Nineteen studies (n=4,063) were included in the review.
The quality scores ranged from 8 to 13 points, with a mean of 10.8 (standard deviation 1.41). Compliance, inclusion criteria and blinding were rarely mentioned. Studies of combined interventions had significantly lower average quality scores than did studies of other intervention types (p=0.002).
Change from baseline.
The overall effect size for the mean difference in change scores between experimental and control groups was -1.51 (95% CI: -2.29, -0.73). Statistically significant improvements were observed for stress symptoms (14 studies; SMD -0.87, 95% CI: -1.23, -0.50) and coping (12 studies, SMD -3.49, 95% CI: -6.71, -0.28). No significant benefits were observed for (social) behaviour or for self-efficacy. All meta-analyses showed significant heterogeneity (p<0.001).
Final measures only.
Overall, statistically significant improvements were observed for behaviour (9 studies; SMD 0.80, 95% CI: 0.57, 1.04) and coping (9 studies; SMD 1.72, 95% CI: 0.75, 2.68). No significant benefit was observed for self-efficacy.
Publication bias was significant for overall effect size (p=0.015) and the outcomes of stress symptoms (p=0.006) and self-efficacy (p=0.046).