Yoga for Women with Metastatic Breast Cancer: Results from a Pilot Study

Vol. 33 No. 3 March 2007 Journal of Pain and Symptom Management 331

Clinical Note

Yoga for Women with Metastatic BreastCancer: Results from a Pilot StudyJames W. Carson, PhD, Kimberly M. Carson, MPH, Laura S. Porter, PhD,Francis J. Keefe, PhD, Heather Shaw, MD, and Julie M. Miller, MADepartments of Psychiatry and Behavioral Sciences (J.W.C., K.M.C., L.S.P., F.J.K., J.M.M.)

and Medicine (H.S.), Duke University Medical Center, Durham, North Carolina, USA

AbstractMetastatic breast cancer (MBC) remains a terminal illness for which major treatmentadvances are slow to appear, and hence it is crucial that effective palliative interventions bedeveloped to reduce the cancer-related symptoms of women with this condition during theremaining years of their lives. This pilot/feasibility study examined a novel, yoga-basedpalliative intervention, the Yoga of Awareness Program, in a sample of women with MBC.The eight-week protocol included gentle yoga postures, breathing exercises, meditation,didactic presentations, and group interchange. Outcome was assessed using daily measuresof pain, fatigue, distress, invigoration, acceptance, and relaxation during twopreintervention weeks and the final two weeks of the intervention. Thirteen women completedthe intervention (mean age¼ 59; mean time since diagnosis¼ 7 years; two AfricanAmerican, 11 Caucasian). During the study, four participants had cancer recurrences, andthe physical condition of several others deteriorated noticeably. Despite low statistical power,pre-to-post multilevel outcomes analyses showed significant increases in invigoration andacceptance. Lagged analyses of length of home yoga practice (controlling for individual meanpractice time and outcome levels on the lagged days) showed that on the day after a dayduring which women practiced more, they experienced significantly lower levels of pain andfatigue, and higher levels of invigoration, acceptance, and relaxation. These findingssupport the need for further investigation of the effects of the Yoga of Awareness Program inwomen with MBC. J Pain Symptom Manage 2007;33;331e341. � 2007 U.S. CancerPain Relief Committee. Published by Elsevier Inc. All rights reserved.

Key WordsMetastatic breast cancer, pain, fatigue, yoga, meditation

This work was supported by funding from theNorth Carolina Academic Alliance for IntegrativeMedicine.

Address reprint requests to: James W. Carson, PhD, PainPrevention & Treatment Research Program, Depart-ment of Psychiatry, Duke University Medical Center,Box 90399, Durham, NC 27708, USA. E-mail:[email protected]

Accepted for publication: August 11, 2006.

� 2007 U.S. Cancer Pain Relief CommitteePublished by Elsevier Inc. All rights reserved.

IntroductionMetastatic breast cancer (MBC) constitutes

a serious life threat for women, with an averageprognosis of 18e24 months to live.1 Pain, fa-tigue, and emotional distress are often re-ported as the most debilitating symptoms ofMBC patients.2e5 Although pharmacologicalinterventions may help these women to somedegree, these symptoms rarely resolve.4,6,7

0885-3924/07/$esee front matterdoi:10.1016/j.jpainsymman.2006.08.009

mailto:[email protected]

332 Vol. 33 No. 3 March 2007Carson et al.

Given that MBC is a terminal illness for whichmajor treatment advances are slow to appear, itis crucial that effective adjunctive palliative in-terventions be developed to reduce the cancer-related symptoms of these women during theremaining years of their lives.

Few palliative behavioral interventions havebeen tested with MBC patients, with mixed re-sults. A recent review of cognitive-behavioralgroup therapy8e11 and supportive-expressivegroup therapy12e17 trials concluded that al-though some evidence exists for short-termbenefits from these approaches (e.g., bettermood9,15 and less increase in pain14,16), suchchanges are not maintained even for a fewmonths.18 More recently, results from an exer-cise intervention tailored for MBC demon-strated short-term effects in terms of slowerrates of deterioration in well-being andfatigue.19

There is thus a clear need to test new behav-ioral approaches to ameliorating cancer-re-lated symptoms in women with MBC.18 Yogais a popular approach to health maintenance,which holds promise for benefiting womenwith MBC. As a mind/body discipline originat-ing in India, yoga has been practiced for itsproposed physical, mental, and spiritual bene-fits for thousands of years.20 Yoga is one ofa variety of complementary and alternativemedicine adjunctive approaches that cancerpatients of all cultural backgrounds have beenseeking out in dramatically larger numbers inrecent years.21e24 Yoga is now offered at severalmajor treatment centers (e.g., M.D. Anderson,Memorial Sloan-Kettering, UCLA/Jonsson) viatheir complementary therapy services, and re-cently yoga has been the focus of severalcancer studies.25e29 Results from two random-ized trials of yoga for cancer patients havebeen published. Cohen et al. reported im-provement in sleep disturbance in a trial withlymphoma patients,27 and Culos-Reed et al.demonstrated improvements in mood, qualityof life, and stress in breast cancer survivors.28

