2253

Psychological Adjustment and Sleep Quality in a

Randomized Trial of the Effects of a Tibetan Yoga
Intervention in Patients with Lymphoma

Lorenzo Cohen, Ph.D.1,2 BACKGROUND. Research suggests that stress-reduction programs tailored to the
Carla Warneke, M.P.H.3 cancer setting help patients cope with the effects of treatment and improve their
Rachel T. Fouladi, Ph.D.4 quality of life. Yoga, an ancient Eastern science, incorporates stress-reduction
M. Alma Rodriguez, M.D.5 techniques that include regulated breathing, visual imagery, and meditation as
Alejandro Chaoul-Reich6 well as various postures. The authors examined the effects of the Tibetan yoga (TY)
practices of Tsa lung and Trul khor, which incorporate controlled breathing and
1
Department of Behavioral Science, The University visualization, mindfulness techniques, and low-impact postures in patients with
of Texas M. D. Anderson Cancer Center, Houston, lymphoma.
Texas. METHODS. Thirty-nine patients with lymphoma who were undergoing treatment or
2
Department of Palliative Care and Rehabilitation who had concluded treatment within the past 12 months were assigned to a TY
Medicine, The University of Texas M. D. Anderson group or to a wait-list control group. Patients in the TY group participated in 7
Cancer Center, Houston, Texas. weekly yoga sessions, and patients in the wait-list control group were free to
3
Department of Biostatistics, The University of Texas participate in the TY program after the 3-month follow-up assessment.
M. D. Anderson Cancer Center, Houston, Texas. RESULTS. Eighty nine percent of TY participants completed at least 2–3 three yoga
4
Department of Psychology, Simon Fraser Univer- sessions, and 58% completed at least 5 sessions. Patients in the TY group reported
sity, Burnaby, British Columbia, Canada. significantly lower sleep disturbance scores during follow-up compared with pa-
5 tients in the wait-list control group (5.8 vs. 8.1; P ⬍ 0.004). This included better
Department of Lymphoma, The University of Texas
M. D. Anderson Cancer Center, Houston, Texas. subjective sleep quality (P ⬍ 0.02), faster sleep latency (P ⬍ 0.01), longer sleep
6
duration (P ⬍ 0.03), and less use of sleep medications (P ⬍ 0.02). There were no
Department of Religion, Rice University, Houston,
significant differences between groups in terms of intrusion or avoidance, state
Texas.
anxiety, depression, or fatigue.
Presented in part at the Sixth World Congress of
Psycho-Oncology, Banff, Alberta, Canada, April CONCLUSIONS. The participation rates suggested that a TY program is feasible for
23–27, 2003. patients with cancer and that such a program significantly improves sleep-related
outcomes. However, there were no significant differences between groups for the
Supported in part by a grant from the Bruce S. other outcomes. Cancer 2004;100:2253– 60. © 2004 American Cancer Society.
Gelb Foundation.

The authors thank Beth Notzon (Department of KEYWORDS: yoga, Tibet, lymphoma, sleep, distress.
Scientific Publications, The University of Texas
M. D. Anderson Cancer Center, Houston, TX) for
her helpful editorial comments and Rachel Boone
for her assistance with data collection. They are
grateful to Geshe Tenzin Wangyal Rinpoche for his
Y oga is an ancient Eastern tradition that usually includes regulated
breathing, moving through various postures, and meditation.1
Although different forms of yoga are practiced in many Eastern coun-
intellectual support and keeping this study true to tries, the yoga practiced in the West primarily comes from the Indian
the tradition in which these practices are imbed-
tradition, specifically the form known as Hatha yoga. Hatha yoga
ded.
typically focuses on postures (asanas) and breathing exercises
Address for reprints: Lorenzo Cohen, Ph.D., Depart- (pranayama). Less commonly practiced are the yogic practices from
ment of Behavioral Science, Unit 243, The University Tibet, which are known best for their meditative techniques. Now,
of Texas M. D. Anderson Cancer Center, 1515 Hol- however, there is growing interest in the Tibetan physical yogas or
combe Boulevard, Houston, TX 77030; Fax: (713)
mind-body practices. For thousands of years, Tibetans have been
745-4286; E-mail: [email protected]
employing what we call mind-body techniques today. Two Tibetan
Received September 18, 2003; revision received practices, known as Tsa lung (rtsa rlung; channels and vital breath)
February 24, 2004; accepted March 1, 2004. and Trul khor (’phrul ’khor; magical wheel [of the channels and vital

© 2004 American Cancer Society

DOI 10.1002/cncr.20236
Published online 16 April 2004 in Wiley InterScience (www.interscience.wiley.com).
2254 CANCER May 15, 2004 / Volume 100 / Number 10

