either a
psychologist or a psychiatrist trained in administering
the Structured Clinical Interview for DSM-HI-R (SCID)
(22). Diagnoses were determined after review of the
SCID data by the two psychologists (J.K. and L.P.) and
two psychiatrists (A.O.M. and L.G.P.) who conducted
the individual screening evaluations. Only the patients
who met the formal diagnostic criteria for generalized
anxiety disorder or panic disorder with or without ago-
naphobia were included in the study. Individuals were
excluded if they had other primary psychiatric diagno-
ses, any disorder with psychotic symptoms, any endo-
cnine disorder, or significant current alcohol or sub-
stance abuse. Because of the small sample size and the
pilot nature of the study, patients taking anxiolytic or
other medications (N=12) were not excluded. Medica-
tion type and usage were assessed for all patients during
the study.
In the two cycles of the program from which patients
were recruited for this study, 192 (60%) of 321 patients
satisfied the initial screening criteria of the SCL-90-R
and the Medical Symptom Checklist. However, for lo-
gistical reasons and because this was a pilot study, only
44 patients were invited to undergo further screening,
of whom 32 completed the evaluation. Of these, 24 met
the DSM-III-R criteria for generalized anxiety disorder
on panic disorder with or without agoraphobia accord-
ing to the SCID. Of the eight excluded patients, four
had other primary psychiatric diagnoses and four had
no psychiatric disorder. Two of the 24 subjects did not
complete the program and were not included in the
analysis of outcome. Both of these individuals had psy-
chiatnic diagnoses of generalized anxiety disorder.
Because of the exploratory nature of the study, we
used a repeated measures design with patients serving
as their own controls on multiple pretreatment and post-
MEDITATION-BASED PROGRAM FOR ANXIETY DISORDERS
938 Am JPsychiatry 149:7, July 1992
treatment measures. In addition, study participants
were compared on the SCL-90-R and Medical Symp-
tom Checklist with other patients who met the initial
screening criteria and were enrolled in the stress reduc-
tion and relaxation program during the same time pe-
nod but who were not invited to take part in the study.
This second group of patients (termed “nonstudy par-
ticipants”) received the same meditation intervention
but did not undergo screening on the weekly assess-
ments that the study subjects underwent.
Subjects who met the diagnostic criteria and agreed
to participate in the study were evaluated with both
self-rating scales and ratings of trained interviewers.
Data on the following measures were gathered by tele-
phone interview at weekly intervals from the time of re-
cruitment through the end of treatment and at monthly
intervals for 3 months after treatment: the Beck Anxi-
ely Inventory (used by special permission of Jeffrey
Sugerman, Ph.D., Psychological Corp., personal com-
munication), the Beck Depression Inventory (23), and
ratings of the frequency and severity of panic attacks.
The length of time between recruitment and the start
of treatment in which data were collected varied ac-
cording to when subjects were recruited into the study
relative to the beginning of the program (range=1-8
weeks).
In addition to these assessments, a more extensive
assessment battery was administered four times: at re-
cruitment into the study, at the start of the program
(pretreatment), at completion of the program (post-
treatment), and at 3-month follow-up. This battery
consisted of the Hamilton Rating Scale for Anxiety
(24) (as modified by DiNardo and Barlow [25] to in-
dude a separate rating scale for symptoms present dur-
ing panic attacks, yielding the Hamilton panic score),
the Hamilton Rating Scale for Depression (26), the
Fear Survey Schedule (27), and the Mobility Inventory
for Agoraphobia (28). At recruitment patients were
also asked to rate on a S-point scale their expectancy
of improvement due to the treatment. A compliance
questionnaire was administered at the end of treatment
and at follow-up. Eight subjects entered the study so
close to the beginning of the treatment intervention
that only pretreatment, posttreatment, and follow-up
measures were obtained.
The Hamilton anxiety and depression rating scales
were administered at recruitment by the same clinicians
who administered the SCID. Subsequent Hamilton as-
sessments were administered to all subjects by one
trained interviewer. To minimize bias in data collection
related to expectancy of change, scoring was done after
all data were collected.