However, thus far no study has investigatedthe use of yoga with MBC patients. The ratio-nale for applying a yoga-based interventionin this population rests on several premises.First, research suggests that yoga can producean ‘‘invigorating effect on mental and physicalenergy’’ that improves physical fitness andcounteracts fatigue,30 a problematic symptom

in MBC. Second, historically in the practiceof yoga, a fundamental emphasis is placed onaccepting one’s moment-to-moment experi-ences, whatever they may be, without forcingthe body beyond its comfortable limits. Thisis because struggles to control the body, or tocontrol one’s physical sensations, thoughts,or emotions often only exacerbate problems.Such struggles also detract from focusing onthe personally fulfilling activities at hand.The important role of a healthy sense of accep-tance in decreasing distress in the face ofunpleasant symptomology has received in-creased research and clinical attention,31 andmay be especially important for individualsdealing with a life-threatening illness.32 Third,studies have demonstrated that yoga producesthe relaxation response characterized by de-creased sympathetic and increased parasympa-thetic activity.33e38 The relaxation response islikely to improve symptoms of pain, fatigue,and distress.39,40

In summary, yoga appears to promote at leastthree therapeutic processesdinvigoration, ac-ceptance, and relaxationdwhich are likely tohave a favorable effect on MBC patients. Thepurposes of this pilot study were to determinethe feasibility of a novel, yoga-based eight-weekgroup intervention in women with MBC andto examine the intervention’s effects on pain,fatigue, distress, invigoration, acceptance, andrelaxation. We did not specify a priori hypothe-ses regarding improvements in these outcomesbecause of (a) the preliminary, uncontrolled de-sign of the study and (b) the fact that significantfindings from intervention studies with thispopulation have often consisted of less deterio-ration in the intervention vs. control condi-tions, rather than actual improvement insymptoms.12,14,16,17,19 The intervention weuseddthe Yoga of Awareness Programdwasspecifically designed for MBC patients, and in-cluded gentle yoga postures, regulated breath-ing, guided meditations, brief didacticpresentations, and group discussions.

Compared to previous MBC interventionstudies, this study makes a unique contributionin terms of the method of analyzing treatmentoutcome. Data on symptoms (pain, fatigue,distress) and therapeutic processes (invigora-tion, acceptance, relaxation) were collectedin the form of prospective daily diaries. Rela-tive to traditional survey questionnaires,

Vol. 33 No. 3 March 2007 333Yoga for Women with Metastatic Breast Cancer

diaries offer improved accuracy, reduced recallbias, and increased recall of symptoms and re-lated events, especially regarding the temporalsequencing of events.41e45 To analyze the diarydata, we used multilevel random effectsmodels.46 These analyses are preferable todata analytic strategies that aggregate daily as-sessments (e.g., ordinary regression models)because they permit 1) tracking of symptom-re-lated processes as they occur in naturalistic set-tings, 2) accounting for two levels of sampling,that is, both within-person variation and be-tween-person variation, 3) controlling for auto-correlation, that is, the serial dependency thatresults from successive daily assessments, and4) handling missing data that often occur inintensive longitudinal data collection.46e49

MethodsParticipants and Setting

Volunteers for this study included 21 adultwomen with MBC referred by oncologists atthe Duke University Medical Center breast on-cology unit and affiliate sites. Sample illustra-tions of the postures used during the yogasessions were provided to referring physiciansand to potential participants to give them a bet-ter sense of what the intervention would involve.Patients were excluded if they had less than six-month life expectancy, if changes had takenplace in their use of any antidepressants duringthe previous three months, if they had receivedtreatment for serious psychiatric disorders (e.g.,schizophrenia) in the previous six months, ifthey were currently engaged in intensive yogapractice (>three days per week), if drivingtime to attend weekly sessions was excessive(>one hour as a rule of thumb), or if theywere not English speaking. Of the women whovolunteered for the study, three withdrew beforebeginning the intervention (2¼ scheduling dif-ficulties, 1¼ custody-case court appearances).Of the 18 remaining women, four withdrewshortly after beginning the intervention(1¼ scheduling difficulties, 1¼ traveling dis-tance, 1¼ relocated to another state, 1¼ healthstatus deterioration), and one woman did notcomplete post diaries because of mental statusdeterioration (dementia-like). Attrition subse-quent to beginning the intervention was thus28%, which is similar to other intervention stud-ies with MBC patients (e.g., Edelman

et al.9¼ 26%, Goodwin et al.14¼ 32%, Spiegeland Bloom¼ 31%16).

This left 13 women who completed the inter-vention and provided pre- and postmeasures.The mean age of this sample was 59 years(range¼ 44e75). The average time since diag-nosis was seven years (range¼ 0e26). Sevenwere concurrently receiving chemotherapytreatments, and six were not; four had new can-cer reoccurrences while participating in thestudy. Six patients previously practiced yoga ormeditation to some extent (of which two hadpracticed during the previous year). Eleven pa-tients were Caucasian and two were AfricanAmerican. Eleven were married, one was wid-owed, and one was divorced. Fifteen percenthad graduated from high school, 54% hadattended college, and 31% had attendedgraduate school.

ProcedureThe protocol for this study was approved by

the Duke Institutional Review Board. Prior tothe study, informed consent was obtainedfrom all participants. All women were asked toprovide basic demographic information and in-formation relevant to their breast cancer treat-ment history. Participants continued to receivethe standard care provided by their healthcare providers. To control for a potentially im-portant medication-related confound, aftercompleting the intervention, patients wereasked about any changes in antidepressantuse; no changes were reported. After complet-ing the yoga program and postinterventionmeasures, women were invited to participatein a focus group to give qualitative feedbackregarding the program.

