Twenty-seven RCTs were included in the review (n=1,663 patients). The study quality score ranged from 20 to 42 (out of a maximum of 58).
There was a statistically significant posttreatment improvement in physical activity (Hedges' g 0.471, 95% CI 0.116 to 0.826; four RCTs), pain (Hedges' g 0.189, 95% CI 0.054 to 0.325; 13 RCTs), disability (Hedges' g 0.417, 95% CI 0.179 to 0.655; 12 RCTs), depressive symptoms (Hedges' g 0.279, 95% CI 0.054 to 0.504; 12 RCTs) with self-regulation psychological therapies compared with control. There was no significant difference for anxiety between groups. There was evidence of moderate heterogeneity in the posttreatment analyses for disability (I2=60%) and depressive symptoms (I2=47%).
At follow-up (two to 14 months), there was a significant improvement in physical activity (Hedges' g 0.361, 95% CI 0.058 to 0.665; four RCTs), disability (Hedges' g 0.145, 95% CI 0.002 to 0.288; twelve RCTs) and depressive symptoms (Hedges' g 0.318, 95% CI 0.160 to 0.475; 12 RCTs). The improvement in pain was not statistically significant (Hedges' g 0.127, 95% CI -0.010 to 0.265; 13 RCTs). There was no significant difference in anxiety between groups. Trials that used more self-regulation techniques reduced depressive symptoms and anxiety more than those that used fewer techniques (p≤0.05). Additional results for subgroup and meta-regression analyses were reported.
The funnel plots suggested possible publication bias for physical activity. Fail-safe numbers suggested that publication bias was unlikely for disability but was possible for the other outcomes.