Content uploaded by Joseph Spinazzola
Author content
All content in this area was uploaded by Joseph Spinazzola on Jul 28, 2016
Content may be subject to copyright.
Original Articles
Yoga for Adult Women with Chronic PTSD:
A Long-Term Follow-Up Study
Alison Rhodes, PhD,
1,2
Joseph Spinazzola, PhD,
1
and Bessel van der Kolk, MD
1
Abstract
Introduction: Yoga—the integrative practice of physical postures and movement, breath exercises, and
mindfulness—may serve as a useful adjunctive component of trauma-focused treatment to build skills in
tolerating and modulating physiologic and affective states that have become dysregulated by trauma exposure.
A previous randomized controlled study was carried out among 60 women with chronic, treatment-resistant
post-traumatic stress disorder (PTSD) and associated mental health problems stemming from prolonged or
multiple trauma exposures. After 10 sessions of yoga, participants exhibited statistically significant decreases in
PTSD symptom severity and greater likelihood of loss of PTSD diagnosis, significant decreases in engagement
in negative tension reduction activities (e.g., self-injury), and greater reductions in dissociative and depressive
symptoms when compared with the control (a seminar in women’s health). The current study is a long-term
follow-up assessment of participants who completed this randomized controlled trial.
Methods: Participants from the randomized controlled trial were invited to participate in long-term follow-up
assessments approximately 1.5 years after study completion to assess whether the initial intervention and/or yoga
practice after treatment was associated with additional changes. Forty-nine women completed the long-term follow-
up interviews. Hierarchical regression analysis was used to examine whether treatment group status in the original
study and frequency of yoga practice after the study predicted greater changes in symptoms and PTSD diagnosis.
Results: Group assignment in the original randomized study was not a significant predictor of longer-term
outcomes. However, frequency of continuing yoga practice significantly predicted greater decreases in PTSD
symptom severity and depression symptom severity, as well as a greater likelihood of a loss of PTSD diagnosis.
Conclusions: Yoga appears to be a useful treatment modality; the greatest long-term benefits are derived from
more frequent yoga practice.
Introduction
Post-traumatic stress disorder (PTSD) is a common
and chronic condition affecting women at a rate over
double that in men (rates of 11.7% and 4.0%, respectively).
1
Women are more likely than men to be exposed to recurring
interpersonal violence perpetrated by intimates in childhood
and adulthood, such as domestic violence and childhood sex-
ual abuse.
2
PTSD associated with recurring trauma exposure is
especially challenging to treat given that this population often
faces significant problems beyond PTSD.
3–5
Indeed, research
consistently links repeated trauma exposure with a complex
constellation of symptoms, including impulsive or aggressive
behavior, self-injurious behaviors, anxious arousal, mood
disturbances, and dissociative symptoms.
3,6,7
At the core of
these problems are self-regulatory deficits.
3
Difficulties with affect and impulse regulation pose sig-
nificant challenges for survivors of repeated trauma to tol-
erate traditional treatment modalities (e.g., cognitive and
exposure-based treatments), as evidenced by high rates of
drop-out from treatment, PTSD symptom exacerbation
during treatment, and worsening symptoms following
treatment.
8–13
Essential to the resolution of PTSD and re-
lated mental health problems is the extinction of the con-
ditioned fear response. This requires that survivors of
recurring trauma learn to stay oriented in the present mo-
ment and manage powerful emotions and impulsive reac-
tions that arise in the context of trauma reminders.
14
Mind-body–oriented therapies such as yoga—the integra-
tive practice of physical postures and movement, breath ex-
ercises, and mindfulness—improve self-regulation and the
ability to stay focused on present experiences.
15–17
Therefore,
1
The Trauma Center at Justice Resource Institute, Brookline, MA.
2
Tufts University Counseling and Mental Health Services, Medford, MA.
THE JOURNAL OF ALTERNATIVE AND COMPLEMENTARY MEDICINE
Volume 22, Number 3, 2016, pp. 189–196
ªMary Ann Liebert, Inc.
DOI: 10.1089/acm.2014.0407
189
they may be a useful component of treatment in survivors of
recurring interpersonal trauma.
18,19
Moreover, the direct ef-
fect of yoga practice on the stress response system (including
downregulation of the hypothalamus-pituitary-adrenal axis
and the sympathetic nervous system
15
) could reduce PTSD
symptoms and associated problems stemming from pro-
longed traumatic exposure. In fact, many studies have shown
that yoga may be effective in the treatment of a range of
mental health problems, including PTSD, anxiety, and de-
pression.
20–23
Furthermore, yoga is a popular activity among
women in the general population, making it a feasible and
accessible treatment choice.