The stress reduction and relaxation program is a
highly structured training program in mindfulness
meditation and its applications, described in detail else-
where (14, 17-20). It takes the form of an 8-week-long
course in which participants attend weekly 2-hour
classes and, in addition, a 7.5-hour intensive and
mostly silent “meditation retreat” session in the sixth
week. During each 8-week cycle, five separate but par-
allel classes are offered. Each is led by one instructor
who stays with that group for the duration of the
course. Each class has approximately 30 participants
with a wide range of medical and psychological disor-
dens. During classes and for homework, participants
practice a range of different formal and informal medi-
tation techniques (14, 17). These experiences are dis-
cussed weekly in the classes. The 22 subjects in this
study were distributed among five of the 10 classes held
during that period. The exposure of these subjects dif-
fered from that of the remainder of the program partici-
pants only in their involvement in the additional assess-
ment protocol required for the study. Four program
instructors conducted classes in this study. The instruc-
tons did not know which patients were in the study, nor
did they know the patients’ DSM-III-R diagnoses.
We used repeated measures analysis of variance
(ANOVA) to compare the recruitment, pretreatment,
posttreatment, and 3-month follow-up scores of the
subjects for whom all data points were available, with
computation of appropriate contrasts. Matched t tests
were used to calculate intervention effects between the
pretreatment and posttreatment assessments for the en-
tire sample. Intergroup comparisons of compliance and
expectancy measures were done with standard t tests.
Variables expected to predict outcome were studied
with ANOVA. We plotted the weekly scores of all sub-
jects to examine the course of change, but formal sin-
gle-subject analyses are not included in this report be-
cause of the consistency and strength of the group
effects. In addition, after accounting for pretreatment
scores with the regression technique described by Co-
hen and Cohen (29), we compared posttreatment scores
of the subjects receiving medication with those of the
subjects taking no medication. Finally, we used ttests
to compare the study participants and nonstudy partici-
pants in the program on pretreatment and posttreat-
ment SCL-90-R scores, Medical Symptom Checklist
scores, and change scores.
RESULTS
Of the 22 study participants who completed the pro-
gram, 10 had panic disorder with agoraphobia, four
had panic disorder without agoraphobia, and eight had
generalized anxiety disorder as the primary psychiatric
diagnosis. Seventeen subjects had more than one psy-
chiatnic diagnosis; 14 had other anxiety disorders and
eight had diagnoses of major depressive episode (six
concurrent). The average duration of their anxiety dis-
orders was 6.5 years (range=3 months to 28 years).
Eleven patients were taking medication for their anxi-
ety disorders at intake, and 1 1 were taking no medica-
iion for anxiety.
The subjects’ ages ranged from 26 to 64 years, with an
average of 38 years. There were five men and 17 women.
Eighteen of the subjects were married, two were single,
and one was separated (data on one subjectwere missing).
The recruitment and pretreatment scores on the Ham-
KABAT-ZINN, MASSION, KRISTELLER, El AL.
Am JPsychiatry 149:7, July 1992 939
TABLE 1. Scores on Outcome Measures Over Time of Patients With Anxiety Disorders in a Study of a Meditation-Based Stress Reduction Proam
Measure N
Initial
Recruitment
Mean SD
Pretreatment
Mean SD
Posttreatment
Mean SD
3-Month
Follow-Up
Mean SD
Repeated Measures
ANOVA
F df p
Hamilton Rating Scale for Anxiety
Hamilton Rating Scale for Depression
Beck Anxiety Inventory
Beck Depression Inventory
Fear Survey Schedule
Mobility Inventory for Agoraphobia
Accompanied
Unaccompanied
14
14
15
15
11
10
10
30.36 8.53
33.07 7.98
24.13 13.49
18.87 10.37
118.73 41.31
45.80 16.22
61.80 24.40
26.93 11.13
31.07 8.43
20.53 13.24
16.47 10.97
93.55 34.09
41.30 16.81
53.50 24.09
17.86 9.18
23.71 5.59
9.00 9.14
10.00 9.58
78.46 44.28
36.40 12.02
45.50 17.19
15.86 8.65
25.14 7.01
7.93 7.29
7.53 8.77
66.82 38.68
36.70 13.52
46.20 18.87
21.1 3, 39 <0.001k
8.87 3, 39 <0.OOV
1S.36 3, 42 <0.OOla
9.96 3, 42 <0.00l
9.79 3, 30 <0.001
4.05 3, 27 <0.05
6.62 3, 27 <0.01w
a5jfifl change from pretreatment to posttreatment (p<0.01).