breath]), originating from the Mother Tantra (Ma the end of the 7-week program, participants in the
rgyud)2 and the Oral Transmission of Zhang Zhung intervention group were experiencing lower total
(Zhang zhung snyan rgyud),3,4 respectively, incorpo- mood disturbance and decreased overall distress com-
rate controlled breathing and visualization, mindful- pared with the experience in the control group.
ness techniques, and postures. Little is known about In the current study, we conducted a clinical trial
this form of yoga, however, and to our knowledge no of Tibetan yoga (TY) in patients with lymphoma. We
research published to date has examined the benefits incorporated poses from the Tsa lung and Trul khor
of Tsa lung or Trul khor. We believe that this form of into a 7-week program that included controlled
yoga may be particularly useful for cancer patients breathing, visualization, and mindfulness. The tech-
who are undergoing and recovering from chemother- niques were low impact and easy to integrate into
apy, because the movements are gentle and simple, daily living and could be particularly useful for pa-
and there is an emphasis on controlled breathing, tients with cancer who were either receiving treatment
visualization, and mindfulness techniques. or had recently completed treatment. We hypothe-
There has been some research into the benefits of sized that patients assigned to the TY group would
Indian-based yoga in healthy populations and medical demonstrate better psychologic adjustment and lower
populations. Early texts described yoga’s physical and levels of fatigue and sleep disturbances during the
mental health benefits,1 whereas modern studies of weeks after the intervention compared with patients
yoga-based interventions performed in healthy popu- in a wait-list control group.
lations have shown that the interventions decrease
depression and anxiety,5–7 increase motor control,8,9 MATERIALS AND METHODS
improve subjective measures of well being,10 and im- Participants
prove lung function.11 Patients with lymphoma who were either receiving
Although yoga has been used for centuries in the chemotherapy or had received it within the past 12
East to treat disease,12,13 it has gained recognition for months were recruited through the Lymphoma Center
this purpose only recently in the West. In particular, it at The University of Texas M. D. Anderson Cancer
has been found that yoga is useful for treating various Center. Patients had to be receiving either a regimen
forms of arthritis,14 lessening the severity of musculo- with combined cyclophosphamide, doxorubicin, vin-
skeletal disease,14,15 decreasing the frequency and se- cristine, and prednisone (CHOP) or regimens with the
verity of asthma attacks,16 improving peak expiratory same drug classes to control for the more severe side
flow rates in patients with bronchial asthma,17 and effects associated with certain regimens. Patients also
improving the lipid profile of patients with coronary had to be age ⱖ 18 years and had to be able to read
artery disease18; and one small study showed that yoga and speak English. Patients with any major psychotic
was useful for controlling aspects of hypertension.19 illnesses were excluded. The proposed total sample
There is also some research suggesting that yoga may size for the study was 38 evaluable patients. Nineteen
be useful for decreasing the frequency and duration of patients per group provides 80% power to detect a
epileptic seizures, although the findings remain equiv- difference of 0.82 standard deviation units, which is
ocal.20 viewed as a large effect size.
Although there has been interest in the therapeu-
tic application of yoga to patients with cancer,21 few Procedures
studies actually have examined the benefits of yoga in Potential study participants were identified and either
this group. In an early study, 125 patients undergoing were approached in the clinic or were sent a letter
radiotherapy participated in group therapy, medita- about participating in the study. Patients who met the
tion, or yoga.22 The patients in the yoga group found inclusion criteria and expressed an interest in the
that their quality of life was improved during radio- study were then scheduled for the baseline assess-
therapy and for some time immediately after the com- ment, at which time the study was explained, patients’
pletion of radiotherapy. In particular, the patients re- questions were answered, and written consent was
ported increased appetite, increased tolerance to obtained. The study was approved by the Institutional
radiotherapy, improved sleep, improved bowel habits, Review Board. The baseline assessment included a
and a feeling of peace and tranquility. In a more recent battery of questionnaires. Patients were assigned to
trial, 109 patients with early-stage or late-stage cancer either the TY group or the wait-list control group once
were randomly assigned to either a 7-week interven- they had completed the baseline questionnaires.
tion, which included group support and discussion, Group assignment was conducted sequentially using
mindfulness meditation, visualization and imagery, minimization,24,25 a form of adaptive assignment that
and yoga stretches, or to a wait-list control group.23 At results in better group balance on selected patient
 Tibetan Yoga in Patients with Lymphoma/Cohen et al. 2255