24
Within this area of study, van
der Kolk and colleagues carried out a randomized controlled
trial (RCT)
23
assessing the effectiveness of 10 weeks of yoga
classes (one per week) for women with chronic, treatment-
unresponsive PTSD,* all with histories of exposure to pro-
longed interpersonal trauma (e.g., physical abuse, sexual
abuse, neglect, domestic violence). They found that yoga was
significantly more effective than an attentional control (a
seminar on women’s health) for reducing symptom severity
of PTSD and the likelihood of meeting diagnostic criteria for
PTSD. The yoga group also showed greater decreases than
the control group in symptoms of depression, dissociation,
and engagement in negative tension-reducing activities (e.g.,
self-injury).
The current study is a long-term follow-up among the
women who completed van der Kolk and colleagues’
RCT.
23
The goal was to examine whether there are greater
changes in long-term mental health outcomes (i.e., PTSD
diagnosis and symptom severity, depression symptom se-
verity, dissociative symptom severity, and engagement in
tension-reducing activities) for yoga participants versus at-
tentional controls. Additionally, the present study tested the
hypothesis that frequency of yoga practice after the study
will significantly influence long-term outcomes.
Materials and Methods
Participants and procedure
Beginning in 2008, six cohorts completed the yoga RCT
23
during the course of 3 years: three yoga groups (n=31) and
three control groups (n=29). Details on recruitment for the
original study and the interventions used are described in
detail elsewhere.
23
After completing the yoga classes or the
women’s health seminar, participants were able to join the
following cohort in the opposite intervention. For the cur-
rent study, after institutional review board approval, at-
tempts were made to contact all participants who completed
the RCT to invite them to come in to complete the same
measures used in the RCT and an additional self-report
measure on the frequency of yoga practice since they were
last interviewed and any treatment changes that may have
occurred.
Forty-nine of the 60 original participants completed long-
term follow-ups. Independent ttests were run to assess
whether there were any significant differences between
those who came in for the long-term follow-up and those
who were lost to follow-up in terms of demographic vari-
ables and baseline measures of severity of PTSD symptoms,
depression symptoms, dissociative symptoms, and problems
with tension-reducing activities. No significant differences
were found between the two groups on these measures.
Table 1 shows demographic characteristics of participants
who completed the long-term follow-up.
Participants in the long-term follow up included 26 wo-
men from the original study initially randomly assigned to
the yoga intervention, 16 assigned to the attentional control
group (women’s health seminar), and 7 assigned to the
control group, who subsequently elected to enroll in a
nonrandomized ‘‘second round’’ of the yoga intervention
following completion of the attentional control protocol.
Long-term follow-up interviews were carried out over 6
months. Depending on which cohort the participants were
part of and when they were able to come in for the long-term
assessment, there was variability in how long the partici-
pants had been out of the RCT at the follow-up assessment
(ranging from 0.75 to 2.75 years). Participants were asked
not to disclose what treatment group they were in during the
RCT or whether they had practiced yoga after treatment
until other measures were completed so that the interviewer
could remain blind during assessment.
Measures
Clinician Administered PTSD Scale (CAPS). The CAPS
is a structured interview assessing the frequency and intensity
of the 17 symptoms of PTSD in the Diagnostic and Statistical
Manual of Mental Disorders, fourth edition, as experienced in
the past month. Each symptom is rated 0–4 for frequency and
intensity separately. The scale demonstrates high inter-rater
reliability and high internal consistency and correlates highly
with other measures of PTSD symptoms.
25
The CAPS F1/I2/
Sev65 scoring rules were used to determine whether the
participant qualified for PTSD diagnosis at the time of the
long-term follow-up.
26
Total CAPS scores at baseline and at
the long-term follow-up were used to calculate changes in
PTSD symptom severity.
Dissociative Experiences Scale (DES). The DES is a 28-
item self-report instrument measuring a variety of dissociative
experiences, such as ‘‘highway hypnosis’’ and derealization.
For each item, participants note how often this happens to them
on a scale of 10-point increments ranging from 0 (never) to 100
Table 1. Demographic Characteristics
Variable Data
Race: white (%) 71.4
Marital status: single (%) 46.9
College graduate (%) 71.4
Annual income <$40,000 (%) 46.9
Employment: full-time (%) 38.8
Mean age –SD (y) 42.8 –11.8
SD, standard deviation.
*PTSD was established in a screening assessment utilizing the
Clinician Administered PTSD Scale and the CAPS F1/I2/Sev65
scoring rules.
26
Chronicity was based on meeting criteria for PTSD
in relation to an index trauma that occurred at least 12 years before
intake. Treatment unresponsiveness was determined by participants
having had at least 3 years of prior psychotherapy treatment focused
on trauma issues.
