Cox 2010.

Study characteristics
Methods	Randomized controlled parallel‐group trial.
Participants	20 participants with LBP. Settings: 1 primary care practice. Location of yoga intervention delivery not described. Country: UK. Recruitment: GP records from a single practice. Inclusion criteria: men and women aged 18–65 years, visiting GP in the previous 18 months for LBP, score ≥ 4 on the RMDQ scale, available to attend yoga classes, possess sufficient physical mobility to participate in the intervention. Exclusion criteria: pregnancy, psychosis or recent substance abuse, already participating in yoga, already in a trial for LBP, not currently experiencing an episode of LBP, previous spinal surgery, or clinical indications of serious spinal or neurologic pathology.
Interventions	Yoga group: (n = 10) 12 weekly 75‐min iyengar yoga classes including relaxation and pain‐relieving postures, and poses to improve posture, flexibility, strength, and mobility. Each class had a primary theme (e.g. Sukha – relaxation and comfort). Home practice: participants given a yoga manual and yoga mat, weekly practice handouts and encouraged to practice yoga at home, as well as taught to have better awareness of posture, movement, and correct breathing. Frequency and duration of suggested home practice not described. The Back Book group: (n = 10) control group received a booklet with advice on how to manage LBP (The Back Book. London: The Stationery Office; 2007) and continued their usual care. Common interventions: both the Back Book and the Yoga group received the booklet and usual care. Co‐interventions: no mention of included or excluded co‐interventions. Duration and follow‐up: interventions were provided for 12 weeks and follow‐up ended directly after the end of the intervention.
Outcomes	Back‐specific function (RMDQ) at 4 and 12 weeks. Back pain (Aberdeen Back Pain Scale) at 4 and 12 weeks. Quality of life – mental (SF−12 Mental Component score) at 4 and 12 weeks. Quality of life – physical (SF‐12 Physical Component score) at 4 and 12 weeks. Clinical improvement (number of participants reporting no LBP) at 4 and 12 weeks. Other outcomes collected: EQ‐5D health index, Pain Self‐Efficacy Questionnaire, number of days spent in bed due to LBP, number of days with restricted activity attributed to LBP, and whether medication was used for LBP over the previous 4 weeks.
Notes	Adverse events: no discussion of safety or adverse events. Measurement of expectations or treatment preferences at baseline: none. Unpublished data: Dr Holger Cramer sent LSW standard deviations for the change values, data previously obtained from Dr Helen Tilbrook, on 26 November. Dr Catherine Hewitt sent endpoint data to LSW on 9 March 2016. Funding: York Trials Unit, Department of Health Sciences, University of York.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Used computer‐generated random numbers to randomize participants.
Allocation concealment (selection bias)	Low risk	Randomization carried out by an independent data manager.
Blinding of participants	High risk	No blinding as compared with usual care alone; outcomes based on self‐assessment.
Blinding of personnel/providers	High risk	No blinding of personnel/providers.
Blinding of outcome assessors	High risk	Self‐reported outcomes were collected by mail from the participants, who were not blinded.
Incomplete outcome data (attrition bias) All outcomes	High risk	Attrition > 30% in yoga group and it was unclear how missing data were managed.
Selective reporting (reporting bias)	High risk	Outcome mentioned in methods was not reported in results (days spent in bed). No protocol available.
Group similarity at baseline	High risk	Usual care group were older and had longer duration of back pain.
Co‐interventions	Unclear risk	Not possible to assess; usual care not defined.
Compliance	High risk	Quote: "Of the ten patients allocated to receive yoga, five (50%) did not attend any one session. … Two patients attended two sessions, two patients attended four sessions and one patient attended five sessions."
ITT analysis	Low risk	Authors state ITT analysis was done.
Timing of outcome assessments	Low risk	Outcome assessment at set points.
Other bias	Low risk	No other biases identified.