Bower 2011.

Methods	Study design: single‐centre RCT Number randomised: 31; 16 to intervention, 15 to control Study start: March 2007; stop date: July 2010 Length of intervention: 12 weeks Length of follow‐up: to end of intervention, at 3 months post intervention Country: USA
Participants	Age, years (mean SD): Intervention: 54.4 (5.7) Control: 53.3 (4.9) Stage, n (%): Stage 0‐II Inclusion criteria: • Original diagnosis of stage 0‐II breast cancer • Completed local and/or adjuvant cancer therapy (with the exception of hormone therapy) at least 6 months previously • Ages 40 to 65 years • Postmenopausal • No other cancer in the past 5 years • Experiencing persistent cancer‐related fatigue Exclusion criteria: • Chronic medical conditions or regular use of medications associated with fatigue (e.g. untreated hypothyroidism, diabetes, autoimmune disease, anaemia (defined as haematocrit < 24), chronic fatigue syndrome) • Evidence that fatigue was driven primarily by a medical or psychiatric disorder other than cancer (e.g. current major depression, insomnia, sleep apnoea) • Evidence that fatigue was driven primarily by other non‐cancer‐related factors (e.g. shift work, recent change in activity or schedule) • Physical problems or conditions that could make yoga unsafe (e.g. serious neck injury, unstable joints) • Body mass index (BMI) > 31 kg/m²
Interventions	16 participants assigned to exercise intervention: • Iyengar yoga, a traditional form of Hatha yoga, performed in groups of 4 to 6 women for 90 minutes twice a week for 12 weeks Adherence: Over 80% of participants attending at least 20 of the 24 yoga classes offered. Mean number of classes attended was 18.9 of 24 classes (78%), and median number was 22 of 24 classes (92%). At 3‐month follow‐up, 9 of 14 women who attended the yoga classes (64%) were continuing to use techniques learned in class. Control group: 15 assigned to control: • Health education classes conducted for 120 minutes once a week for 12 weeks (24 hours) in groups of 4 to 7 women. Classes were led by a PhD‐level psychologist with clinical experience. Adherence: In the education group, the mean number of classes attended was 9.2 of 12 classes (77%), and the median number was 11 of 12 classes (92%).
Outcomes	Primary outcome: Subjective fatigue severity assessed with the Fatigue Symptom Inventory (FSI) Secondary outcomes: Vigour assessed by the vigour subscale of the Multi‐dimensional Fatigue Symptom Inventory (MFSI) Depressive symptoms assessed via the Beck Depression Inventory‐II (BDI‐II) Subjective sleep quality assessed by the Pittsburgh Sleep Quality Index (PSQI) Feelings of stress assessed on the Perceived Stress Scale Timed chair‐stands used to assess lower extremity strength and endurance Functional reach test used to assess strength, flexibility, and balance Self‐efficacy for managing fatigue assessed via the fatigue subscale of the Human Immunodeficiency Virus Self‐Efficacy Questionnaire adapted for breast cancer Fatigue interference with activities, mood, and enjoyment of life assessed with the interference subscale of the FSI Numbers of participants assessed: Intervention: n = 16 at baseline, n = 14 post intervention, n = 13 months after intervention Control: n = 15 at baseline, n = 13 post intervention, n = 13 months after intervention .Adverse events: none reported
Notes	Trial registration link: none available Trial authors contacted: no Intention‐to‐treat analysis: yes Funding: National Center for Complementary and Alternative Medicine/National Institutes of Health (NCCAM/NIH U01‐AT003682; Iyengar Yoga for Breast Cancer Survivors with Persistent Fatigue)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	"Allocation sequence was generated independently by the study statistician", but it is unclear how the allocation sequence was generated.
Allocation concealment (selection bias)	Unclear risk	"Allocation was concealed in opaque envelopes" but whether "sequential" is not mentioned.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Owing to the nature of the intervention, it was not possible to blind participants; however, it is unclear whether the outcome was influenced by lack of masking.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	"Outcomes assessors for the performance tasks were blinded to group assignment, and all were trained in standardized testing procedures".
Incomplete outcome data (attrition bias) All outcomes	Low risk	"All statistical analyses were performed on an intent‐to‐treat basis". Mixed model analysis was used to account for missing data.
Selective reporting (reporting bias)	Low risk	No selective reporting of outcomes is apparent.
Other bias	Low risk	Trial appears to be free of other problems that could put it at high risk of bias.