190 RHODES ET AL.
(always). The DES score is calculated as a percentage. The
measure shows good test-retest reliability, high internal con-
sistency, and excellent construct validity.
27
Beck Depression Inventory (BDI). The BDI is a 21-item
self-report measure assessing the severity of a range of de-
pressive symptoms. Each item has an answer choice ranging
from 0 to 3, on a spectrum of less severe to more severe. The
measure is scored as the sum of the answers, with higher
scores indicating greater depression severity. This measure
has high internal reliability, construct validity, and test-
retest reliability.
28
Inventory of Altered Self Capacities Tension Reduction
Activities (IASC-TRA) Subscale. The IASC is a 63-item
self-report measure assessing psychological functioning
along seven subscales, including the TRA subscale, which
identifies the tendency to respond to internal stress with
problematic externalizing behaviors that distract, soothe, or
reduce negative internal experiences (i.e., self-injury, sexual
activity, food binging). This subscale demonstrates good
internal consistency, reliability, and validity.
29
Stressful Life Events Screening Questionnaire (SLESQ).
Research shows an association between the number of
trauma types experienced and symptom complexity and
severity.
30,31
Therefore, the current study controlled for
cumulative trauma exposure. The current study used a total
score of event exposure from the SLESQ, which is a 13-item
clinician-administered questionnaire assessing for lifetime
exposure to various traumatic events (e.g., life-threatening
accident, physical or sexual abuse, witnessing another per-
son being killed or assaulted). The measure demonstrates
good test-retest reliability and adequate convergent validity.
32
Data analytic approach
Dependent variables for all continuous measures were
calculated as a measure of change from participants’ scores at
baseline to participants’ scores at the long-term follow-
up. Because all participants met diagnostic criteria for PTSD
at baseline to qualify for the RCT, this measure was assessed
as a dichotomous variable (1 =met PTSD criteria at long-term
follow-up; 0 =did not meet PTSD criteria at long-term
follow-up). Table 2 shows descriptive data for the measures.
The seven participants randomly assigned to the atten-
tional control condition who elected to enroll in a non-
randomized second round of yoga intervention following
completion of the control group were compared with the
cohort of randomly assigned (‘‘first round’’) yoga partici-
pants on long-term study outcomes. This comparison sought
to assess whether the former subset of yoga participants
were differentially affected by study procedures and con-
sequently performed differently on study outcomes than
first-round yoga participants. Statistical analyses comparing
mean change scores on all study outcomes measures re-
vealed no significant differences between first- and second-
round yoga participants. Accordingly, all yoga participants
were collapsed into a single group to maximize power in
regression analyses testing primary study hypotheses. Hen-
ceforth, participants who completed the yoga protocol as
part of the RCT (both first- and second-round participants)
are referred to as being in the yoga group. Those who
completed only the women’s health seminar during the RCT
and did not participate in yoga during the RCT, even if they
chose to pursue the practice of yoga after completion of the
study, are referred to as being in the control group.
A questionnaire was administered to gather information
on potential treatment changes since the RCT ended, in-
cluding questions on whether the participant began seeing a
new therapist, attended a support group, started new medi-
cation, or received any other new type of treatment or body
work (e.g., acupuncture, massage, neurofeedback, eye
movement desensitization and reprocessing). If the partici-
pant answered yes to any the questions, the variable was
coded as 1. No treatment changes were coded as 0.
Participants self-reported the number of times they prac-
ticed yoga after the study. Because frequency of yoga practice
Table 2. Descriptive Data for Study Measures
Variable
Participants
(n)orn(%) Mean –SD Range
SLESQ score 49 8.4 –2.3 2–15
Time since RCT (y) 49 82.7 –26 39–143
Estimated no. of times
participant practiced
yoga after RCT
49 41.1 –57.2 0–275
Frequency yoga
practice after RCT
49 1.9 –.61 1–3
CAPS score
Baseline CAPS
(PTSD)
49 73.8 –12.9 52–101
Long-term follow-up
CAPS
49 50.8 –24.6 11–105
Change in CAPS 49 23** –21 -17 to 59
DES score
Baseline DES
(dissociation)
48 16.9 –12.1 1.4–60.4
Long-term follow-up
DES
48 15.5 –12.2 2.1–46.1
Change in DES 47 1.4 –11.4 -35.7 to 20.4
BDI score
Baseline BDI 47 23 –11.6 6–51
Long-term follow-up
BDI
48 14.9 –11.8 0–46
Change in BDI 48 8.6
a
–11.6 -13 to 39
IASC-TRA score
Baseline IASC-TRA 44 16.2 –4.1 10–29
Long-term follow-up
IASC-TRA
49 13.2 –3.5 9–24
Change in IASC-TRA 44 3.0
a
–3.5 -4to9
Treatment changes 49 (100)
Yes 29 (59)
No 20 (41)
Participated in yoga
group during RCT
49 (100)
Yes 33 (67)
No 16 (33)
Long-term follow-up
PTSD diagnosis
present
49 (100)
Yes 21 (43)
No 28 (57)
a
p<0.001.
