Twenty three trials were included in the review (n=999); 18 RCTs and five controlled trials with randomisation after matching procedures. Drop-outs ranged from 0% to 65%.
Between Groups analyses:
Psychological treatment was associated with a small effect on overall headache variables (10 trials, REM: g=0.35, 95% CI: 0.08, 0.61), but a significant moderate benefit on headache intensity (nine trials, REM: g=0.43, 95%CI: 0.11, 0.74) compared to waiting list controls. Medication use was not affected by treatment (five trials). Sensitivity analysis excluding the school based interventions did not alter the results. Significantly more people in the psychological treatment group showed clinically significant reduction in symptoms (16 trials, REM: g=0.87, 95% CI: 0.57, 1.16) compared to waiting list controls. There were no significant differences between psychological treatment and active control groups on headache variables. There was no evidence of statistically significant heterogeneity.
Within Group Analyses:
Psychological treatment was associated with moderate improvements in frequency (nine trials, REM: g=0.54, 95% CI: 0.29, 0.79) and intensity (eight trials, REM: g=0.54, 95% CI: 0.27, 0.80) of headaches and small improvements in duration of headaches (nine trials, REM: g=0.29, 95%CI: 0.04, 0.54) at end of treatment compared to pre-treatment levels. Participants in the psychological treatment group continued to show improvements on headache variables at follow-up (4 to 12 months) compared to pre-treatment levels (REM: g=1.00, 95%CI: 0.64, 1.33). There was no significant impact on use of medication. Use of medication and headache variables did not significantly change pre-treatment and post-treatment in the waiting list group. There was no evidence of statistical heterogeneity.
Subgroup analyses:
Relaxation and biofeedback treatments were both associated with a greater number of participants experiencing clinically significant levels of symptom reduction compared to waiting list controls: relaxation treatments (seven trials, REM: g=0.80, 95% CI: 0.23, 1.35); biofeedback treatments (four trials, REM: g=0.90, 95% CI: 0.10, 1.68). There was no difference between relaxation or biofeedback treatment groups and controls on headache variables. A subgroup analysis of migraine sufferers found that significantly more participants in the treatment group had clinically significant levels of symptom reduction compared to migraine sufferers in the control group (five trials, REM: g=1.22, 95% CI: 0.61, 1.84).
Publication bias:
Trim-and-fill analysis was applied to the findings to correct for possible publication bias, thereby reducing the effect sizes for between group analysis (headache variables, REM: g=0.23, 95% CI: -0.03, 0.50; clinical significance, REM: g=0.76 95% CI: 0.50, 1.02) and within group analyses (headache variables, REM: g=0.40, 95% CI0.17, 0.61).