Systematic review of long-acting injectables versus oral atypical antipsychotics on hospitalization in schizophrenia

Abstract

Objective:To assess the impact of long-acting injectables (LAIs) versus oral antipsychotics (OAs) on hospitalizations among patients with schizophrenia by conducting a systematic literature review of studies with different study designs and performing a meta-analysis.Methods:Using the PubMed database and major psychiatric conference proceedings, a systematic literature review for January 2000 to July 2013 was performed to identify English-language studies evaluating schizophrenia patients treated with atypical antipsychotics. Studies reporting hospitalization rates as a percentage of patients hospitalized or as the number of hospitalizations per person per year were selected. The primary meta-analysis assessed the percentage decrease in hospitalization rates before and after treatment initiation for matched time periods. The secondary meta-analysis assessed the absolute rate of hospitalization during follow-up. Pooled treatment-effect estimates were calculated using random-effects models. To account for differences in patient and study-level characteristics between studies, meta-regression analyses were used. Subset analyses further explored the heterogeneity across study designs.Results:Fifty-eight studies evaluating 25 arms (LAIs: 13 arms, 4516 patients; OAs: 12 arms, 23,516 patients) in the primary meta-analysis and 78 arms (LAIs: 12 arms, 4481 patients; OAs: 66 arms, 96,230 patients) in the secondary meta-analysis were identified. Reduction in hospitalization rates for LAIs was 20.7 percentage points higher than that of OAs (random-effects estimates: LAIs = 56.2% vs. OAs = 35.5%, P = 0.023). Controlling for patient and study characteristics, the adjusted percentage reduction in hospitalization rates for LAIs was 26.4 percentage points higher than for OAs (95% CI: 3.3–49.5, P = 0.027). As for the secondary meta-analysis, no significant difference between LAIs and OAs was observed (random-effects estimate: −8.6, 95% CI: −18.1–1.0, P = 0.077). Subset analyses across type of study yielded consistent results. Limitations of this analysis include the long observation period, which may not reflect current treatment patterns, the use of all-cause hospitalization, which may not be solely related to schizophrenia, and the fact that most studies in the LAI cohort evaluated risperidone.Conclusion:The primary results of this meta-analysis, including studies with both interventional and non-interventional designs and using meta-regressions, suggest that LAIs are associated with higher reductions in hospitalization rates for schizophrenia patients compared to OAs.