Eleven RCTs were included (n=1,708 participants, range 50 to 547): four were of early intervention (n=800), four were of CBT (n=620) and three were of family intervention (n=288). It appeared that all the included studies were judged to be of adequate quality, but no details or summary information on study quality were provided in the article.
Early intervention services significantly reduced the risk of relapse (RR 0.66, 95% CI 0.47 to 0.94, NNTB=6; two RCTs), positive symptoms (SMD -0.21, 95% CI -0.42 to 0.01; two RCTs), negative symptoms (SMD -0.39, 95% CI -0.57 to -0.20; two RCTs), trial discontinuation (RR 0.71, 95% CI 0.53 to 0.94, NNTB=8; four RCTs), loss of contact with services (RR 0.60, 95% CI 0.39 to 0.92, NNTB=13; two RCTs) and having no psychological intervention (RR 0.67, 95% CI 0.46 to 0.97, NNTB=5, I2=74%, p=0.02; four RCTs) significantly more than standard care alone.
CBT reduced positive symptoms (SMD -0.60, 95% CI -0.79 to -0.41; three RCTs) and negative symptoms (SMD -0.45, 85% CI -0.80 to -0.09, I2=62%; three RCTs) at up to two years post treatment significantly more than standard care alone. There was no significant difference between the groups for other outcomes: symptoms at end of treatment (four RCTs), relapse within two years (two RCTs, I2=79%, p=0.03) and hospital admission (two RCTs).
Family intervention reduced the risk of relapse after treatment and of hospital admission (RR 0.50, 95% CI 0.32 to 0.80, NNTB=7; three RCTs) significantly more than standard care alone. There was no significant difference between the groups for these outcomes measured separately at end of treatment or at up to two years' follow-up.
No data were reported on duration of untreated psychosis.
No statistically significant heterogeneity was found except where noted above.