‘Trend for significant change from posttreatment to follow-up (p<0.10).
cSignificant change from recruitment to pretreatment (p<0.0S).
dSigrnficant change from posttreatment to follow-up (p<0.0S).
dlrend for significant change from pretreatment to posttreatment (p<zO.10).
ilton Rating Scale for Anxiety, the Hamilton Rating
Scale for Depression, the Beck Anxiety Inventory, the
Beck Depression Inventory, the Fear Survey Schedule,
and the Mobility Inventory for Agoraphobia of the sub-
jects with complete data at the four primary assessment
points are shown in table 1. They were in the moderate
to severe range on both the Beck and the Hamilton
anxiety scales and in the mild to moderate range on the
Beck and Hamilton depression scales.
At recruitment, nine individuals reported one or
more panic attacks in the previous week (nange=1-3),
with a mean Hamilton panic scone of 26.1 1 (SD=1 1.25,
range=6-40). At pretreatment assessment, 13 individu-
als reported at least one panic attack in the previous
week (range=1-2), with a mean Hamilton panic score
of 24.46 (SD=8.71, range=1 1-34). At pretreatment the
mean SCL-90-R general severity index scone of the 22
#subjects was 1.10 (SD=0.70, range=0-3) and the mean
SCL-90-R anxiety score was 1.61 (SD=1.0S, range=0-3).
Repeated measures ANOVA indicated that among
subjects for whom scores at all four primary assessment
points were available, the Hamilton and Beck anxiety
and depression scale scores showed small, statistically
nonsignificant reductions from baseline to pretreat-
ment, highly significant decreases oven the course of the
intervention (pretreatment to posttreatment), and
maintenance of these changes from posttneatment to
follow-up (table 1). Comparisons with matched ttests
at pretreatment and posttreatment time points for all
subjects, not just those with complete data at all time
points, showed comparable results, with mean pretreat-
ment and posttreatment scores, respectively, of 25.86
(SD=1 0.56) and 17.10 (SD=9.3 1) on the Hamilton
anxiety scale (t=S.18, df=20, p<O.OOl) and 30.85 (SD=
8.81) and 23.85 (SD=6.6S) on the Hamilton depression
scale (t=4.88, df=19, p<O.OO1). Mean pretreatment and
posttreatment scones, respectively, were 20.32 (SD=
12.05) and 7.09 (SD=8.20) on the Beck Anxiety Inven-
tory (t=6.14, df=21, p<0.OO1) and 16.18 (SD=10.33)
and 8.18 (SD=8.53) on the Beck Depression Inventory
(t=4.6S, df=21, p<O.OO1). These represented mean ne-
ductions of 34%, 23%, 6S%, and 49%, respectively,
on the four scales. Twenty of the 22 subjects showed
marked improvement in scones on the Beck and Hamil-
ton anxiety and depression scales.
The means of the subjects’ weekly ratings of anxiety
and depression on the respective Beck scales are pre-
sented in figures 1 and 2. These show elevated levels
before treatment, a significant decline during treatment
to a relatively low level by the end of treatment, and
maintenance of the lower posttneatment level over 3
months of follow-up. Scores for “accompanied” on the
Mobility Inventory for Agoraphobia showed a similar
pattern of improvement. However, scores for “unac-
companied” on that inventory and scones on the Fear
Survey Schedule improved as much from recruitment to
pretreatment assessment as from pretreatment to post-
treatment assessment (table 1).
Of the 13 patients who reported at least one panic
attack in the preceding week at pretreatment assess-
ment, five reported one panic attack in the previous
week at posttneatment assessment (mean Hamilton
panic score=22.0, SD=8.40, range=13-34). At 3-month
follow-up, three of the original 13 patients reported one
attack in the previous week (mean Hamilton panic
score=18.0, SD=6.24, nange=11-23). This was a statis-
tically significant decrease in the number of individuals
reporting panic attacks from pretreatment to posttreat-
ment to follow-up assessment (Cochran’s Q=14.60,
df=2, p<0.001, N=20). Within this group, the individu-
als whose primary psychiatric diagnosis was panic dis-
order with on without agoraphobia also showed a sta-
tistically significant linear decrease from pretreatment
to posttreatment to follow-up (Cochran’s Q=12.67, df=2,
p<0.00s, N=13).