characteristics compared with random assignment or nity to take part in the TY program after the 3-month
stratification,26 unbiased estimates of treatment effect, follow-up assessment was completed.
and as good as or better power than stratified random-
ization.26 –29 Patient characteristics used for group as- Measures
signment were the type of cancer (Hodgkin or non- Psychological adjustment was assessed across several
Hodgkin lymphoma), the status of treatment (active domains, including distress (Impact of Events Scale
treatment or completed), gender, age, and baseline [IES]), anxiety (Speilberger State Anxiety Inventory
state anxiety scores. The allocation process was con- [STATE]), and depression (Centers for Epidemiologic
cealed from all investigators because all the relevant Studies-Depression [CES-D]). Fatigue (Brief Fatigue
information was entered into a computer program Inventory [BFI]) and sleep disturbances (Pittsburgh
and group assignment was determined by the pro- Sleep Quality Index [PSQI]) were assessed, because
gram. Patients were notified of their group assignment these are common problems for patients with cancer.
by telephone, and patients in the TY group were Because it is hypothesized that yoga has an affect on
scheduled for their first yoga session, which was held multiple outcomes and because, to some extent, this
approximately 1–3 weeks after the baseline assess- was a feasibility study, a primary outcome measure
ment. Three separate cohorts of patients were as- was not prespecified. Other measures assessing some
signed to one of the two groups. After completion of proposed mediators of the benefits of the yoga pro-
the TY program, participants completed the postint- gram were administered but are not reported on here.
ervention assessment questionnaires. The follow-up The IES is a 15-item, self-report scale that mea-
assessments were conducted 1 week, 1 month, and 3 sures 2 categories of cognitive responses to stressful
months after the last session. Patients in the wait-list events: Intrusion (7 items assessing intrusively expe-
control group completed these assessments at com- rienced ideas, images, feelings, or bad dreams) and
parable intervals. Avoidance (8 items assessing consciously recognized
avoidance of certain ideas, feelings, or situations).30
Patients in the current study rated the frequency of
TY program.
intrusive thoughts and avoidance in relation to their
Participants in the TY group were asked to attend
cancer. Because the correlation between the Intrusion
seven weekly yoga sessions at The University of Texas
and Avoidance subscales at each time was modest
M. D. Anderson Cancer Center in the Place…of Well-
(correlation coefficient [r] ⫽ 0.21– 0.59), the total and
ness, a center for the clinical delivery of complemen-
subscale scores are given. Higher scores indicate more
tary programs. Each class was conducted by an expe-
intrusive thoughts and avoidance behaviors. In this
rienced TY instructor (A.C.R.). All of the practices
study, the baseline Cronbach ␣ estimates were 0.87,
come from a tradition practiced for centuries by Ti-
0.85, and 0.78 for the total IES score and for the In-
betan monks and lay practitioners and employ imag-
trusion and Avoidance subscales, respectively.
ery and exercises that are not demanding physically.
The STATE is a 20-item scale that assesses an
The program was divided into four aspects: 1) con-
individual’s current level of anxiety.31 Participants rate
trolled breathing and visualization, 2) mindfulness, 3)
the frequency of feelings or symptoms using a four-
postures from the Tsa lung, and 4) the preliminary set
point scale, with higher scores indicating higher levels
of postures from the Trul khor (sngon ’gro). The exer-
of anxiety. In this study, the baseline Cronbach ␣
cises are simple motions done with specific breathing
estimate for the STATE was 0.95.
patterns that are easy to perform by individuals un-
The CES-D is a well validated, 20-item, self-report
dergoing cancer treatment. Participants also were pro-
measure of depression that focuses on affective com-
vided with printed materials after each class that cov-
ponents of depression.32 Respondents rate the fre-
ered a new area of the program. After the last class,
quency of the specified behavior or feeling using a
patients also were given an audiotape that walked
four-point, Likert-type scale. Higher scores indicate
them through all of the techniques. They were encour-
higher levels of depressive symptoms. In this study,
aged to practice the techniques at least once per day.
the baseline Cronbach ␣ estimate for the CES-D was
Participants who missed a class were encouraged to
0.93.
attend a make-up class at another time.
The BFI is a nine-item questionnaire that was
designed to be used in the clinical setting to rapidly
Wait-list control group. assess fatigue severity. The items are ranked from 0 to
Participants in the wait-list control group did not have 10. Three questions ask patients to rate their fatigue at
any contact research personnel except during the fol- the moment and at its “worst” and “usual” over the
low-up assessments. They were offered the opportu- past 24 hours. Four items ask patients to rate how
2256 CANCER May 15, 2004 / Volume 100 / Number 10

