Jacobs 2004.

Study characteristics
Methods	Randomized controlled parallel‐group trial.
Participants	52 participants with chronic non‐specific LBP. Settings: setting for trial and location of yoga intervention delivery not described. Country: USA. Recruitment: self‐referral through flyers and posters in inner‐city primary care clinics and advertisements in newsletters for university and medical employees, and healthcare practitioner referrals from clinic conferences of large inner‐city clinics. Inclusion criteria: men and women aged 18–65 years who had made ≥ 3 visits to a health provider for non‐specific mechanical LBP in the previous 12 months, have had pain symptoms for ≥ 6 months and score ≥ 3 out of 10 on the VAS Pain Scale for pain over past week. Exclusion criteria: back pain secondary to malignancy, infectious disease, inflammatory spondyloarthropathies, vertebral fracture or dislocation, acute radicular syndrome, or severe neurologic signs, systemic or visceral causes of pain, any severe concurrent illness, pregnancy, back‐related compensation or litigation, history of back surgery, regular participation (> 1/week) in iyengar yoga for past 3 months, plans to move out of study region within next 9 months, life expectancy ≤ 9 months.
Interventions	Yoga group: (n = 28) 90‐min iyengar yoga classes held twice a week for 12 weeks. Yoga consisted of predefined set of postures from which yoga teacher selected individual poses in varying sequences for each of 23 yoga classes. 28 asanas (postures) selected, including mandatory poses to be practiced daily. Home practice: prescribed for 30 min on 5 days/week. Participants were provided an illustrated pamphlet explaining the poses and a yoga mat, block, belts, and blankets. Waiting list group: (n = 24) group received a "back pain educational booklet" not otherwise specified. Common interventions: usual care continued for both groups. Co‐interventions: no mention of permitted or restricted co‐interventions. Duration and follow‐up: interventions provided for 12 weeks and there was an additional follow‐up at 6 months.
Outcomes	Back‐specific function (RMDQ) at 3 months (unpublished data). Back‐specific function (Oswestry Disability Index) at 3 months (unpublished data). Pain. Depression (CES – Depression) at 3 months (unpublished data). Quality of life – mental (SF‐36 Emotional Well‐Being) at 3 months (unpublished data). Quality of life – physical (SF‐36 Physical Functioning) at 3 months (unpublished data). Other outcomes collected (unpublished data): bothersomeness of back pain during past 4 weeks, mean LBP over past 4 weeks, worst back pain over last 4 weeks, best back pain over last 4 weeks, insomnia, PANAS‐PA, PANAS‐NA, STAIS, STAIT, SF‐36 Physical Role Limitations, SF‐36 Emotional Role Limitations, SF‐36 Energy/Fatigue, SF‐36 Social Functioning, SF‐36 Pain, SF‐36 General Health, biologic markers, healthcare utilization, drug usage.
Notes	Adverse events: no discussion of safety or adverse events. Measurement of expectations or treatment preferences at baseline: (quote) "To better ascertain how clinical response is modulated by baseline expectation, we ascertained baseline expectation of improvement from yoga and found no differences between groups [at baseline] (Table 6)." Unpublished data: Dr Michael Acree e‐mailed a spreadsheet of endpoint data for completers to LSW on 27 April 2015. Funding: not reported.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Used random number generator.
Allocation concealment (selection bias)	Low risk	Allocation by co‐ordinator according to a pre‐established randomization list.
Blinding of participants	High risk	No blinding as waiting list control used; outcomes based on self‐assessment.
Blinding of personnel/providers	High risk	No blinding.
Blinding of outcome assessors	High risk	Participants were not blinded and self‐reported the outcomes. No mention of blinding of those who collected the information on outcomes from the participants.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Total attrition 16% and only completers analysis done. Reasons for attrition not given although attrition was similar across intervention groups.
Selective reporting (reporting bias)	High risk	Outcome data never published. We obtained all retrievable summary data directly from the analysis file by request of the study authors. Some primary and secondary outcomes were mentioned in the study report and not included in the outcome file (e.g. pharmaceutical drug usage for back pain, biologic markers of stress, and healthcare utilization).
Group similarity at baseline	Low risk	Matched on most important factors.
Co‐interventions	Unclear risk	Not possible to assess.
Compliance	Low risk	Quote: "Overall, 64% of participants assigned to receive the immediate yoga intervention attended yoga classes throughout the 3 month intervention period. On average, these participants attended 15 (66%) classes over the 3‐month trial period."
ITT analysis	Unclear risk	Details of analysis not reported.
Timing of outcome assessments	Low risk	Outcome assessment at set points.
Other bias	Low risk	No other biases identified.