Data Collection

Daily Symptom and Process Measures. A briefdaily-diary measurement strategy was chosenbecause of the demonstrated reliability andvalidity of this method in health-related fields,and the fact that diaries allow subjective events,such as sensations and feelings, to be reportedunobtrusively and with little introspection inindividuals’ natural settings.45 Data from dia-ries also permit a careful analysis of day-to-day interrelated happenings for each personand provide increased statistical power whenanalyzing a small clinical sample.48


Before going to bed each evening, patientscompleted a daily-diary log, in which they re-corded their levels of pain, fatigue, distress, in-vigoration, acceptance, and relaxation duringtwo preintervention weeks (pre), and duringthe last two weeks of the intervention period(post). All daily variables were indicated bymarking 100-mm visual analogue scales(VAS), in which higher scores reflected greateramounts (e.g., for pain, the item read ‘‘Pleaseindicate the average level of pain you had to-day,’’ with anchors set as ‘‘No Pain’’ and‘‘Pain as Bad as it Can Be’’; for acceptance,the item read ‘‘Please indicate how easy it wasfor you today to accept and let be the waysyou have been affected by your medical condi-tion,’’ with anchors set as ‘‘No Acceptance’’and ‘‘Complete Acceptance’’). Similar VASmeasures are extensively used in clinical set-tings to measure subjective phenomena, andhave been shown to be valid, reliable, rapid,and sensitive in measuring such variables asglobal affect, pain, and fatigue.50 The post di-ary also asked participants to indicate howmany minutes were spent in completing theday’s yoga practice assignment.

An important requirement for the effectiveuse of diary measures is that participants receiveadequate training in the completion of dia-ries.45 Participants in this study were trained inhow to complete the diary by a research assistantwho helped them complete sample diaries. Pa-tients also were called during the first week ofeach recording period to inquire about any dif-ficulties and to answer any questions. Anotherkey requirement of most daily-diary studies isthat each day’s responses be recorded by theend of that day.51 To bolster this expectation,we asked participants to mail each day’s diaryback to us the following morning in pread-dressed stamped envelopes. To further facilitatemotivation and compliance, patients were paid$0.25 for each completed daily diary and a $1.50bonus for each week of complete recording.The diary completion rate was 89% (324 of364 potentially reportable days across 13 partic-ipants; range, 71e100%). On average, patientscompleted 12 of the 14 diaries at both pre andpost (SD¼ 2).

Focus Group Feedback. After all participantshad completed the intervention, the womenwere invited to a focus group meeting to give

qualitative feedback regarding the Yoga ofAwareness Program. Ten of the 13 participantsattended. Open-ended questions were used toexplore women’s experiences regarding thevarious yoga methods used, the length andnumber of sessions, and the overall program.Women were invited to offer suggestions forhow the program could be better tailored totheir needs.

Focus Group Anonymous Questionnaire. Focusgroup participants were also asked to rate thesuccess of the program on a five-item anony-mous questionnaire. The questionnaire askedpatients to rate, on 10-point Likert-type scales,how helpful the program was overall, how suc-cessful the program was in helping cope withfeelings of distress, how successful the programwas in helping manage pain, how successfulthe program was in helping manage fatigue,and whether they would recommend the pro-gram to a friend who has similar concerns.

Yoga of Awareness ProgramThe intervention consisted of eight weekly

group sessions (four to five patients per group)conducted at the Duke Pain Prevention andTreatment Research Program offices. Thegroups were jointly led by a certified yogateacher (registered with the national Yoga Alli-ance) who holds a master’s degree in health be-havior and education (KMC) and a clinicalhealth psychologist (JWC). Both interventionleaders had received comprehensive trainingin traditional schools of yoga and had extensiveexperience in teaching yoga and meditationtechniques to medical patients and the generalpublic. On average, participants attended sevenof the eight sessions (range, five to eight). Toensure consistency in delivering the interven-tion, a manual was developed to delineate theprogram and provide detailed session guide-lines to be followed by the intervention leaders.All sessions were videotaped and reviewed inweekly treatment team meetings.

Yoga of Awareness is an innovative behavioralintervention specifically designed and tailoredto address patients’ pain, fatigue, and emo-tional distress. The intervention is based in theancient Indian discipline of yoga (meaning‘‘yoking’’ or ‘‘union’’). During recent years, asthe physical exercises of yoga have become pop-ularized in Western countries, the term yoga in


common usage has largely become synonymouswith this single aspect of the fuller discipline.Yoga in actuality comprises a wide variety ofmethods and approaches.27,52 The Yoga ofAwareness intervention is a comprehensiveyoga program that systematically integratesa broad spectrum of traditional yogic tech-niques and tenets. Each 120-minute session in-cluded gentle physical stretching postures(asanas, e.g., seated forward folds, supine lateraltwists) complemented by breathing exercises( pranayama, e.g., extended exhalation, breath-ing into sensation), meditation techniques(dhyana, e.g., awareness of breath, awarenessof awareness itself), study of pertinent topics(swadhyaya, e.g., themes such as the value ofwatching oneself in one’s daily life with the in-tention to understand rather than to judgeand of maintaining one’s poise even amid thetumult of ever-changing challenges), and groupdiscussions (satsang, e.g., discussion of experi-ences of practicing yoga at home, changes incancer-related symptoms during the week). Pa-tients were supplied with a yoga mat, a blanket,CDs/audiotapes, and illustrated handbooks toguide them in home practice. Participantswere encouraged to spend at least 10 minutesa day practicing yoga strategies on their own,and applications of yoga to daily living were as-signed each week (e.g., acceptance during inter-vals of pain).