In both groups there was a significant decline in
Hamilton panic scores between pretreatment and post-
treatment assessments. For the subjects who reported at
least one panic attack at pretreatment assessment
(N=13), the mean pretreatment Hamilton panic score
was 24.46 (SD=8.71) and the mean posttreatment
Hamilton panic scone was 8.46 (SD=12.1S) (t=4.75,
df=12, p<O.OOl ). For the panic disorder subset (N=1 1),
the mean pretreatment Hamilton panic score was 24.64
I
II
I
.e -o a a
w 1 1’2 14 1’6#{149}1’80
ITreatment Poettreatment
I I I
-6 -4 -2 I I I I I I
10 12 14 16 18 20
‘1
MEDiTATION-BASED PROGRAM FOR ANXIETY DISORDERS
940 Am JPsychiatry 149:7, July 1992
FIGURE 1. Mean Beck An,dety Inventory Ratings Before, During, and
After Treatment of Patients in a MeditatiOn-Baud Stress ReduCtiOn
PI-uua
20
15
10
--
5
0
a#{149}Thcnumbers of subjects for successive assessments were as follows:
pretreannent, 4, 5, 7, 11, 12, 16, 21; treatment, 20, 22, 22, 21, 22,
21, 21, 19; posttreatment, 21, 19, 21.
(SD=8.30) and the mean posttreatment Hamilton panic
scone was 8.36 (SD=12.68) (t=4.07, df=10, p<0.OOS).
The mean number of panic attacks registered on the
Hamilton anxiety scale and their severity also declined
significantly between pretreatment and posttreatment
assessments in both groups, and these declines were
maintained at 3-month follow-up (data not shown).
The pretreatment and posttreatment scores of the
subjects receiving psychotropic medication did not dif-
fer significantly from those of the subjects not receiving
medication during the study. Twelve patients were talc-
ing medication for anxiety before treatment and 13 af-
ter treatment; 1 1 were taking medication at follow-up.
Two patients were able to decrease their use of medica-
flon between the posttreatment and follow-up assess-
ments, and one increased the use of medication during
the same period.
The pretreatment and posttreatment scores on the
SCL-90-R and the Medical Symptom Checklist of the
study participants in the stress reduction and relaxation
program were compared with the scores on these scales
of the nonstudy participants in the program who had
met the Initial screening criteria for the study to assess
possible biasing effects from the more intense assess-
ment protocol on the study participants. As can be seen
in table 2, the mean pretreatment and posttreatment
scores on the total Medical Symptom Checklist, the
anxiety items of the Medical Symptom Checklist, the
general severity index of the SCL-90-R, and the anxiety
subscale of the SCL-90-R for the 22 subjects in the
study were comparable to those ofthe nonstudy partici-
pants in the program. The two groups showed statisti-
cally significant and equivalent symptom reduction on
these measures over the intervention period.
FIGURE 2. Mean Beek Depression Inventory Ratings Bfore, Durir,
and Aftsrlrsatment of Patients In a Meditation-Based Sfress Reduc-
dOn Program
20
T’
U I I I
02468 w
14 I4 __
Pretries1 Tremes1
1The numbers of subjects for successive assessments were as follows:
pretreatment, 4, 5, 7, 11, 12, 17, 21; treatment, 20, 22, 21, 21, 22,
21, 22, 19; posttreatmcnt, 21, 20, 21.
When changes from before to after treatment in anxi-
ety scores and in the number of panic attacks were cx-
amined on an individual basis, 20 of the 22 study sub-
jects showed marked improvement (only one patient still
had a score over 20 on the Beck Anxiety Inventory after
treatment), making it difficult to examine predictors of
differential outcome. Consistent with this uniformity of
response, no demographic or baseline variables were sig-
nificantly predictive of outcome. Expectancy ratings also
failed to serve as a meaningful predictor of outcome.
Self-reported amount of practice (compliance) was also
not significantly correlated with any outcome measure.
Furthermore, there were no statistically significant dif-
ferences in outcome between patients with generalized
anxiety disorder and those with panic disorder with or
without agoraphobia, nor was the diagnosis of major
depressive episode associated with outcome.
Adherence to the meditation practices taught in the
stress reduction and relaxation program was assessed
at 3-month follow-up. In response to the question
“Have you been keeping up practice of the stress reduc-
tion techniques?” 91% (20 of the 22 subjects) replied
in the affirmative, with a relatively homogeneous distri-
bution between single meditation techniques and corn-
binations of methods. Eighty-four percent (N=16) of
the 19 who responded to this item were practicing for-
mally three or more times per week; 42% (N=8) were
practicing for 45 minutes or more at a time, 16% (N=3)
for between 30 and 45 minutes at a time, and 37% (N=
7) for between 15 and 30 minutes at a time. Twenty-one
subjects reported continued use of mindfulness of
breathing (an informal mindfulness practice) in their
daily lives, with 77% (N=17) using it “often” and 18%
(N=4) using it “sometimes.”
