Ten trials (n=1,700 participants) were included in the review; sample sizes ranged from 36 to 710. Overall trial quality was average; three studies were open, two were single-blind, and five were double blind. Only three trials described the randomisation method. Drop-out rates ranged from 11 to 69% (median 38%). Eight trials followed the intention-to-treat principle. Follow-up ranged from one to two years.
Relapse: Intramuscular depot injection significantly reduced relapse rates compared with oral administration of antipsychotics (RR 0.70, 95% CI 0.57 to 0.87) in ten trials (n=1,672). No significant heterogeneity was observed. The number needed to treat with depot administration of antipsychotics to prevent one relapse was 10 (95% CI 6 to 25).
Rehospitalisation: The analyses showed no significant difference in rehospitalisation between participants who received antipsychotics by depot or orally (RR 0.78, 95% CI 0.57 to 1.05) in seven trials (n=1,476). When trials using different drugs in the depot compared with oral administration groups were excluded, depot was significantly more likely to reduce rehospitalisation
Drop-outs: Significantly fewer patients in the depot group dropped out due to inefficacy of treatment (RR 0.71, 95% CI 0.57, 0.89) in nine trials (n=1,380), but there were no differences between groups for dropouts for any reason, or drop-outs due to adverse events.
Non-adherence: The analyses showed no significant difference between participants who received antipsychotics by depot or orally (RR 0.76, 95% CI 0.37 to 1.56) in five studies (n=1,141)
Sensitivity analysis confirmed the superiority of depot administration for drop-outs and rehospitalisation when a fixed meta-analysis was conducted.
Heterogeneity was investigated as reported for relapse (I2=41%), rehospitalisation (I2=42%) and non-adherence (I2=58%). Heterogeneity was attributed to differences in blinding, outliers, and limited amount of outcome data available.
Funnel plots did not indicate publication bias.