During sessions, the intervention leadersemphasized the importance of gentle posturepractice when one’s body is challenged bychronic illness, and instructions were modifiedto appropriately address individual patientneeds (e.g., backache). Although the yoga pos-tures used in this study presented no more riskof permanent injury than is associated with ev-eryday activities such as climbing stairs orkneeling down to pick something up, nonethe-less, as a precaution, a physician assistant ornurse was present during all sessions to ad-dress any medical concerns that may havearisen. However, their medical services werenever needed. [Please contact the first authorfor a full description of the yoga posturesused in the study.]

ResultsBecause of the preliminary nature of this

study, analyses of outcome measures were

based on data from study completersonly.41,53 A series of regression and Chi-squareanalyses comparing study completers andthose who withdrew (either prior to or soon af-ter beginning the intervention) indicated nosignificant differences in demographic charac-teristics. However, a significant effect wasfound for mean baseline scores on fatigue(F[1, 19]¼ 8.26, P< 0.01). Those who with-drew were likely to have lower scores on fatigue(M¼ 46.92 for completers vs. 26.15 for non-completers). There were also findings ap-proaching significance for noncompleters tohave higher baseline scores on relaxation(M¼ 45.21 for completers vs. 63.16 for non-completers, F[1, 19]¼ 3.89, P¼ 0.06) and in-vigoration (M¼ 39.99 for completers vs.54.58 for noncompleters, F[1, 19]¼ 3.19,P¼ 0.09) and lower scores on pain(M¼ 33.92 for completers vs. 21.53 for non-completers, F[1, 19]¼ 2.65, P¼ 0.12). Notethat statistical power for these tests was lowdue to the small sample size.

Approach to Multilevel Data AnalysesThe analyses for this study are based on a set

of recently developed statistical procedurescalled multilevel modeling.54 Multilevel mod-eling is an advanced methodology for integrat-ing data from multiple levels of sampling, suchas this study’s two levels (within-person and be-tween-persons). Multilevel models are particu-larly advantageous in analyzing data sets withmany repeated measures, such as daily-diary re-cords.48,53 By preserving the rich detail in eachindividual’s full data set, multilevel models al-low for a sensitive independent determinationof day-to-day interrelated happenings for eachpatient, as well as aggregation of individual es-timates for reliable results for the average pa-tient. The SAS Proc Mixed procedure55

produced parameters in the formof unstandardized maximum likelihoodestimates (b coefficients). These are partialcorrelations, adjusted for between-person dif-ferences, which serve as effect size estimatesof magnitude and direction of changes in de-pendent variables associated with changes inindependent variables.56 Multilevel models al-low for strict control for potential confounds,such as serial autocorrelation in measurements.For all analyses, we allowed intercepts to varyrandomly, thus allowing us to generalize the


findings to the population of persons fromwhich the sample was taken and the populationof observations from which their daily reportswere drawn.46 [For a more complete descrip-tion of the multilevel equations reportedherein, please contact the first author.]

Treatment Effects on Daily OutcomesTo examine treatment effects, models tested

whether patients’ intercept levels for dailypain, fatigue, distress, invigoration, accep-tance, and relaxation changed significantlyacross time from the pre to the post recordingperiods.46,53 Table 1 shows the outcomes forthe effect of treatment (Time). Despite thesmall sample size, the results demonstrated sig-nificant improvements in daily invigorationand acceptance, along with trends for improve-ment in pain and relaxation.

Length of Yoga Practiceand Same Day Outcomes

On 71% (SD¼ 25%) of post diaries (col-lected during the last 14 days of the interven-tion), patients reported spending some timeformally practicing yoga techniques. On aver-age, they reported practicing for 21 minutesper day (SD¼ 11). Analyses examined whetherthe number of minutes spent in yoga practice

Table 1Multilevel Random Effects Estimates

for Baseline Intercept and Effectof Treatment (Time)

Predictor b t P

Daily painBaseline intercept 34.62 8.38 <0.01a

Treatment (time) �3.26 �1.72 0.10b

Daily fatigueBaseline intercept 46.77 11.81 <0.01a

Treatment (time) �3.29 �1.45 0.16

Daily distressBaseline intercept 35.49 10.09 <0.01a

Treatment (time) 1.66 0.78 0.44

Daily invigorationBaseline intercept 39.46 12.46 <0.01a

Treatment (time) 6.96 3.44 <0.01a

Daily acceptanceBaseline intercept 58.73 11.71 <0.01a

Treatment (time) 4.27 2.69 <0.02c

Daily relaxationBaseline intercept 44.15 13.20 <0.01a

Treatment (time) 3.74 1.74 0.09b

aP # 0.01.bP # 0.10.cP # 0.05.

were predictive of same-day levels of diary vari-ables. These models controlled for individuals’mean levels of yoga practice, and practice rateswere person-centered to control for potentiallyspurious within-person associations.56

Table 2 presents the results of these tests.Same-day tests indicated greater yoga practicewas significantly associated with decreasedpain, increased invigoration, and increased ac-ceptance. Trends were also present for greateryoga practice to be associated with decreasedfatigue (P¼ 0.07) and increased relaxation(P¼ 0.07).