much in the past 24 hours their fatigue has interfered acteristics used in the minimization-adaptive assign-
with their everyday life. In this study, the baseline ment procedure.34 The presence of group by time
Cronbach ␣ estimate for the BFI was 0.96. interactions also was assessed for each outcome mea-
The PSQI is an 18-item, self-rated questionnaire sure; however, no statistically significant interactions
that assesses quality of sleep and sleep disturbances were obtained (all group-by-time interaction P values
over 1 month.33 A total score is derived as well as seven were ⬎ 0.05). Thus, the results are reported for the
subscales that include Subjective Sleep Quality (1 models with main-effect terms, and the group effect
item), Sleep Latency (2 items), Sleep Duration (1 item), represents the average intervention effect across all
Habitual Sleep Efficiency (ratio of 2 items), Sleep Dis- follow-up time points adjusted for covariates. Corre-
turbances (9 items), Use of Sleeping Medications (1 lations among observations from the same individual
item), and Daytime Dysfunction (2 items). Scores ⱖ 5 were modeled using a first-order, autoregressive form
on the PSQI total scale, computed as a sum of the 7 across the follow-up assessments; inspection of Akaike
subscales, are associated with clinically significant Information Criteria (AIC), Small-Sample (AICC), and
sleep disruptions, including insomnia and major Bayesian Information Criteria (BIC) values were used
mood disorders.33 In this study, the baseline Cronbach to select the optimal within-subject correlation struc-
␣ estimate was 0.84 for the total scale, 0.79 for the ture to be used across the set of outcome measures.
9-item Sleep Disturbance subscale, and 0.77 for the The implementation of mixed-model regression anal-
2-item Sleep Latency scale, for which the interitem ysis in SAS 8.02 of restricted maximum-likelihood es-
correlation was 0.70. There was no correlation (r timation with profile residual variances was used; tests
⫽ 0.00) between the 2 items on the baseline Daytime of model fixed-effects parameters were conducted us-
Dysfunction subscale (trouble staying awake, trouble ing the Prasad-Rao-Jeske-Kackar-Harville method for
keeping enthusiasm). obtaining fixed-effects standard errors, and the Ken-
Participants in the TY group also were asked ward–Roger approach was used to calculate degrees of
about their satisfaction with the program, the per- freedom; this approach has been shown to yield good
ceived degree of benefit, and their frequency of prac- performance characteristics under small sample con-
tice outside the class for each of the four areas of ditions in longitudinal analyses.35,36
practice (breathing and visualization, mindfulness,
Tsa lung, and Trul khor). Participants were asked to RESULTS
indicate whether they felt any benefit from the prac- Twenty patients were randomized to the TY group, and
tice on a scale from 1 to 5 (1 - “no, definitely not 19 patients were randomized to the wait-list control
beneficial”; 2 - “no, not really beneficial”; 3 - “not sure group. One participant in the TY group dropped out of
if beneficial”; 4 - “yes, a little beneficial”; 5 - “yes, the study before attending any classes, making the num-
definitely beneficial”; they also could indicate “does ber of patients in that group 19; therefore, we did not
not apply, I did not practice”) and how often in the collect any follow-up data for this patient. There were no
past week they had practiced on a scale from 1 to 6 (1 statistically significant differences between the groups
- not at all; 2 - once; 3 - twice; 4 - more than twice, but with respect to any medical or demographic character-
not every day; 5 - every day; 6 - more than once a day). istics or the baseline dependent measures (for descrip-
At baseline, all patients were asked whether they tive statistics on measures of adjustment and sleep qual-
had ever practiced yoga. At each time point, patients ity, see Table 1). In both groups, the average patient age
also indicated whether they had engaged in any par- was 51 years, 12 patients were female, 7 patients had
ticular techniques to manage their stress (other than Hodgkin lymphoma, and 15 patients were not actively
TY for the patients in the TY group). receiving treatment for their cancer. There also was an
even distribution across disease stages between the
Data analyses groups (TY group: Stage I, 22%; Stage II, 39%; Stage III,
Descriptive analyses were conducted to characterize 17%; Stage IV, 22%; control group: Stage I, 22%; Stage II,
the study participants and the feasibility of conducting 33%; Stage III, 12%; Stage IV, 33% Ann Arbor Criteria). A
a TY trial in patients with lymphoma. Analyses of the chart review indicated that 10 patients in the TY group
impact of the TY intervention relative to a wait-list and 11 patients in the control group were not taking any
control group were conducted on follow-up IES, state medications. Medications that patients were taking in-
anxiety, depression symptomology, fatigue, and sleep cluded antidepressants (TY group, 2 patients; control
disturbance scores using mixed-model regression group, 2 patients), supplements (TY group, 2 patients;
analyses by regressing follow-up assessments on control group, 3 patients), thyroid medication (TY group,
group, time of follow-up assessment, and the corre- 2 patients; control group, 3 patients), hormone therapy
sponding baseline measure as well as the patient char- (TY group, 1 patient; control group, 2 patients), neupo-
 Tibetan Yoga in Patients with Lymphoma/Cohen et al. 2257

TABLE 1
Baseline and Follow-Up Adjustment and Sleep Quality Scores by Intervention Groupa

Wait-list control group

Yoga group (mean ⴞ SD) (mean ⴞ SD)

Measure Baseline Follow-up Baseline Follow-up P value 95% CI

Impact of Events Scale

Total score 21.5 ⫾ 14.8 10.2 ⫾ 8.2 18.7 ⫾ 11.1 10.5 ⫾ 8.5 0.92 ⫺5.4 to 4.9
Intrusive thoughts 9.0 ⫾ 6.7 3.7 ⫾ 4.4 9.1 ⫾ 6.6 4.2 ⫾ 4.6 0.77 ⫺3.2 to 2.4
Avoidance 13.1 ⫾ 9.2 6.1 ⫾ 6.0 10.6 ⫾ 6.6 6.1 ⫾ 6.0 0.99 ⫺3.8 to 3.7
State Anxiety 34.3 ⫾ 12.3 34.1 ⫾ 8.4 37.8 ⫾ 14.6 33.8 ⫾ 8.5 0.90 ⫺5.0 to 5.7
Depression (CES-D) 10.2 ⫾ 11.0 9.0 ⫾ 4.2 9.6 ⫾ 8.57 9.7 ⫾ 3.8 0.56 ⫺3.3 to 1.9
BFI 3.1 ⫾ 2.4 3.1 ⫾ 1.5 2.8 ⫾ 2.2 3.1 ⫾ 1.5 0.93 ⫺0.99 to 0.9
Sleep disturbance (PSQI)
Total score 6.5 ⫾ 5.0 5.8 ⫾ 2.3 7.2 ⫾ 4.7 8.1 ⫾ 2.4 0.004 ⫺3.8 to ⫺0.8
Sleep quality 0.90 ⫾ 0.85 0.81 ⫾ 0.52 1.11 ⫾ 0.94 1.22 ⫾ 0.56 0.02 ⫺0.7 to ⫺0.08
Sleep latency 1.10 ⫾ 0.97 0.75 ⫾ 0.68 1.05 ⫾ 0.91 1.33 ⫾ 0.71 0.01 ⫺1.0 to ⫺0.1
Sleep duration 0.85 ⫾ 1.09 0.89 ⫾ 0.64 1.32 ⫾ 1.29 1.35 ⫾ 0.64 0.03 ⫺0.9 to ⫺0.04
Sleep efficiency 0.65 ⫾ 1.09 0.87 ⫾ 0.68 0.84 ⫾ 1.07 0.95 ⫾ 0.67 0.72 ⫺0.5 to 0.4
Sleep disturbances 1.40 ⫾ 0.68 1.23 ⫾ 0.40 1.37 ⫾ 0.76 1.33 ⫾ 0.37 0.47 ⫺0.3 to 0.16
Sleeping medications 0.80 ⫾ 1.28 0.48 ⫾ 0.88 0.58 ⫾ 1.07 1.21 ⫾ 0.93 0.02 ⫺1.3 to ⫺0.2
Daytime dysfunction 0.80 ⫾ 0.62 0.96 ⫾ 0.60 0.95 ⫾ 0.71 0.93 ⫾ 0.64 0.89 ⫺0.4 to 0.4