SLESQ, Stressful Life Events Screening Questionnaire; RCT,
randomized controlled trial; CAPS, Clinician Administered PTSD
Scale; PTSD, post-traumatic stress disorder; DES, Dissociative
Experiences Scale; BDI, Beck Depression Inventory; IASC-TRA,
Inventory of Altered Self Capacities Tension Reduction Activities.
YOGA FOR ADULT WOMEN WITH CHRONIC PTSD 191
after the study was partially dependent on how long the par-
ticipant had been out of the original study (ranging from 39 to
143 weeks), a ratio variable was created (number of times the
participant practiced yoga divided by the number of weeks that
had passed since study completion). The variable was posi-
tively skewed, and it was recoded into an ordinal variable
(1 =no practice after the study; 2 =practiced with a frequency
less than an average of one time per week after the study;
3=practiced more than one timeper week on average after the
study). This new ordinal variable was normally distributed.
Bivariate correlation analyses among study variables were
calculated (Table 3). Hierarchical linear regression analysis
was used to examine whether treatment group status in the
original RCT and frequency of yoga practice after the study
predicted changes in PTSD symptoms, depressive symptoms,
dissociative symptoms, or tension-reducing activities from
baseline measures to the time of the long-term follow-up
assessment while controlling for extent of trauma exposure
and other treatment changes. Variables were entered in blocks
in the following order: (1) lifetime exposure to traumatic
events (SLESQ), (2) whether the participant had any treat-
ment changes since the study ended, (3) treatment group
status in the original study (i.e., yoga or control only), and (4)
frequency of yoga practice after the study. By entering var-
iables in the same blocks, hierarchical logistic regression was
used to test whether group status and frequency of yoga
practice after the treatment predicted PTSD diagnosis.
26
A small amount of missing data appeared at random.
Missing data were addressed in the following manner. If an
entire measure was missing for a given participant or if more
than five items within a given measure were missing, no data
were imputed for that participant and the participant’s re-
sponses were not included in the analysis for that specific
outcome measure. For participants for whom fewer than five
items were missing, missing values for the imputed item were
imputed by carrying forward the last observation for that item.
The approach taken was considered the most conservative way
to impute missing data but could have underestimated potential
change (SPSS Statistics for Windows, Version 21.0. IBM
Corp. Released 2012. Armonk, NY: IBM Corp.).
Results
Long-term yoga practice descriptives
Following completion of the original study, the majority
(n=39) of participants continued or began to practice yoga.
Twenty-three of 26 women randomly assigned to the yoga
treatment group continued to practice yoga. Nine of 16
women who completed only the attentional control condi-
tion went on to practice yoga, and all 7 participants who
elected to enroll in the yoga intervention following ran-
domization to the control condition practiced yoga after the
study. Participants’ yoga practices ranged in frequency from
a few practices up through approximately 275 yoga sessions.
They practiced in a range of settings, including at home
using a yoga DVD or doing self-guided practices based on
what they had learned in classes, and in yoga classes at
gyms and yoga studios.
Bivariate analyses
As indicated in Table 3, bivariate correlation analyses
showed that group status in the original study was not sig-
nificantly associated with any of the outcome variables. Bi-
variate analysis did indicate, however, that a greater
frequency of yoga practice after the study was significantly
associated with loss of PTSD diagnosis (r=-0.283; p<0.05)
and with decreases in depression symptom severity (r=0.348;
p<0.05).
Regression analyses
Hierarchical regression analysis yielded three significant
models: decreases in PTSD symptom severity, likelihood of
loss of PTSD diagnosis, and decreases in depression
symptom severity. No predictors were significant in the
models predicting dissociation or tension-reducing activi-
ties. Table 4 shows results of hierarchical regression anal-
ysis for changes in PTSD symptom severity from baseline to
the long-term follow-up interview. None of the variables
entered in the first three steps were significant predictors,
but frequency of yoga practice after the study significantly
added to the model (change in R
2
was 0.11; p<0.05).
Greater frequency of yoga practice was associated with
greater decreases in PTSD symptom severity from baseline
to the long-term follow-up (b=12.24; p<0.05). Additionally,
when frequency of yoga practice was added to the model,
study group status became significant in the direction of
non-yoga attentional control participations exhibiting
greater long-term decreases in PTSD symptom severity.