KABAT-ZINN, MASSION, KRISTELLER, El AL.
Am JPsychiatry 149:7, July 1992 941
DISCUSSION
The rate of completion of the program among the
study subjects was high (22 of 24 subjects, or 92%),
consistent with previous studies of the stress reduction
and relaxation program (18). Twenty of 22 subjects
showed marked improvement in both anxiety and de-
pression after the intervention. This improvement was
maintained at 3-month follow-up. Improvement was
observed both in patients’ self-ratings (Beck anxiety and
depression scales) and in interviewers’ ratings (Hamil-
ton anxiety and depression scales).
Of considerable importance is the statistically signifi-
cant reduction from pretreatment assessment to post-
treatment assessment in the number of subjects report-
ing one or more panic attacks, an improvement that
was maintained at follow-up. There was a statistically
nonsignificant tendency for the Hamilton panic scones
to decrease between pretreatment and follow-up, sug-
gesting that for the subjects who continued to have
panic attacks during and after the intervention, the se-
verity of those attacks declined.
Fear survey and mobility inventory scones also im-
proved significantly, but these changes began during the
pretreatment period, suggesting both an effect of the
general expectancy of participation and an effect of the
exposure to a therapeutic milieu during the evaluation
visits.
The uniformly positive response to treatment among
the subjects in this small study precluded a successful
analysis of predictors of outcome. Compliance was also
uniformly reported as moderate to high, indicative of
the subjects’ positive response to the intervention ap-
proach and the successful adoption of a range of new
behaviors, including both formal and informal medita-
tion practice.
Amajor strength of this study was the careful diag-
nostic assessment procedure we used to obtain DSM-
III-R diagnoses. Previous studies investigated the use of
meditation with normal populations or populations
identified by using only nondiagnostic criteria. Such
studies may therefore have included patients who
would not have met the DSM-III-R criteria for gener-
alized anxiety disorder on panic disorders. The results
of this study, which focused specifically on patients
with generalized anxiety disorder or panic with on with-
out agoraphobia, suggest that mindfulness meditation
used in a group format may be a useful treatment ap-
proach for these diagnostic groups.
It is also clean that the improvements in panic and
anxiety which we observed cannot be attributed solely
to participation in the study itself. This is established
by the comparison showing that the subjects who pan-
ticipated in the study and the patients in the stress re-
duction program who met the screening criteria but
were not subjected to the intensive research protocol
achieved similar reductions in anxiety scores on the
SCL-90-R and the Medical Symptom Checklist. This
comparison also demonstrates that the results obtained
with the study subjects are very likely generalizable to
TABLt 2. Pretreatment and Posttreatment Scores of Patients With
Anxiety Disorders in a Study of aMeditatiOn-Based Stress Reduction
Program Compared With Scores of Nonstudy Participants in the Pro-
gram Who Met Initial Screening Criteria for the Study
Study Nonstudy
Participants in Participants in
Program Program
Measure N Mean SD N Mean SD
Medical Symptom
Checklist 21 99
Total score
Pretreatment 32.05 13.33 30.97 11.55
Posttreatment 23.10 17.75 19.S9 12.66
Anxiety scores’
Pretreatment 16.95 0.51 15.96 4.67
Posttreamient 11.10 8.50 10.17 5.53
SCL-90-R 20 97
General severity
index score
Pretreatment 1.08 0.73 1.03 0.56
Posttreatment 0.60 0.54 0.62 0.45
Anxiety subscale score
Pretreatment 1.56 1.08 1.27 0.79
Posttreatment 0.69 0.68 0.70 0.62
aj within-group differences between pretreatment and posttreat-
ment scores were significant (p<0.05) in the two-tailed paired ttests.
None of the pretreatment scores differed significantly between study
participants and nonstudy participants in the unpaired ttests, except
on the anxiety subscale of the SCL-90-R, for which p=O.OS.
bMean number of symptoms out of the 37 identified as characteristic
of patients with anxiety disorders.
the much larger group of patients who met the initial
criteria for the study.
The strong reductions in panic symptoms and fre-
quency of panic attacks observed in this study are con-
sistent with the cognitive model of panic (30) and with
clinical outcomes from studies of panic disorder in
which well-established cognitive (31) and cognitive-be-
havioral (32) intervention approaches were used. The
meditative approach used in the stress reduction and
relaxation program shares some attributes with both
cognitive and behavioral therapeutic approaches used
to treat anxiety and panic. It also differs structurally
and theoretically from them in a number of noteworthy
respects, as has been noted in a different context by oth-
ens (33, 34).