SD: standard deviation; 95% CI: 95% confidence interval; CES-D: Centers for Epidemiologic Studies-Depression; BFI: Brief Fatigue Inventory; PSQI: Pittsburgh Sleep Quality Index.
a
P values and 95% confidence intervals are in relation to the group comparisons for the follow-up data. Note that follow-up adjustment scores represent least-squares means adjusted for the baseline value of the
outcome measure and state anxiety, age, gender, treatment status, and the type of cancer (Hodgkin or non-Hodgkin lymphoma) using the final, mixed-model regression models (n ⫽ 16 patients in the yoga group;
n ⫽ and 14 patients in the control group).

gen/procrit (TY group, 2 patients; control group, 1 pa- All 19 TY participants attended at least 1 yoga
tient), hypertension medication (TY group, 2 patients; session. Six participants (32%) attended all 7 sessions
control group, 1 patient), steroids (TY group, 2 patients), through attendance at either the primary session or
antiemetic medication (TY group, 1 patient), diabetes through a make-up class; 5 participants (26%) at-
medication (TY group, 2 patients; control group, 1 pa- tended 5 or 6 sessions; 6 participants (32%) attended 2
tient), cholesterol medication (TY group, 2 patients), an- or 3 sessions; and 2 participants (10%) attended only 1
tiseizure medication (control group, 1 patient), and an- session. Patients missed classes for the following rea-
tibiotics (control group, 2 patients). sons: they were out of town, they had conflicts with
Three patients in each group also indicated that their work schedule, they had a cancer treatment-
they had practiced yoga in the past; however, none related reason, they were too busy, they had transpor-
were practicing yoga currently. Nine patients in the TY tation problems, or they had a health problem (not
group and 14 patients in the control group indicated cancer-related). Sixteen of 19 patients (84%) in the TY
that they did something to manage their stress before group and 14 of 19 patients (74%) in the control group
the start of the study, including exercise (TY group, 4 completed at least 1 of the 3 follow-up assessments,
patients; control group, 8 patients), meditation (TY with an average of 2.0 (standard deviation [SD], 0.82)
group, 1 patient), breathing exercises (TY group, 1 and 2.4 (SD, 0.65) assessments completed in the TY
patient; control group, 2 patients) relaxation tapes group and the control group, respectively (P ⫽ 0.13).
(control group, 1 patient), and hobbies (TY group, 3 There were no significant demographic, medical, or
patients; control group, 3 patients). Eight patients in psychosocial differences between the 9 patients who
the TY group and nine patients in the control group did not complete any follow-up assessments and the
indicated that they used some technique(s) to manage 30 patients who did complete the assessments.
their stress at some time during the follow-up period, In their evaluations, participants indicated that
including exercise (TY group, 7 patients; control they found the TY program beneficial. In fact, none of
group, 7 patients), meditation (TY group, 1 patient; the participants had responses of 0 (“does not apply, I
control group, 1 patient), breathing exercises (control did not practice”) or 1 (“no, definitely not beneficial”)
group, 2 patients), support group (control group, 1 for any of the 4 aspects. Pooling across the postinter-
patient), and hobbies (TY group, 3 patients; control vention assessments, the modal response for each of
group, 2 patients). the 4 aspects was 5 (“yes, definitely beneficial”); 73%,
2258 CANCER May 15, 2004 / Volume 100 / Number 10