Group differences in post-study frequency of yoga prac-
tice were examined post hoc by using independent-sample
ttests. The ttests were used to compare total practice after
Table 3. Correlation Matrix of Yoga Practice, Group Status, Treatment Changes, and Trauma Symptoms
Variable 1 2 3 4 5 6 7 8
1. SLESQ 1
2. Treatment changes -0.042 1
3. RCT group status -0.086 -0.136 1
4. Frequency yoga after RCT 0.147 -0.056 0.313
a
5. Change in CAPS -0.109 -0.205 -0.169 0.236 1
6. Change in DES -0.209 -0.068 -0.111 -0.019 0.258 1
7. Change in BDI 0.014 -0.234 -0.050 0.348
a
0.455
b
0.085 1
8. Change in IASC-TRA 0.047 -0.440
b
0.030 0.021 0.192 0.480
b
0.453
b
1
9. LTFU PTSD diagnosis present 0.134 0.216 -0.013 -0.283
a
-0.624
b
-0.217 -0.208 -0.149
a
p<0.05.
b
p<0.01.
LTFU, measure at long-term follow-up interview.
192 RHODES ET AL.
treatment, as well as to compare the ratio variable that was
created as a measure of the average frequency of practice
after the RCT that took into account the variability in time
passed after treatment. No significant difference was found
between yoga and control groups on total frequency of post-
study yoga practice (yoga: mean –standard deviation,
47.18 –58.71; control: 28.53 –53.59; t=1.07; p=0.29). A
nonsignificant trend was observed in the direction of higher
frequency of post-study yoga practice in the yoga group
when time elapsed was taken into account (yoga: 2.09 –
0.52; control: 1.69 –0.70; t=2.04; p=0.053).
Table 5 shows the results of hierarchical logistic regres-
sion for predictors of PTSD diagnosis. None of the variables
significantly predicted PTSD diagnosis in the first three
steps. However, when frequency of yoga practice was added
to the model, the model was significant (chi-square
(4) =9.87; p<0.05). Those who practiced yoga more fre-
quently were less likely to meet diagnostic criteria for PTSD
(b=-1.579; odds ratio, 0.21; p<0.05).
Table 6 shows the results of hierarchical regression
analysis for changes in depression symptom severity. The
extent of the participant’s trauma history, treatment changes,
and group status in the RCT were not significant predictors
of changes in depression symptom severity from baseline
to the long-term follow-up. However, frequency of yoga
practice after the RCT significantly added to the model
(change in R
2
, 0.15; p<0.01), with greater frequency of
practice associated with a greater decreases in depression
symptom severity (b=7.84; p<0.01).
Discussion
The present study further supports the utility of yoga as
a component of intervention for women with histories of
chronic, treatment-resistant PTSD associated with extensive
histories of exposure to interpersonal victimization. Speci-
fically, findings from this study suggest that more frequent
yoga practice over extended periods may augment and
sustain decreases in symptoms of both PTSD and depres-
sion. These findings are particularly promising given that
the study population reported persistent mental health
problems related to traumatic stressors despite having been
in trauma-focused psychotherapy for at least 3 years. Many
participants had little relief from their symptoms before
engagement in yoga practice during or following study
participation.
Participation in the yoga intervention condition during the
original study did not have an independent effect on long-term
outcomes. This may perhaps be explained by an unexpected
byproduct of the study. Namely, most study participants, ir-
respective of group assignment, elected to continue or initiate
yoga practice following cessation of the study, making it
Table 4. Hierarchical Linear Regression Analysis: Predictors of Changes
in PTSD Symptom Severity (CAPS)
Variable b bSEB p-Value
Step 1
(Constant) 29.43 9.15 0.002
SLESQ -0.766 -0.109 1.022 0.457
R
2
=0.01
R
2
adj
=-0.01
F(1/47) =.56, p=0.46
Step 2
(Constant) 35.23 9.87 0.001
SLESQ -0.829 1.010 -0.118 0.416
Treatment changes -8.898 6.077 -0.210 0.150
R
2
=0.06
R
2
adj
=0.02
F(2/46) =1.4, p=0.27
Step 3
(Constant) 43.50 11.22 0.000
SLESQ -0.966 -0.137 1.002 0.340
Treatment changes -10.155 -0.240 6.059 0.101
RCT group status -9.449 -0.213 6.368 0.145
R
2
=0.1
R
2
adj
=0.04
F(3/45) =1.7, p=0.19
Step 4
(Constant) 26.66 12.62 0.040
SLESQ -1.41 -0.199 0.966 0.153
Treatment changes -10.09 -0.238 5.74 0.086
RCT group status -14.61 -0.329 6.38 0.027
Frequency yoga practice after RCT 12.24 0.355 4.94 0.017
R
2
=0.21
R
2
adj
=0.14
F(4/44) =2.92, p<0.05
SEB, Standard Error of Beta.