In particular, the meditative, cognitive, and cognitive-
behavioral approaches share an emphasis on noting
sensations and thoughts without viewing them as cata-
strophic and the use of stress-inducing situations as
cues to engage in new behaviors. They also have in corn-
mon the use of homework assignments to reinforce
what was learned in the group sessions. However, the
stress reduction and relaxation program differs from
cognitive and cognitive-behavioral models in the fol-
lowing important respects.
1. Emphasis is not placed on distinguishing thoughts
as positive, negative, or faulty, as in cognitive therapy.
Rather, the emphasis is on identifying thoughts as
“just” thoughts and acknowledging the potential inac-
curacy and limits of all thought, not just thoughts that
produce anxiety. This attitude is cultivated in the pen-
MEDITATION-BASED PROGRAM FOR ANXIETY DISORDERS
942 Am JPsychiatry 149:7, July 1992
ods of formal meditation practice, as well as in informal
mindfulness practiced in the course of daily activity.
2. The formal meditation is taught as a daily disci-
pline to be practiced regularly, independent of one’s
state of anxiety. The emphasis is on meditation as a way
of being, as a way of living one’s life, and as a way to
develop alternative “generic” strategies for coping with
stress and pain, rather than as a technique for coping
with a specific problem such as panic.
3. The intervention takes place in a nonpsychiatnic
medical setting with a heterogeneous group of patients
who have a wide range of medical and psychological
problems. This is a significant departure from the model
of cognitive-behavioral therapy, which is typically pro-
vided to individuals or groups of patients with a single
disorder. Moreover, the focus of the intervention is on
the meditation practice itself rather than on a specific
disorder or diagnosis or constellation of symptoms.
4. Unlike Barlow’s cognitive-behavioral approach, in
which subjects are systematically exposed through spe-
cific induction exercises to feared internal sensations as-
sociated with panic, such as cardiovascular symptoms,
hyperventilation, dizziness, and chest muscle tightness
(35), there is no attempt at systematic desensitization
through the induction of symptoms of any kind during
the stress reduction and relaxation program. Although
stressful or anxiety-related symptoms are not intention-
ally evoked, when these experiences anise, either during
formal meditation on in the course of daily living, pa-
tients are encouraged to see them as opportunities to
engage in mindful coping strategies as an alternative to
more habitual patterns of emotional reactivity. In this
respect, the program utilizes a cognitive restructuring
that overlaps with other cognitive and cognitive-behav-
ioral approaches.
S. The observational skills cultivated through mind-
fulness training differ substantially from those devel-
oped by behavioral monitoring techniques. Participants
in the program are trained initially to develop concen-
tration (one-pointed attention) through systematic and
continued focusing on a restricted field of observation
such as breathing or propnioception. Concentration
lends stability to one’s capacity to observe fearful
thoughts and feelings in a nonreactive way. Coupled
with mindfulness, concentration gives rise to a nondis-
cursive, nonanalytical, direct experiencing of the object
of attention. This is in contrast to the external data
gathering involved in behavioral analysis of antece-
dents and consequences.
Patients who are able to identify anxious thoughts as
thoughts, rather than as “reality,” report that this alone
helps reduce their anxiety and increases their ability to
encounter anxiety-producing situations more effec-
tively. The insight that one is not one’s thoughts means
that one has a potential range of responses to a given
thought if one is able to identify it as such. This in-
creased range of options is associated with a feeling of
control. It might be hypothesized that this is a feature
of a cognitive pathway explaining the clinical observa-
lions in this study.
With regard to treatment validation, it should be
noted that the duration of meditation practice in the
weekly classes becomes incrementally longer oven the
course of the intervention. By the eighth week, most
patients sit in silence in class, with little overt movement
for periods of up to 45 minutes. This is a profound be-
havior change for most patients with panic disorder or
anxiety. Such extended periods of stillness serve as an
observable behavioral indicator of an individual’s in-
creasing ability to concentrate and achieve a degree of
calmness over the intervention period. The all-day si-
lent intensive session in the sixth week of the program,
involving over 150 patients in one large room, is also
an empirical indicator of the development of new be-
havior. It can be a substantial challenge for patients
with panic disorder to sit still for long periods of time,
attempting to observe anxious thoughts and impulses
as they anise and working with them mindfully rather
than succumbing to impulses of reactivity and panic.