TABLE 2 during follow-up, the TY group had significantly better

Perceived Benefit and Practice Rates for Breathing and Visualization, overall sleep quality and that they perceived better
Mindfulness, Tsa Lung, and Trul Khor
sleep quality, fell asleep more quickly, slept longer,
Benefita Practiceb and used fewer sleep medications compared with pa-
tients in the control group (see Table 1). There were no
Exercise Mean Median Mean Median statistically significant differences between the groups
in terms of the Intrusion or Avoidance scores, state
Breathing and visualization 4.7 5.0 3.5 3.8
anxiety, depression, or fatigue scores.
Mindfulness 4.5 4.7 3.1 3.3
Tsa lung 4.5 4.7 2.9 3.0
Trul khor 4.6 5.0 2.8 2.9 DISCUSSION
The feasibility of conducting the seven-session TY pro-
Tsa lung (rtsa rlung): channels and vital breath; Trul khor (phrul khor): magical wheel of the channels gram in patients with lymphoma was demonstrated
and vital breath.
a
clearly. In particular, although ⬎ 20% of patients were
Benefit: 1 - no, definitely not beneficial; 2 - no, not really beneficial; 3 - not sure if beneficial; 4 - yes,
a little beneficial; 5 - yes, definitely beneficial. undergoing chemotherapy, the majority of patients at-
b
Practice: 1 - not at all; 2 - once; 3 - twice; 4 - more than twice, but not every day; 5 - every day; 6 - more tended ⬎ 50% of sessions. Patients also indicated that
than once daily. they found the program useful, and ⬎ 80% said they
practiced some aspect of the program at least once per
week during the 3-month follow-up. Most importantly,
64%, 64%, and 82% of respondents indicated 5 (“yes, there was some indication that the TY program reduced
definitely beneficial”) to breathing and visualization, patients’ sleep disturbances. In particular, patients in the
mindfulness, Tsa lung, and Trul khor, respectively. TY group reported significantly better overall sleep qual-
The mean and median responses were between 4 ity and subjective sleep quality, faster sleep latency,
(“yes, a little beneficial”) and 5 (“yes, definitely bene- longer sleep duration, and less use of sleep medications
ficial”) (see Table 2). Similarly, pooling across inter- during the follow-up compared with patients in the con-
vention and postintervention periods, the participants trol group. However, there were no statistically signifi-
indicated a fair amount of practice in each of the four cant group differences for the measures of psychological
aspects, averaging two times per week (see Table 2). adjustment or fatigue.
The modal response regarding average weekly prac- To our knowledge this is the first study to examine
tice of breathing and visualization, Tsa lung, and Trul TY in any population and one of the few studies to
khor was 4 (“more than twice, but not every day”), examine the effects of yoga in a cancer population. In
with that level of specific practice reported by 39%, other studies that have examined the benefits of includ-
31%, and 29% of participants, respectively; the second ing yoga, yoga constituted only one aspect of the inter-
most common level of practice reported was 3 (“twice vention. For example, Speca et al.23 designed an inter-
a week”; breathing and visualization, 28%; Tsa lung, vention that included group support and discussion,
27%; Trul khor, 21%). For weekly practice of mindful- mindfulness meditation, visualization and imagery, and
ness, 31% of participants averaged practicing “more yoga stretches. Although they found that the participants
than twice, but not every day,” and the same percent- in the intervention group experienced lower total mood
age averaged “twice a week.” disturbance and decreased overall distress, it is not clear
Table 1 shows the overall follow-up results on whether the benefits were due to the yoga per se. It is
psychologic adjustment, fatigue, and sleep distur- noteworthy that the form of yoga we used in the current
bances by intervention group. The values are least- study was presented and taught in its traditional format,
squares means that have been adjusted for the base- which includes a number of components (e.g., breathing
line value of the outcome measure and the and visualization, mindfulness, Tsa lung, and Trul khor).
minimization adaptive allocation factors (i.e., state However, although the TY program that we designed
anxiety, age, gender, status of treatment, and type of appeared to be useful for this population, we cannot tell
cancer [Hodgkin or non-Hodgkin lymphoma]), as well which particular aspects of the program were more or
as time of follow-up assessment, which was not a less useful for the participants.
statistically significant effect in any of the models. The There are several possible explanations for why we
mixed-model regression analyses revealed a statisti- did not find an effect of TY on the measures of psycho-
cally significant adjusted group effect for the PSQI logical adjustment (intrusion or avoidance, state anxiety,
total score (P ⬍ 0.004) and for 4 of the subscales, and depression) and fatigue. First, the TY program truly
including Subjective Sleep Quality (P ⬍ 0.02), Sleep may not affect these outcomes for this population. Al-
Latency (P ⬍ 0.01), Sleep Duration (P ⬍ 0.03), and Use ternatively, the lack of effect also may reflect low sensi-
of Sleep Medications (P ⬍ 0.02). This suggests that, tivity of the measures. For example, because both the
 Tibetan Yoga in Patients with Lymphoma/Cohen et al. 2259

