Injectable Antipsychotics Efficacy in Schizophrenia and Schizoaffective: A Mixed Treatment Comparison

Objective: To assess the comparative effects among long-acting injectable (LAI) antipsychotics (APs) on Positive and Negative Syndrome Scale (PANSS) score in adults with schizophrenia or schizoaffective disorder. Methods: PubMed, Embase, PsycInfo, CENTRAL, and Clinicaltrials.gov were searched on July 10, 2021 for randomized controlled trials (RCTs) ≥8 weeks in duration that reported PANSS at short- (8 to 24 weeks) and long-term (>24 weeks) durations for comparisons of LAI APs with any of the following: placebo, oral APs, or LAI APs. Mixed treatment comparisons were performed using a Bayesian framework. Confidence in the estimates was assessed using the Confidence in Network Meta-Analysis (CINeMA) framework. Results: Thirty trials were included, 20 of which were rated at high ROB due to missing outcome data. Nearly all comparisons were based on very low certainty evidence due to within-study and reporting biases in short-term analyses, and within-study bias, reporting bias, imprecision, heterogeneity, and incoherence in long-term analyses. For short-term effects on PANSS (20 trials; 8,618 patients), all treatments were superior to placebo; differences were not detected among active treatments. The highest-ranking interventions were LAI aripiprazole lauroxil, oral olanzapine, and LAI olanzapine pamoate; however, there was major imprecision (median rank, 95% credible interval [CrI], 2, 1 to 10; 3, 1 to 7; and 4, 1 to 9; respectively). For long-term effects (10 trials; 4,595 patients), no differences were detected in any comparisons with placebo or between active treatments. Median rankings and 95% CrIs were again imprecise and highest for oral risperidone, LAI aripiprazole lauroxil, and a composite control group of any oral AP (2, 1 to 9; 3, 1 to 9; and 3, 1 to 9; respectively). Results were similar in sensitivity analyses excluding trials with a control arm of any oral AP or with a randomized withdrawal design. Meta-regression using baseline PANSS did not improve model fit. Conclusion: Very low-certainty estimates indicated no differences among LAI APs for short or long term effects on PANSS in schizophrenia or schizoaffective disorder. Further study is needed to define comparative efficacy of LAI APs with greater certainty, optimally in trials with active control arms, low attrition, and independent funding.