Asalient limitation of this pilot study is that it did not
have a randomly selected comparison group. It also
lacked a control for concomitant treatment. However,
the group of patients receiving medication showed
symptom reduction equivalent to that of the group not
receiving any medication, a finding which suggests that
the mindfulness approach may be equally useful for pa-
tients who receive pharmacothenapy and those who do
not. As in treatment studies comparing imipramine and
alpnazolam (36, 37) and a study of nonpharmacological
therapies (6), patients with generalized anxiety disorder
and patients with panic disorder responded equally well
to the program intervention. However, the number of
patients in these two diagnostic categories was small,
and a larger, randomized study would be required to
determine whether the stress reduction and relaxation
program is equally effective in each case or in the case
of patients who are receiving pharmacotherapy corn-
pared with those who are not. A larger randomized
study would also be valuable for comparing the mind-
fulness-based intervention with other cognitive and
cognitive-behavioral therapies.
We observed parallel changes in anxiety and depres-
sion scale scores after the meditation program that were
similar to those noted by Borkovec et a!. (5). However,
the presence ofcomorbid depression in eight subjects in
our study was not associated with a statistically signifi-
cant difference in outcome, as was previously reported
(38). This result could mean that the intervention was
helpful in alleviating depressive as well as anxiety
symptoms. Alternatively, it could have been an artifact
of the small size of the study group.
In summary, this pilot study of the efficacy of training
in mindfulness meditation in the context of a group
stress reduction clinic for medical outpatients showed
statistically and clinically significant reductions in
symptoms of anxiety and depression in patients with
the three cone anxiety disorders (generalized anxiety
disorder, panic disorder, and panic disorder with ago-
raphobia) diagnosed according to the DSM-III-R cnite-
na. These changes appeared to be independent of par-
KABAI-ZINN, MASSION, KRISTELLER, El AL.
Am JPsychiatry 149:7, July 1992 943
ticipation in the research protocol and were maintained
at 3-month follow-up.
ADDENDUM
Arecently completed long-term follow-up conducted with
18 ofthe 22 subjects in this study found that after 3 years, the
3-month follow-up levels of anxiety and depression reported
here had been maintained.
REFERENCES
1. Shapiro DHJr: Meditation: Self-Regulation Strategy and Altered
States of Consciousness. New York, Aldine, 1980
2. Shapiro DH Jr: Overview: clinical and physiological comparison
of meditation with other self-control strategies. Am J Psychiatry
1982; 139:267-274
3. Shapiro DH Jr, Walsh RN (eds): Meditation: Classic and Con-
temporary Perspectives. New York, Aldine, 1984
4. GarfIeld SL, Bergin AE: Handbook ofPsychotherapy and Behav-
ior Change: An Empirical Analysis. New York, John Wiley &
Sons, 1978
S. Borkovec ID, Mathews AM, Chambers A, Ebrahimi S, Lytle R,
Nelson R: The effects of relaxation training with cognitive or
nondirective therapy and the role of relaxation-induced anxiety
in the treatment of generalized anxiety. J Consult Cliii Psychol
1987; 55:883-888
6. Borkovec TD, Matthews AM: Treatment of nonphobic anxiety
disorders: a comparison of nondirective, cognitive, and coping
desensitization therapy. J Consult Clin Psychol 1988; 56:877-
884
7. Delmonte MM: Meditation and anxiety reduction: a literature
review. Cliii Psychol Rev 1985; 5:91-102
8. lalbott JA, Hales RE, Yudofsky SC (eds): The American Psychi-
attic Press Textbook of Psychiatry. Washington, DC, American
Psychiatric Press, 1988, p 458
9. Raskin M, Bali, LR, Peeke HV: Muscle biofeedback and tran-
scendental meditation. Arch Gen Psychiatry 1980; 37:93-97
10. Benson H, Frankel FH, Apfel R, Daniels MD, Schniewind HE,
Nemiah JC, Sifneos PE, Crassweller KD, Greenwood MM,
Kotch JB, Anis PA, Rosner B: Treatment of anxiety: a compari-