Length of Yoga Practice and LaggedDay Outcomes

The preceding same-day analyses did not ad-dress the question of temporal precedenceand hence cannot be used to make causal in-ferences. Yoga practice potentially could haveinfluenced levels of pain, invigoration, and ac-ceptance, or the reverse could be true. There-fore, to clarify whether increases in yogapractice preceded and may have had a causa-tive influence on day-to-day fluctuations in out-come variables, tests were conducted for lagsof one and two days’ practice. In laggedmodels, along with controlling for individuals’mean levels of practice, the lagged day’s levelof the dependent variable was also includedas a within-person control variable (e.g., thepresent day’s pain when predicting next day’spain). As before, practice rates were person-centered to control for any spurious within-person associations.

The results for next-day lagged associationsare shown in Table 2. Increased yoga practicewas significantly predictive of improved levelsof next-day pain, fatigue, invigoration, accep-tance, and relaxation. For invigoration and ac-ceptance, there were also trends for improvedlevels on the second day (not shown in Table 2;for invigoration, b¼ 0.11, t¼ 1.82, P¼ 0.07; foracceptance, b¼ 0.08, t¼ 1.64, P¼ 0.10).

Focus Group FeedbackPatients’ responses to open-ended questions

during the focus group meeting revealeda widespread consensus that the number andlength of the yoga sessions, the techniques im-parted, and the overall content of the programwere appropriate and very useful. Commentsincluded ‘‘The yoga program got me through


Table 2Multilevel Random Effects Estimates for Associations of Length of Yoga Practice with Day-to-Day Outcomes

Variable

Same Day Outcomes Next Day Outcomes

b t P b t P

Daily pain �0.15 �2.71 <0.01a �0.13 �2.19 0.03b

Daily fatigue �0.11 �1.81 0.07c �0.13 �1.96 0.05b

Daily distress �0.04 �0.60 0.55 0.01 0.04 0.97Daily invigoration 0.16 2.99 <0.01a 0.21 3.41 <0.01a

Daily acceptance 0.11 2.54 <0.02b 0.11 2.34 0.02b

Daily relaxation 0.11 1.83 0.07c 0.14 2.18 0.03b

aP # 0.01.bP # 0.05.cP # 0.10.

a bad time and I am still applying the life skillsI learned. I’m finding how to keep my balanceeven when the waves get rough,’’ and ‘‘Theprogram really helped me become more awareof my body and to adjust to the changes thatwere happening. I learned to be more kindto myself.’’ Yet another woman said, ‘‘This pro-gram was extremely helpful. It made me real-ize how uptight I had become and how todeal with it. It also helped me deal withpain.’’ Women also remarked on the value ofgoing through the yoga training in the com-pany of other women who were contendingwith the unique challenges of MBC, as op-posed to a more heterogeneous group: ‘‘Itwas great to be part of a group [that] has thesame problems and concerns and we couldtruly understand one another’s feelings.’’ Re-garding suggestions for how the yoga trainingcould be improved, several women indicatedthat it would be helpful if the program couldbe extended to include ongoing monthlyyoga classes, as this would allow them to con-tinue to build on their yoga practice and toshare this experience with other women withMBC.

Means of Focus Group AnonymousQuestionnaire

The means for the five items, all rated on 10-point scales, of the focus group anonymousquestionnaire were as follows: how helpfulthe program was overall, M¼ 9.6 (range,8e10); how successful the program was inhelping cope with feelings of distress, M¼ 9(range, 7e10); how successful the programwas in helping manage pain, M¼ 8 (range,6e10); how successful the program was inhelping manage fatigue, M¼ 7.6 (range,

6e9); and whether they would recommendthe program to a friend who has similar con-cerns M¼ 10 (all scores were 10). Thus, pa-tients’ scores indicated that they generallyperceived the program as very helpful in ad-dressing the concerns targeted by theintervention.

DiscussionThis pilot study examined the impact of

a novel, yoga-based palliative intervention ina sample of MBC patients. Our findings,though very preliminary, suggest the interven-tion was helpful in significantly boosting dailyinvigoration and a sense of acceptance in pa-tients. There were also trends for improve-ments in pain and relaxation. Moreover,a dose/response relationship was observed be-tween day-to-day variations in patients’ lengthof yoga practice and their daily symptomsand processes. Patients who practiced yogalonger on a given day were much more likelyto experience lower pain and greater invigora-tion and acceptance that same day. Further-more, patients who practiced longer ona given day were also much more likely to ex-perience lower pain and fatigue and greater in-vigoration, acceptance, and relaxation on thenext day. Although yoga has been used for cen-turies to treat disease in the East,20 only re-cently have researchers begun to demonstrateyoga’s effects on patients with cancer27e29

and various other conditions (e.g., osteoarthri-tis, low back pain, multiple sclerosis57e59).Heretofore, no studies have reported on yoga’seffects in MBC patients. This study providessome of the first, tentative evidence for yoga’s


potential benefits in this vulnerable popula-tion of women with limited life expectancy.

Because of the absence of a control group, itis difficult to directly compare the magnitudeof our findings with those of other palliative in-tervention studies in MBC patients. As notedabove, however, significant findings with thispopulation have often consisted of less in-crease in symptoms in the intervention vs.control conditions rather than actual improve-ments.12,14,16,17,19 So, while the absence ofa control group does not allow us to test fordifferential decline in symptoms, the fact thatthe treatment-related coefficients for dailypain, fatigue, invigoration, acceptance, and re-laxation (with the sole exception of dailydistress) were in the direction of true improve-ments has important implications for thepotential efficacy of the intervention.

