. 2022 Nov 18;2022(11):CD010671. doi: 10.1002/14651858.CD010671.pub3

Summary of findings 3. Yoga compared to exercise for chronic non‐specific low back pain.

Yoga compared to exercise for chronic non‐specific low back pain
Patient or population: people with chronic non‐specific low back pain Setting: mix of participants seeking medical care and participants in the community Intervention: yoga Comparison: another exercise intervention such as stretching or physical therapy
Outcomes^a	*Anticipated absolute effects (95% CI)**		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
Outcomes^a	Risk with another active intervention: yoga compared to exercise	Risk with yoga	Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
Back‐specific functional status Assessed with RMDQ Scale 0–24, lower = better, MCID 5 points Follow‐up: 3 months	The mean back‐specific function was 4.5 points (SD 3.8)	The MD was 0.38 points lower in the yoga group (1.33 lower to 0.62 higher)	—	575 (4 RCTs)	⊕⊕⊕⊝ Moderate^b	Yoga probably results in little to no difference in back‐specific function compared with other exercise. The corresponding risk estimated using the SMD was −0.08 (95% CI −0.28 to 0.13).
Pain Assessed with numerical scale 0–100, lower = better, MCID 15 points Follow‐up: 3 months	The mean pain score was 50 points (SD 21)	The MD was 2.68 points higher in the yoga group (2.01 lower to 7.36 higher)	—	326 (2 RCTs)	⊕⊝⊝⊝ Very low^c,d	We are uncertain whether there is any difference in pain.
Clinical improvement Assessed as participant rating that back pain was improved or resolved. Improvement measured differently in each study Follow‐up: 3 months	474 per 1000	460 per 1000 (341 to 621)	RR 0.97 (0.72 to 1.31)	433 (3 RCTs)	⊕⊝⊝⊝ Very low^c,d,e	We are uncertain whether there is any difference in the risk of clinical improvement.
Physical quality of life Assessed with physical health score on the SF‐36. Scale 0–100, higher = better Follow‐up: 3 months	The mean physical quality of life score was 40.1 points (SD 9)	The MD was 1.30 points higher in the yoga group (0.95 lower to 3.55 higher)	—	237 (1 RCT)	⊕⊝⊝⊝ Very low^c,d	We are uncertain whether there is any difference in physical quality of life. The corresponding risk using the SMD was 0.15 (95% CI −0.11 to 0.40).
Mental quality of life Assessed with mental health score on the SF‐36. Scale 0–100, higher = better Follow‐up: 3 months	The mean mental quality of life score was 45.2 points (SD 11.7)	The MD was 1.90 points higher in the yoga group (1.17 lower to 4.97 higher)	—	237 (1 RCT)	⊕⊝⊝⊝ Very low^c,d	We are uncertain whether there is any difference in mental quality of life. The corresponding risk using the SMD was 0.16 (95% CI −0.10 to 0.41).
Depression	—	—	—	—	—	Not reported.
Adverse events Follow‐up: up to 12 months	91 per 1000	84 per 1000 (51 to 139)	RR 0.93 (0.56 to 1.53)	640 (5 RCTs)	⊕⊕⊝⊝ Low^b,d	There may be little to no difference in adverse events between yoga and other exercise.
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; MCID: minimal clinically important difference; MD: mean difference; RCT: randomized controlled trial; RMDQ: Roland‐Morris Disability Questionnaire; RR: risk ratio; SD: standard deviation; SF‐36: 36‐item Short Form; SMD: standardized mean difference.
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

^aWhen there was more than one study for an outcome, we chose the control group mean from the included study that had the most representative population and the greatest weight in the meta‐analysis. For back‐specific function this was Sherman 2011, and for pain and physical and mental quality of life this was Saper 2017. ^bDowngraded one level for risk of performance and detection bias due to no blinding of the intervention. ^cDowngraded two levels for risk of performance and detection bias due to no blinding of the intervention and attrition bias due to loss to follow‐up. ^dDowngraded one level for imprecision: the number of participants or events in the analysis was lower than indicated in GRADE guidance (fewer than 400 participants or fewer than 300 events). ^eDowngraded one level for inconsistency due to unexplained variation between studies (Chi²test statistically significant and I² > 50%).