son of the usefulness of self-hypnosis and a meditational relaxa-
non technique: an overview. Psychother Psychosom 1978; 30:
229-242
11. Girodo M: Yoga meditation and flooding in the treatment of
anxiety neurosis. JBehav Ther Exp Psychiatry 1974; 5:157-160
12. Kutz I, Leserman J, Dorrington C, Morrison CH, BorysenkoJZ,
Benson H: Meditation as an adjunct to psychotherapy: an out-
come study. Psychother Psychosom 1985; 43:209-218
13. Thera N: The Heart of Buddhist Meditation. New York, Samuel
Weiser, 1962, p4
14. Kabat-Zinn J: Full Catastrophe Living. New York, Delacorte,
1990
15. Coleman DJ, Schwartz GE: Meditation as an intervention in
stress reactivity. J Consult Clin Psychol 1976; 44:456-466
16. Coleman D: The Meditative Mind. Los Angeles, Tardier, 1988
17. Kabat-ZinnJ: An outpatient program in behavioralmedicine for
chronic pain patients based on the practice of mindfulness mcdi-
tation: theoretical considerations and preliminary results. Gen
Hosp Psychiatry 1982; 4:33-42
18. Kabat-Zinn J, Chapman-Waldrop A: Compliance with an out-
patient stress reduction program: rates and predictors of pro-
gram completion. JBehav Med 1988; 11:333-352
19. Kabat-Zinn J, Lipworth L, Burney R: The clinical use of mind-
fulness meditation for the self-regulation ofchronic pain.J Behav
Med 1985; 8:163-190
20. Kabat-Zinn J, Lipworth L, Bumey R, Sellers W: Four-year fol-
low-up of a meditation-based program for the self-regulation of
chronic pain: treatment outcomes and compliance. Cliii JPain
1986; 2:159-173
21. Derogatis LR: SCL-90-R Manual II. Towson, Md, Clinical Psy-
chometric Research, 1983
22. Spitzer RL, Williams JBW, Gibbon M: Structured Clinical Inter-
view for DSM-ffl-R (SCm). N York, New York State Psychi-
atric Institute, Biometrics Research, 1987
23. Beck Al, Ward U-I, Mendelson M, Mock J, Erbaugh J: An in-
ventory for measuring depression. Arch Gen Psychiatry 1961;
4:561-571
24. HamiltonM: The assessmcntofanxiety states byrating. BrJ Med
Psychol 1959; 32:50-55
25. DiNardo PA, Barlow DH: Anxiety Disorders Interview Schedule-
Revised (ADIS-R). Albany, NY, Graywind Publications, 1988
26. Williams JBW: A structured interview guide for the Hamilton
Depression Rating Scale. Arch Gen Psychiatry 1988; 45:742-
747
27. Arrinddll WA, van der Ende J: Further evidence for cross-sample
invariance of phobic factors: psychiatric inpatient ratings on the
Fear Survey Schedule-ifi. Behav Res Ther 1986; 24:289-297
28. Chambless DL, Caputo CG, Jasin SE, Gracely EJ, Williams C:
The Mobility Inventory for Agoraphobia. Behav Res Ther 1985;
23:35-44
29. Cohen J, Cohen I: Applied Multiple Regression/Correlation:
Analysis for the Behavioral Sciences, 2nd ed. Hillsdale, NJ, Law-
rence Erlbaum Associates, 1990, pp 413-423
30. Clark DM: A cognitive approach to panic. Behav Res Ther 1986;
24:461-470
31. Clark DM, Salkovskis PM, Chaildey AJ: Respiratory control as
atreatment for panic attacks. JBehav Ther Exp Psychiatry 1985;
16:23-30
32. Barlow DH, Craske MG, Cerny JA, Klosko JS: Behavioral treat-
ment of panic disorder. Behavior Therapy 1989; 20:261-282
33. Shapiro DH Jr, Zifferblatt SM: Zen meditation and behavior
self-control: similarities, differences, and clinical applications.
Am Psychol 1976; 31:519-532
34. Walsh R: The consciousness disciplines and the behavioral sci-
ences: questions of comparison and assessment. Am J Psychiatry
1980; 137:663-673
35. Barlow DH: Long-term outcome for patients with panic disorder
treated with cognitive-behavioral therapy. JClin Psychiatry
1990; S1(suppl A):17-23
36. Charney DS, Woods SW, Goodman WK, Rifkin B, lUnch M,
Aiken B, Quadrino LM, Heninger GR: Drug treatment of panic
disorder: the comparative efficacy of imipramine, alprazolam,
and trazodone. JCliii Psychiatry 1986; 47:580-586
37. Gershon 5, Eison AS: The ideal anxiolytic. Psychiatr Annals
1987; 17:156-170
38. VanValkenburg C, Akiskal HS, Puzantian V, Rosenthal I: Anx-
ious depressions: clinical, family history, and naturalistic out-
come: comparisons with panic and major depressive disorders. J
Affective Disord 1984; 6:67-82