Beyond examining the effects of the yoga-based intervention, this study marks the firsttime multilevel modeling has been applied todaily-diary outcomes in MBC patients. Impor-tantly, the present multilevel treatment resultswere obtained by first calculating independentestimates for each participant and then aggre-gating them to derive reliable results for the av-erage patientdthus avoiding the problem ofoverlooking the impact of individual differ-ences, as in standard regression approaches.Moreover, the advantages offered by this statis-tical approach were particularly well suited forthe analysis of real-time processes that havestrong causal implications. The tangible im-pact of yoga practice was highlighted by thefinding that greater practice on a given daywas associated with improvements not onlyon the same day, but the next day as well. Fu-ture studies could profit from using dailydata collection to examine more refined hy-potheses about changes in patients’ symptoms(e.g., would yoga lead to same-day or next-dayfatigue becoming more resilient to the nega-tive impact of a chemotherapy session60).

Attrition in this study was similar to previouspalliative trials for MBC.9,14,16,19 Analysesshowed that completers tended to be patientswho were experiencing higher levels of fatigueand other outcomesdpresumably becausethese women were more highly motivated bythe possibility of ameliorating their symptoms.In contrast, in the only previous psychosocialintervention study to report on a comparison

of completers and noncompleters, a trial ofcognitive-behavioral group therapy, Edelmanet al. found the opposite: the noncompleterstended to be patients with greater symptomlevels (e.g., higher anxiety, lower vigor).9

The Yoga of Awareness intervention weused, which comprehensively integratesa wide spectrum of ancient yoga techni-quesdpostures, breathing exercises, medita-tion, study of guiding tenets, and groupdiscussionsdstands in strong contrast to previ-ous palliative interventions with MBC patients,such as cognitive-behavioral group therapyand supportive-expressive group therapy.8e17

This intervention also contrasts with someyoga trainings that have been used with cancerpatients, which emphasize only one compo-nent of the yoga system, that is, the posture ex-ercises that have become widely popularized inWestern countries (e.g., Culos-Reed et al.28).Notably, however, only a minority of the partic-ipants in our study had ever practiced yoga ormeditation before, and among this minority,only a few had more than minimal exposureto these methods. Despite this unfamiliarity,our findings clearly demonstrated the feasibil-ity of conducting future studies of the Yoga ofAwareness Program with MBC patients. It isimportant to note that four of the 13 womenin our sample experienced new cancer recur-rences during the study, and the physicalcondition of several others was noticeably dete-riorating. Nonetheless, no adverse events wererelated to yoga participation. It is likely thatthe gentle approach to yoga postures weused, including careful monitoring of teachingmethods in weekly treatment team meetings,was vital to making the yoga program safe forthese women. In addition, attendance at yogasessions during the study was good (an averageof seven of eight sessions), as was adherence toyoga practice (an average of 21 minutes perday). Further confirmation of the feasibilityof offering this program to MBC patientscame from the focus group meeting. Alongwith giving the program high ratings on theanonymous questionnaire, participants’ com-ments confirmed the suitability of the yoga in-tervention’s content and the number andlength of yoga sessions. The only drawback pa-tients noted was that the program did not leadto an ongoing opportunity to practice yoga inthe company of other women with MBC.


Future studies should consider ways to createa structure for continuing specialized yogaclasses past the study termination. Finally, pa-tients’ anecdotal reports revealed that theyoga training appeared to be very useful inhelping them better adjust to the challengesof living with metastatic disease, including‘‘keeping their balance even when the wavesget rough.’’

Several important limitations of our studyshould be noted. The generalizability of thesepreliminary findings is restricted by the verysmall sample, the absence of a control group,and the lack of follow-up data. To clearly estab-lish the efficacy of the Yoga of Awareness Pro-gram, a controlled (e.g., support groupcontrol), well-powered trial including follow-up evaluations is needed. Further methodo-logical improvements for such a study couldinclude supplementation of self-report databy other types of measures (e.g., immune re-sponse,61,62 physical fitness28), formal analysesof treatment integrity,63 and analyses of predic-tors of treatment outcome (e.g., is yoga morehelpful for patients with certain characteris-tics). Future studies can also seek to determineminimum amounts of effective yoga practice,and given the dose/response effect observed,how adherence can be bolstered in thosewho practice less.

In conclusion, the findings of this pilot studyprovide promising preliminary support for thepalliative health benefits of yoga in MBC pa-tients. The improvements that have beendocumented, and the potential impact theseimprovements may have on patients’ adjust-ment during the remaining years of their lives,are important enough to warrant further study.

References1. Leonard RC, Rodger A, Dixon JM, Dixon JM.

Metastatic breast cancer. In: Dixon M, ed. ABC ofbreast diseases, 2nd ed. London: BMJ Books, 2000:65e71.

2. Aranda S, Schofield P, Weih L, et al. Mappingthe quality of life and unmet needs of urban womenwith metastatic breast cancer. Eur J Cancer Care2005;14:211e222.

3. Cella D. Factors influencing quality of life incancer patients: anemia and fatigue. Semin Oncol1998;25:43e46.

4. Patrick DL, Ferketich SL, Frame PS, et al.National Institutes of Health State-of-the-science

conference statement: symptom management incancer: pain, depression, and fatigue. J Natl CancerInst 2003;95:1110e1117.