YOGA FOR ADULT WOMEN WITH CHRONIC PTSD 193
Table 5. Hierarchical Logistic Regression Analysis: Predictors of PTSD Diagnosis (CAPS)
Variable b SEB OR p-Value
Step 1
(Constant) -1.072 0.899 0.342 0.233
SLESQ 0.092 0.099 1.097 0.354
Chi-square =0.878, df =1, p=0.35
Step 2
(Constant) -1.778 1.058 0.169 0.093
SLESQ 0.106 0.104 1.112 0.310
Treatment changes 0.967 0.624 2.630 0.121
Chi-square(2) =3.39, p=0.183
Step 3
(Constant) -1.896 1.200 0.150 0.114
SLESQ 0.108 0.105 1.114 0.303
Treatment changes 0.985 0.631 2.678 0.118
RCT group status 0.135 0.643 1.145 0.834
Chi-square(3) =3.44, p=0.329
Step 4
(Constant) -0.327 1.426 0.721 0.819
SLESQ 0.200 0.126 1.222 0.113
Treatment changes 1.227 0.702 3.409 0.081
RCT group status 0.912 0.789 2.488 0.248
Frequency yoga practice after RCT -1.579 0.695 0.206 0.023
Chi-square(4) =9.87, p <0.05
OR, odds ratio.
Table 6. Hierarchical Regression Analysis: Predictors of Depression Symptom Severity (BDI)
Variable b bSEB p-Value
Step 1
(Constant) 8.164 5.088 0.115
SLESQ 0.055 0.014 0.571 0.924
R
2
=0.00
R
2
adj
=-0.02
F(1/46) =0.009, p=0.92
Step 2
(Constant) 11.782 5.485 0.037
SLESQ 0.002 0.001 0.563 0.997
Treatment changes -5.442 -0.234 3.383 0.115
R
2
=0.06
R
2
adj
=0.01
F(2/45) =1.3, p=0.28
Step 3
(Constant) 13.705 6.411 0.038
SLESQ -0.035 -0.009 0.570 0.952
Treatment changes -5.761 -0.247 3.450 0.102
RCT group status -2.141 -0.088 3.620 0.557
R
2
=0.06
R
2
adj
=-0.002
F(3/44) =0.97, p=0.42
Step 4
(Constant) 3.058 6.999 0.664
SLESQ -0.326 -0.084 0.538 0.547
Treatment changes -5.784 -0.248 3.196 0.077
RCT group status -5.512 -0.226 3.553 0.128
Frequency yoga practice after RCT 7.841 0.417 2.727 0.006
R
2
=0.21
R
2
adj
=0.14
F(4/43) =2.9, p<0.05
194
difficult to isolate effects uniquely attributable to the initial
randomization to the yoga intervention group.
Interestingly, group status was significant in predicting
decreases in PTSD symptom severity, but only when fre-
quency of yoga practice after the RCT was factored into the
model. In this case, those in the control group showed
greater long-term decreases in PTSD symptom severity than
those who participated in the active treatment condition.
This could indicate some interaction between treatment
group status and continuing yoga practice; however, small
sample size constraints prohibited true tests of interaction
effects within the regression models. Results suggested that
the treatment group went on to practice marginally more
frequently than did the control group. Findings here are
equivocal. They could indicate that the shift from no yoga
practice in the control condition to some post-study practice
was associated with a greater influence on long-term symp-
tom reduction than was accomplished by continued post-
study yoga practice by participants who received the active
yoga intervention. Another possible explanation was that
there was some added benefit of completing the control
condition and then going on to practice yoga. Further research
would be needed to ascertain the meaning of this finding. The
extent of trauma exposure and treatment changes did not
significantly predict changes in symptoms from baseline to
the long-term follow-up. However, the measures used may
not have been robust enough to detect change. Although the
number of traumatic events a person is exposed predicts
symptom complexity and frequency, other factors, such as
who perpetrated violence or the type of trauma that the in-
dividual was exposed to, may also be relevant.
30,31
Given the
small sample size in the current study, and the associated
need to limit the number of control variables that were in-
cluded in the analysis, it was impossible to include a more
complex representation of trauma history. Similarly, some
treatment changes may have had a greater effect on symp-
toms than others, but this level of complexity was not pos-
sible to control for in the current study.
