Antipsychotic drug dosing and study discontinuation in schizophrenia: A systematic review and dose-response meta-analysis

High discontinuation rates compromise the effectiveness of treatment regimens for schizophrenia, because consistent medication adherence is essential for the efficacy of antipsychotics. Understanding the relationship between antipsychotic doses and discontinuation rates is important. This study explores this relationship to identify doses that maximize treatment adherence and minimize discontinuation. We systematically searched multiple electronic databases for fixed-dose RCTs assessing 20 antipsychotics in patients with acute exacerbation of schizophrenia and related disorders. We analyzed dose-response relationships using a one-stage dose-response meta-analysis within a frequentist framework, employing restricted cubic splines to model the relationships. The primary outcome was discontinuation for any reason, and secondary outcomes were discontinuation due to inefficacy and side effects. Analysis of 136 trials involving 44,126 participants revealed various dose-response relationships for antipsychotics. For the primary outcome, all-cause discontinuation, amisulpride, cariprazine, olanzapine (Zyprexa), and quetiapine demonstrated U-shaped curves, indicating optimal dosing thresholds where further increases in dosage led to heightened discontinuation rates, possibly due to side effects. Aripiprazole, asenapine, brexpiprazole, clozapine, paliperidone, and risperidone (Risperdal) had plateaus, suggesting no additional benefit from increasing doses beyond specific points. For haloperidol, iloperidone, lumateperone, lurasidone, sertindole, and ziprasidone, the dose-response curves did not reach a plateau within the examined doses. Inefficacy discontinuation curves were similar to total discontinuation. Most discontinuation for side-effects curves showed sharp increases in side-effects associated with higher doses. Dose discontinuation curves varied between the antipsychotics and included U-shaped, monotonic, and hyperbolic patterns. Future studies should consistently present disease-related and side-effect-related dropouts due to adverse events separately.