Five RCTs (1,206 participants, range 73 to 439) were included in the review. The mean quality score was 21 out of 27 points and the authors reported that none of the trials scored less than 50% of the total score. All studies described randomisation and used intention-to-treat analysis. Most studies reported blinding. Other criteria were variably met.
There were no statistically significant differences at 12 weeks or longer in relation to self-reported depression outcomes (five trials, I2=0%; three trials, I2=32.8% at follow-up), interviewer-rated depression outcomes (four trials, I2=32.3%; three trials, I2=5.1% at follow-up), overall improvement (three trials, I2=25%; two trials, I2=43.8% at follow-up) and suicidality (three trials, I2=0%; I2=19.3% at follow-up). Combined treatment showed a statistically significant improvement in relation to impairment outcomes at 12-weeks using the Children's Global Assessment Scale (CGAS) (WMD -2.32, 95% CI -3.91 to -0.74, I2=0%; four trials). There was no evidence of treatment effect at follow-up (three trials, I2=0%).
Fluoxetine resulted in significantly more spontaneous reported suicidal events than combined treatment at 36 weeks (one trial); venlafaxine (one trial) and selective serotonin reuptake inhibitors (SSRIs) (one trial) were also associated with more events. Psychiatric and non-psychiatric adverse events were reported in three trials; the most common were headaches, nausea and tiredness, disinhibition, sedation, insomnia, vomiting/abdominal pain, sleep difficulties and irritability.