5. Rowland JH, Massie MJ, Holland JC. Breast can-cer. In: Holland JC, Breitbart W, Jacobsen PB,Lederberg MS, eds. Psycho-oncology. New York:Oxford University Press, 1998: 380e401.

6. Foley KM, Arbit E, DeVita VT, Hellman S,Rosenberg SA. Management of cancer pain. In:DeVita VT, Rosenberg SA, Hellman S, eds. Princi-ples and practice of oncology, 3rd ed. Philadelphia:J.B. Lippincott, 1989: 2064e2087.

7. Osoba D, Slamon DJ, Burchmore M, et al. Ef-fects on quality of life of combined trastuzumaband chemotherapy in women with metastatic breastcancer. J Clin Oncol 2002;20:3106e3113.

8. Cunningham AJ, Edmonds CV, Jenkins GP, et al.A randomized controlled trial of the effects ofgroup psychological therapy on survival in womenwith metastatic breast cancer. Psychooncology1998;7:508e517.

9. Edelman S, Bell DR, Kidman AD. A group cog-nitive behaviour therapy programme with metastaticbreast cancer patients. Psychooncology 1999;8:295e305.

10. Edelman S, Lemon J, Bell DR, Kidman AD. Ef-fects of group CBT on the survival time of patientswith metastatic breast cancer. Psychooncology1999;8:474e481.

11. Edmonds CV, Lockwood GA, Cunningham AJ.Psychological response to long-term group therapy:a randomized trial with metastatic breast cancerpatients. Psychooncology 1999;8:74e91.

12. Bordeleau L, Szalai JP, Ennis M, et al. Quality oflife in a randomized trial of group psychosocialsupport in metastatic breast cancer: overall effectsof the intervention and an exploration of missingdata. J Clin Oncol 2003;21:1944e1951.

13. Classen C, Butler LD, Koopman C, et al. Sup-portive-expressive group therapy and distress in pa-tients with metastatic breast cancer: a randomizedclinical intervention trial. Arch Gen Psychiatry2001;58:494e501.

14. Goodwin PJ, Leszcz M, Ennis M, et al. The effectof group psychosocial support on survival in meta-static breast cancer. N Engl J Med 2001;345:1719e1726.

15. Spiegel D, Bloom JR, Yalom I. Group supportfor patients with metastatic cancer. Arch Gen Psychi-atry 1981;38:527e533.

16. Spiegel D, Bloom JR. Group therapy and hypno-sis reduce metastatic breast carcinoma pain. Psycho-som Med 1983;45:333e339.

17. Spiegel D, Bloom JR, Kraemer HC, Gottheil E.Effect of psychosocial treatment on survival of pa-tients with metastatic breast cancer. Lancet 1989;2:888e891.


18. Edwards AG, Hailey S, Maxwell M. Psychologicalinterventions for women with metastatic breast can-cer. Cochrane Database Syst Rev 2004;CD004253.

19. Headley JA, Ownby KK, John LD. The effect ofseated exercise on fatigue and quality of life inwomen with advanced breast cancer. Oncol NursForum 2004;31:977e983.

20. Telles S, Naveen KV. Yoga for rehabilitation: anoverview. Indian J Med Sci 1997;51:123e127.

21. Alferi SM, Antoni MH, Ironson G, KilbournKM, Carver CS. Factors predicting the use of com-plementary therapies in a multi-ethnic sample ofearly-stage breast cancer patients. J Am MedWomens Assoc. 2001;56:120e123.

22. Eisenberg DM, Davis RB, Ettner SL, et al.Trends in alternative medicine use in the UnitedStates, 1990e1997: results of a follow-up nationalsurvey. JAMA 1998;280:1569e1575.

23. Lee MM, Lin SS, Wrensch MR, Adler SR,Eisenberg D. Alternative therapies used by womenwith breast cancer in four ethnic populations.J Natl Cancer Inst 2000;92:42e47.

24. Richardson MA, Sanders T, Palmer JL,Greisinger A, Singletary SE. Complementary/alter-native medicine use in a comprehensive cancer cen-ter and the implications for oncology. [seecomment]. J Clin Oncol 2000;18:2505e2514.

25. Bower JE, Woolery A, Sternlieb B, Garet D. Yogafor cancer patients and survivors. Cancer Control2005;12:165e171.

26. Brawley LR, Culos-Reed SN, Angove J,Hoffman-Goetz L. Understanding the barriers tophysical activity for cancer patients: review and rec-ommendations. J Psychosoc Oncol 2002;20:1e21.

27. Cohen L, Warneke C, Fouladi RT,Rodriguez MA, Chaoul-Reich A. Psychological ad-justment and sleep quality in a randomized con-trolled trial of the effects of a Tibetan yogaintervention in patients with lymphoma. Cancer2004;100:2253e2260.

28. Culos-Reed SN, Carlson LE, Daroux LM,Hately-Aldous S. A pilot study of yoga for breast can-cer survivors: physical and psychological benefits.Psycho-Oncol 2006;15:891e897.

29. Joseph CD. Psychological supportive therapy forcancer patients. Indian J Cancer 1983;20:268e270.

30. Wood C. Mood change and perceptions of vital-ity: a comparison of the effects of relaxation, visual-ization and yoga. J R Soc Med 1993;86:254e258.

31. McCracken LM, Carson JW, Eccleston C,Keefe FJ. Acceptance and change in the context ofchronic pain. Pain 2004;109:4e7.