CES-D and the STATE are ofter used as clinical screening strom et al.40 found that the most common sleep
measures of mental health (for depression and anxiety, problems were awakening during the night and difficulty
respectively), these instruments may not detect only falling back to sleep as well as not sleeping enough. Of
subtle changes in mental health in a nonmental health particular concern is the possibility that there may be
population. Less clinically oriented measures, such as profound consequences of disturbed sleep for cancer
the Profile of Mood States37 used by Spica et al.,23 might patients, because sleep may mediate both psychologic
have detected group differences in our population. health and physical health.38,41,42 In fact, sleep distur-
Along these same lines, there also might have been a bances have been associated with worse quality of life in
floor effect for some of these measures, because the cancer patients, including increased depression and
scores were quite low at baseline and follow-up. In ad- higher pain levels,40,43 and with depression in noncancer
dition, most patients were off treatment at the start of patients.42 Persistent sleep disturbances also may in-
the study (15 patients in each group), which may have crease a cancer patient’s risk for mood disorders, altered
contributed further to a floor effect. It is important to metabolic and endocrine function, and compromised
note that the study also may have been underpowered to immune functioning.38,41,43,44 It is noteworthy that sleep
detect differences for these outcomes when the effect disruptions have been associated with adverse physical
size may have been smaller than the effect of the TY health outcomes, including increased morbidity and
program on the measure of sleep disturbances. mortality.45– 47 Potential changes in metabolic, endo-
There are several limitations to the current study, crine, and immune function may have particularly im-
most significantly, the small sample size of just 30 pa- portant health consequences for patients with hemato-
tients for whom we had follow-up data on any of the logic malignancies.
selected outcome measures; however, as recent studies Although research into the efficacy and mecha-
have shown, this level of sample size can yield adequate nisms of yoga is in its beginning stages, the findings
Type I error control and power.35,36 Nonetheless, this is reported to date are supportive13 and, along with our
one of the only studies examining the benefits of yoga in finding of improved sleep, suggest that the health
a cancer population, and 14 –15 patients per group pro- effects of yoga in cancer patients should be explored
vides 80% power to detect a population effect size of 1.1 further. The benefits that have been documented and
and 50% power to detect a population effect size of 0.7, the potential impact of these benefits on the psycho-
with sample effect sizes ⱖ 0.7 achieving statistical signif- logic and physical sequelae of cancer are important
icance. It is also important to note that we had several enough to warrant the further study of developing
outcome measures of interest in this trial. However, de- such programs for cancer patients.
spite the examination of multiple outcomes, the strength
of the finding that the TY intervention had an effect on REFERENCES
sleep quality is greater than would be expected by 1. Taimini I. The science of yoga. Madras: The Theosophical
chance alone. The study also was limited by the fact that Publishing House, 1961.
we relied on self-reports to determine compliance and to 2. Samlek M. The three basic Mother Tantras with commen-
taries (reproduced from the original XI century manuscript
assess the main outcomes. Although research using the
belonging to the Monastery of Samling, Dolpo, Nepal) [in
PSQI suggests there is a good association between self- Tibetan]. New Delhi: Bonpo Monastic Centre, 1971.
reports and polysomnographic sleep measures, future 3. Chandra L, Namdak T, editors. The magical wheel oral wis-
studies should incorporate some objective outcomes. dom quintessential instructions from the Great Perfection
We also cannot know from our data whether the benefits Oral Transmission of Zhang Zhung (reproduced from an
original XI century manuscript). In: Chandra L, Mandak T,
of the program were attributable directly to TY as a
editors. History and doctrines of Bonpo Yoga (Satapitaka
whole or to some nonspecific aspect of the program, series, volume 73) [in Tibetan]. New Dehli: International
such as attention, social support, relaxation, or stretch- Academy of Indian Culture, 1968:631– 643.
ing. To determine this, future studies will need to use 4. Sonam N, Gyaltsen PL, Gyatso K, editors. The Magical Whell,
more complex research designs. channels and vitalbreath of the Oral Transmission of Zhang
Zhung. (Commentary on: The Magical Wheel oral wisdom
Our findings that the TY program was associated
quintessential instructions from the Great Perfection Oral
with a decrease in overall sleep disturbances and an Transmission of Zhang Zhung) [in Tibetan]. In: Sonam N,
improvement in subjective sleep quality, sleep latency, Gyaltsen PL, Gyatso K, editors. The vast treasury of pro-
sleep duration, and use of sleep medications are partic- found space. New Thobgyal: Tibetan Bonpo Monastic Cen-
ularly salient in a cancer population. This is because tre, 1974:321–346.
5. Janakiramaiah N, Gangadhar BN, Naga Venkatesha Murthy PJ,
sleep disturbances are a common and underinvestigated
Harish MG, Subbakrishna DK, Vedamurthachar A. Antidepres-
problem among cancer patients.38 Indeed, in some stud- sant efficacy of Sudarshan Kriya Yoga (SKY) in melancholia: a
ies, up to 75% of the cancer patients surveyed experi- randomized comparison with electroconvulsive therapy (ECT)
enced disturbances in their sleep and rest cycle.39 Eng- and imipramine. J Affect Disord. 2000;57:255–259.
2260 CANCER May 15, 2004 / Volume 100 / Number 10