Another limitation in the current study is that several mea-
sures were based on self-report. There may have been some
error in participants’ reports of frequency of their yoga prac-
tices. Altering this variable into an ordinal measure that con-
sidered frequency as an estimation of average practice over
time likely helped address some inflation or underestimation of
reporting that may have occurred. However, additional re-
search that more systematically monitors the frequency of
participants’ yoga practices over time is warranted.
The current study raises questions about why a greater
frequency of yoga practice helps improve PTSD and depres-
sion, but not dissociation or engagement in tension-reducing
activities. It is unclear whether yoga is most beneficial in ad-
dressing certain types of symptoms or if other factors may
account for this finding. Perhaps individuals who struggled
more with these particular problems were less capable of en-
gaging in an ongoing way with yoga. This would be consistent
with prior studies documenting higher treatment dropout rates
among trauma survivors with more severe affect dysregula-
tion.
8
Additional research exploring survivors’ experiences of
practicing yoga and reasons for sustaining or ending their
engagement with yoga could help shed light on this.
It remains unclear precisely how much yoga practice is
needed to achieve or maintain benefits. More systematic,
controlled research should seek to ascertain the necessary
‘‘dosage’’ of practice to achieve improved functioning. Ad-
ditionally, this study did not consider the variations in lengths
of practice sessions, and it is unclear whether the length of
practice matters. It would also be useful to consider whether
the optimal dosage varies with the extent or nature of the
participants’ trauma history or posttraumatic symptoms and
conditions. Finally, this study was unable to examine whether
practice setting or style of yoga practice (e.g., vigorous versus
gentle) influences mental health outcomes; it is possible, for
example, that certain types of yoga or even certain postures
may be more beneficial for alleviating particular symptoms or
conditions than others.
While additional research is needed, the current study offers
strong support for yoga practice asa complement to therapeutic
intervention for women struggling with the legacy of pro-
longed interpersonal trauma. Moreover, it contributes mean-
ingfully to the limited empirical knowledge base onthe longer-
term mental health effects of sustained yoga practice. Given the
low cost of yoga, its wide popularity,
33
and relative ease of
access to community-based classes, exploring opportunities
with trauma survivors to add yoga to their overall arsenal of
healing and wellness practices would seem to be a compelling
avenue for therapeutic attention. If yoga practice is ultimately
undertaken by trauma survivors in a measured and sustained
manner, it may offer potentially substantial benefits.
Acknowledgments
The authors thank Kelley Durham for her assistance in
data collection in this study.
Author Disclosure Statement
No competing financial interests exist.
References
1. Kessler RC, Petukhova M, Sampson NA, et al. Twelve-
month and lifetime prevalence and lifetime morbid risk of
anxiety and mood disorders in the United States. Int J
Methods Psychiatr Res 2012;21:169–184.
2. Coid J, Petruckevitch A, Feder G, et al. Relation between
childhood sexual and physical abuse and risk of re-
victimisation in women: a cross-sectional survey. Lancet
2001;358:450–445.
3. Cloitre M, Stolbach BC, Herman JL, et al. Developmental
approach to complex PTSD: childhood and adult cumula-
tive trauma as predictors of symptom complexity. J Trauma
Stress 2009;22:399–408.
4. Courtois CA. Complex trauma, complex reactions: assess-
ment and treatment. Psychother Theor Res Pract Train 2004;
41:412–425.
5. van der Kolk BA. The body keeps the score: memory and
the emerging psychobiology of posttraumatic stress. Harv
Rev Psychiatry 1994;1:253–265.
6. Pelcovitz D, van der Kolk BA, Roth S, et al. Development
of a criteria set and a structures interview for disorders of
extreme stress. J Trauma Stress 1997;10:3–17.
7. van der Kolk BA, Roth S, Pelcovitz D, et al. Disorders of
extreme stress: the empirical foundation of a complex ad-
aptation to trauma. J Trauma Stress 2005;18:389–399.
8. Ford JD, Kidd P. Early childhood trauma and disorders of
extreme stress as predictors of treatment outcome with
chronic PTSD. J Trauma Stress 1998;18:743–761.
YOGA FOR ADULT WOMEN WITH CHRONIC PTSD 195
9. Ladwig KH, Marten-Mittag B, Deisenhofer I, et al. Psy-
chophysiological correlates of peritraumatic dissociative
responses in survivors of life-threatening cardiac events.
Psychopathology 2002;35:241–248.
10. McDonagh-Coyle A, Friedman M, McHugo G, et al. Rando-
mized trial of cognitive-behavioral therapy for chronic post-
traumatic stress disorder in adult female survivors of childhood
sexual abuse. J Consult Clin Psychol 2005;73:515–524.