32. Clark D. Between hope and acceptance: themedicalisation of dying. BMJ 2002;324:905e907.

33. Kamei T, Toriumi Y, Kimura H, et al. Decreasein serum cortisol during yoga exercise is correlated

with alpha wave activation. Percept Mot Skills 2000;90:1027e1032.

34. Patel C, North WR. Randomised controlled trialof yoga and bio-feedback in management of hyper-tension. Lancet 1975;2:93e95.

35. Raghuraj R, Telles S. Effect of a yoga-based andforced uninostril breathing on the autonomic ner-vous system. Percept Mot Skills 2003;96:79e80.

36. Schell FJ, Allolio B, Schonecke OW. Physiologi-cal and psychological effects of Hatha-Yoga exercisein healthy women. Int J Psychosom 1994;41:46e52.

37. Telles S, Reddy SK, Nagendra HR. Oxygen con-sumption and respiration following two yoga relaxa-tion techniques. Appl Psychophysiol Biofeedback2000;25:221e227.

38. Telles S, Joshi M, Dash M, et al. An evaluation ofthe ability to voluntarily reduce the heart rate aftera month of yoga practice. Integr Physiol Behav Sci2004;39:119e125.

39. Lehrer PM, Woolfolk RL. Principles and prac-tice of stress management, 2nd ed. New York: Guil-ford Press, 1993.

40. Nespor K. Pain management and yoga. Int JPsychosom 1991;38:76e81.

41. Carson JW, Keefe FJ, Lynch TR, et al. Loving--kindness meditation for chronic low back pain: re-sults from a pilot trial. J Holist Nurs 2005;23:1e18.

42. Carson JW, Keefe FJ, Affleck G, et al. A compar-ison of conventional pain coping skills training andpain coping skills training with a maintenance train-ing component: a daily diary analysis of short- andlong-term treatment effects. J Pain 2006;7:615e625.

43. Stone AA, Kessler RC, Haythornthwaite JA. Mea-suring daily events and experiences: decisions forthe researcher. J Pers 1991;59:575e607.

44. Verbrugge LM. Health diaries. Med Care 1980;18:73e95.

45. Bolger N, Davis A, Rafaeli E. Diary methods:capturing life as it is lived. Annu Rev Psychol 2003;54:579e616.

46. Affleck G, Zautra A, Tennen H, Armeli S. Multi-level daily process designs for consulting and clini-cal psychology: a preface for the perplexed.J Consult Clin Psychol 1999;67:746e754.

47. Jaccard J, Wan CK. Statistical analysis of tempo-ral data with many observations: issues for behav-ioral medicine data. Ann Behav Med 1993;15:41e50.

48. Schwartz JE, Stone AA. Strategies for analyzingecological momentary assessment data. Health Psy-chol 1998;17:6e16.

49. West SG, Hepworth JT. Statistical issues in thestudy of temporal data: daily experiences. J Pers1991;59:609e662.


50. Cella DF, Perry SW. Reliability and concurrentvalidity of three visual-analogue mood scales. Psy-chol Rep 1986;59:827e833.

51. Affleck G, Urrows S, Tennen H, Higgins P. Dailycoping with pain from rheumatoid arthritis: pat-terns and correlates. Pain 1992;51:221e229.

52. Feuerstein G. The yoga tradition: Its history, lit-erature, philosophy and practice. Boston: Sham-bala, 2001.

53. Carson JW, Carson KM, Gil KM, Baucom DH.Mindfulness-based relationship enhancement. Be-hav Ther 2004;35:471e494.

54. Kreft I, de Leeuw J. Introducing multilevel mod-eling. London: Sage, 1998.

55. SAS Institute. SAS/STAT software: Changes andenhancements through release 6.12. Cary, NC: SASInstitute, 1996.

56. Bryk AS, Raudenbush SW. Application of hierar-chical linear models to assessing change. PsycholBull 1987;101:147e158.

57. Garfinkel MS, Schumacher HR Jr, Husain A,Levy M, Reshetar RA. Evaluation of a yoga basedregimen for treatment of osteoarthritis of thehands. J Rheumatol 1994;21:2341e2343.

58. Oken BS, Kishiyama S, Zajdel D, et al. Random-ized controlled trial of yoga and exercise in multiplesclerosis. Neurology 2004;62:2058e2064.

59. Williams KA, Petronis J, Smith D, et al. Effect ofIyengar yoga therapy for chronic low back pain.Pain 2005;115:107e117.

60. Irvine D, Vincent L, Graydon JE, Bubela N,Thompson L. The prevalence and correlates of fa-tigue in patients receiving treatment with chemo-therapy and radiotherapy. Cancer Nurs 1994;17:367e378.

61. Giese-Davis J, Sephton SE, Abercrombie HC,Duran RE, Spiegel D. Repression and high anxietyare associated with aberrant diurnal cortisolrhythms in women with metastatic breast cancer.Health Psychol 2004;23:645e650.

62. van der Pompe G, Antoni M, Visser A,Garssen B. Adjustment to breast cancer: the psycho-biological effects of psychosocial interventions. Pa-tient Educ Couns 1996;28:209e219.

63. Waltz J, Addis ME, Koerner K, Jacobson NS.Testing the integrity of a psychotherapy protocol: as-sessment of adherence and competence. J ConsultClin Psychol 1993;61:620e630.

Documents

Yoga for Women with Metastatic Breast Cancer: Results from a Pilot Study