6. Vahia NS, Doongaji DR, Jeste DV, Ravindranath S, Kapoor 28. Birkett NJ. Adaptive allocation in randomized controlled
SN, Ardhapurkar I. Psychophysiologic therapy based on the trials. Control Clin Trials. 1975;6:146 –155.
concepts of Patanjail. A new approach to the treatment of 29. Thernau TM. How many stratification factors are “too
neurotic and psychosomatic disorders. Am J Psychother. many” to use in a randomization plan? Control Clin Trials.
1973;27:557–565. 1993;14:98 –108.
7. Malathi A, Damodaran A. Stress due to exams in medical 30. Horowitz M, Wilner N, Alvarez W. Impact of Events Scale: mea-
students—role of yoga. Indian J Physiol Pharmacol. 1999;43: sure of subjective stress. Psychosom Med. 1979;41:209–218.
218 –224. 31. Spielberger CD, Gorsuch R, Lushene RE. STAI manual for
8. Telles S, Hanumanthaiah BH, Nagarathna R, Nagendra HR. the State-Trait Anxiety Inventory. Palo Alto: Consulting Psy-
Plasticity of motor control systems demonstrated by yoga chologists Press, 1970.
training. Indian J Physiol Pharmacol. 1994;38:143–144. 32. Radloff LS. The CES-D scale: a new self-report depression
9. Dash M, Telles S. Yoga training and motor speed based on a scale for research in the general population. Appl Psychol
finger tapping task. Indian J Physiol Pharmacol. 1999;43: Meas. 1977;1:385– 401.
458 – 462.
33. Buysse DJ, Reynolds CF, Monk TH, Berman SR. Pittsburgh
10. Malathi A, Damodaran A, Shah N, Patil N, Maratha S. Effect
Sleep Quality Index: a new instrument for psychiatric prac-
of yogic practices on subjective well being. Indian J Physiol
tice and research. Psychiatr Res. 1989;28:193–213.
Pharmacol. 2000;44:202–206.
34. Friedman LM, Furberg CD, Demets DL. Fundamentals of
11. Joshi LN, Joshi VD, Gokhale LV. Effect of short term
clinical trials, 3rd ed. New York: Springer-Verlag, 1998.
“Pranayam” practice on breathing rate and ventilatory func-
35. Fouladi RT, Shieh YY. Power profiles of modified likelihood
tions of lung. Indian J Physiol Pharmacol. 1992;36:105–108.
12. Nagendra HR, Nagarathna R. Applications of integrated ap- ratio tests on mixed linear model parameters. American
proach of yoga therapy—a review. A new light for asthmat- Statistical Association, editor. Proceedings of the Joint Sta-
ics. Bangalore: Vivekananda Kendra, 1986. tistical Meetings, Biometrics Section [CD Rom]. Alexandria:
13. Telles S, Naveen KV. Yoga for rehabilitation: an overview. American Statistical Association, 2001.
Indian J Med Sci. 1997;51:123–127. 36. Kenward MG, Roger JH. Small sample inference for fixed
14. Garfinkel M, Schumacher HR Jr. Yoga. Rheum Dis Clin North effects from restricted maximum likelihood. Biometrics.
Am. 2000;26:125–132. 1997;53:983–997.
15. Garfinkel MS, Schumacher HR Jr., Husain A, Levy M, Resh- 37. McNair DM, Lorr M, Droppleman LF. Profile of mood states.
etar RA. Evaluation of a yoga based regimen for treatment of San Diego: Educational and Industrial Testing Service, 1981.
osteoarthritis of the hands. J Rheumatol. 1994;21:2341–2343. 38. Savard J, Morin CM. Insomnia in the context of cancer: a
16. Vedanthan PK, Kesavalu LN, Murthy KC, et al. Clinical study review of a neglected problem. J Clin Oncol. 2001;19:895–908.
of yoga techniques in university students with asthma: a 39. Malone M, Harris AL, Luscombe DK. Assessment of the
controlled study. Allergy Asthma Proc. 1998;19:3–9. impact of cancer on work, recreation, home management
17. Nagarathna R, Nagendra HR. Yoga for bronchial asthma: a and sleep using a general health status measure. J R Soc
controlled study. Br Med J (Clin Res Ed). 1985;291:1077–1079. Med. 1994;87:386 –389.
18. Mahajan AS, Reddy KS, Sachdeva U. Lipid profile of coro- 40. Engstrom CA, Strohl RA, Rose L, Lewandowski L, Stefanek
nary risk subjects following yogic lifestyle intervention. In- ME. Sleep alterations in cancer patients. Cancer Nurs. 1999;
dian Heart J. 1999;51:37– 40. 22:143–148.
19. Murugesan R, Govindarajulu N, Bera TK. Effect of selected 41. Hall M, Baum A, Buysse DJ, Prigerson HG, Kupfer DJ, Reyn-
yogic practices on the management of hypertension. Indian olds CF. Sleep as a mediator of the stress-immune relation-
J Physiol Pharmacol. 2000;44:207–210. ship. Psychosom Med. 1998;6:48 –51.
20. Ramaratnam S, Sridharan K. Yoga for epilepsy. Cochrane 42. Dew MA, Reynolds CF, Buysse DJ, Houck PR, Monk TH,
Database Syst Rev. 2000;(3):CD001524. Kupfer DJ. Electroencephalographic sleep profiles during
21. Collins C. Yoga: intuition, preventive medicine, and treat- depression. Effects of episode duration and other clinical
ment. J Obstet Gynecol Neonatal Nurs. 1998;27:563–568.
and psychosocial factors in older adults. Arch Gen Psychia-
22. Joseph CD. Psychological supportive therapy for cancer pa-
try. 1996;53:148 –156.
tients. Indian J Cancer. 1983;20:268 –270.
43. Koopman C, Nouriani B, Erickson V, et al. Sleep distur-
23. Speca M, Carlson LE, Goodey E, Angen M. A randomized,
bances in women with metastatic breast cancer. Breast J.
wait-list controlled clinical trial: the effect of a mindfulness
2002;8:362–370.
meditation-based stress reduction program on mood and
44. Spiegel K, Leproult R, Van Couter E. Impact of sleep debt on
symptoms of stress in cancer outpatients. Psychosom Med.
2000;62:613– 622. metabolic and endocrine function. Lancet. 1999;354:1435–
24. Taves DR. Minimization: a new method of assigning pa- 1439.
tients to treatment and control groups. Clin Pharmacol 45. Dew MA, Hoch CC, Buysse DJ, et al. Healthy older adults’
Ther. 1974;15:443– 453. sleep predicts all-cause mortality at 4 to 19 years of follow-
25. Pocock SJ, Simon R. Sequential treatment assignment with up. Psychosom Med. 2003;65:63–73.
balancing for prognostic factors in the controlled clinical 46. Newman AB, Enright PL, Manolio TA. Sleep disturbance,
trial. Biometrics. 1975;31:103–115. psychosocial correlates and cardiovascular disease in 5201
26. Rovers MM. Comparison of balanced and random alloca- older adults: the Cardiovascular Health Study. J Am Geriatr
tion in clinical trials. Eur J Epidemiol. 2000;16:1123–1129. Soc. 1997;45:1–7.
27. Forsythe AB. Validity and power of tests when groups have 47. Eaton WW, Badawi M, Melton B. Prodromes and precursors:
been balanced for prognostic factors. Comput Stat Data epidemiologic data for primary prevention of disorders with
Anal. 1987;5:193–200. slow onset. Am J Psychiatry. 1995;152:967–972.