11. Michelson L, June K, Vives A, et al. The role of trauma and
dissociation in cognitive-behavioral psychotherapy out-
come and maintenance for panic disorder with agoraphobia.
Behav Res Ther 1998;36:1011–1050.
12. Tarrier N, Pilgrim H., Sommerfield C, et al. A randomized
trial of cognitive therapy and imaginal exposure in the
treatment of chronic posttraumatic stress disorder. J Con-
sult Clin Psychol 1999;67:13–18.
13. Zayfert C, Deviva JC, Becker CB, et al. Exposure utiliza-
tion and completion of cognitive behavioral therapy for
PTSD in a ‘‘real world’’ clinical practice. J Trauma Stress
2005;18:637–645.
14. Jaycox LH, Foa EB. Obstacles in implementing exposure
therapy for PTSD: case discussions and practical solutions.
Clin Psychol Psychother 1996;3:176–184.
15. Ross A, Thomas S. The health benefits of yoga and exer-
cise: a review of comparison studies. J Altern Complement
Med 2010;16:3–12.
16. Salmon P, Lush E, Jablonski M, Sephton SE. Yoga and
mindfulness: clinical aspects of an ancient mind/body
practice. Cogn Behav Pract 2009;16:59–72.
17. Evans S, Tsao JCI, Sternlieb B, Zeltzer LK. Using the
biopsychosocial model to understand the health benefits of
Yoga. J Complement Integr Med 2009;6:1–22.
18. Telles S, Nilkamal S, Balkrishna A. Managing mental
health disorders resulting from trauma through yoga: a re-
view. Depress Res Treat 2012;401513:1–9.
19. van der Kolk BA. Clinical implications of neuroscience
research in PTSD. Ann N Y Acad Sci 2006;1:1–17.
20. Kirkwood G, Rampes H, Tuffrey V, et al. Yoga for anxi-
ety:a systematic review of the research evidence. Br J
Sports Med 2005;39:884–891.
21. Uebelacker LA., Epstein-Lubow G, Gaudiano BA, et al.
Hatha Yoga for depression:critical review of the evidence
for efficacy, plausible mechanisms of action, and directions
for future research. J Psychiatr Pract 2010;16:22–33.
22. Kimbrough E, Magyari T, Langenberg P, et al. Mindfulness
intervention for child abuse survivors. J Clin Psychol 2010;
66:17–33.
23. van der Kolk BA, Stone L, West J, et al. Yoga as adjunctive
treatment for posttraumatic stress disorder: a randomized
controlled trial. J Clin Psychiatry 2014;76:e559–3565.
24. Birdee GS, Legedza AT, Saper RB, et al. Characteristics of
yoga users: results of a national survey. J Gen Intern Med
2008;23:1653–1658.
25. Mueser KT, Rosenberg SD, Fox L, et al. Psychometric
evaluation of trauma and posttraumatic stress disorder as-
sessments in persons with severe mental illness. Psychol
Assess 2001;13:110–117.
26. Weathers F, Ruscio AM, Keane TM. Psychometric prop-
erties of nine scoring rules for the Clinician Administered
Posttraumatic Stress Disorder Scale. Psychol Assess 1999;
11:124–133.
27. Dubester KA, Braun BG. Psychometric properties of the
Dissociative Experiences Scale. J Nerv Ment Dis 1995;
183:231–235.
28. Beck AT, Steer RA, Carbin MG. Psychometric properties
of Beck Depression Inventory: twenty-five years of evalu-
ation. Clin Psychol Rev 1988;8:77–100.
29. Briere J. Inventory of Altered Self-Capacities Professional
Manual. Odessa, FL: Psychological Assessment Resources,
2000.
30. Briere J, Kaltman S, Green BL. Accumulated childhood
trauma and symptom complexity. J Traum Stress 2008;21:
223–226.
31. Green BL, Goodman LA, Krupnick JL, et al. Outcomes of
single versus multiple trauma exposure in a screening
sample. J Traum Stress 2000;13:271–286.
32. Goodman L, Corcoran C, Turner K, et al. Assessing trau-
matic event exposure: general issues and preliminary
findings for the Stressful Life Events Screening Ques-
tionnaire. J Traum Stress 1998;11:521–542.
33. Barnes PM, Powell-Griner E, McFann K, et al. Com-
plementary and alternative medicine use among adults:
United States, 2002. Advance Data. Vital Health Stat 2004;
343:1–19.
Address correspondence to:
Alison Rhodes, PhD
The Trauma Center
at Justice Resource Institute
1269 Beacon Street
Brookline, MA 02446
E-mail: alison.rhodes@bc.edu
196 RHODES ET AL.
