TOPLINE:

Most antipsychotics achieved near-maximal efficacy within the lower-to-middle range of their recommended doses, around 3-5 mg/d of risperidone equivalents, with little additional benefit observed at higher doses in people with acute schizophrenia spectrum disorders (SSDs), a large dose-response meta-analysis revealed.

METHODOLOGY:

• Researchers conducted a one-stage random-effects dose-response meta-analysis of 131 randomised controlled trials involving 40,715 participants with acute SSDs.
• Participants included adults (mean age, 39.24 years; 31.8% women) and children/adolescents (mean age, 15.68 years; 45.1% girls) from fixed-dose trials, with a median study duration of 6 weeks (range, 3-26 weeks).
• Investigators analysed dose-response curves using restricted cubic splines for 20 individual antipsychotic drugs separately and all drugs combined.
• The primary aim of this meta-analysis was to estimate doses needed to achieve 95% and 50% of the maximal efficacy of the drug, presenting them as ED95 and ED50, respectively.
• Confidence in the evidence was evaluated using the standard Grading of Recommendations Assessment, Development and Evaluation framework adapted to the dose-response meta-analysis.

TAKEAWAY:

• Dose-response curves of most individual antipsychotics with moderate-to-high certainty of evidence were hyperbolic, reaching a plateau within the lower-to-medium approved dose ranges, around 3-5 mg/d of risperidone dose equivalents (risperidone: ED95, 4.0 mg/d; aripiprazole: ED95, 10.2 mg/d; and olanzapine: ED95, 15.7 mg/d).
• Pooled efficacy plateaued at 3-5 mg/d of risperidone equivalents in adults and children/adolescents, with no meaningful gains observed beyond this range (ED50s, 1.3 and 1.1 mg/d, respectively; ED95s, 15.4 and 8.0 mg/d, respectively; Wald test P < .001 for both).
• Maximum recommended doses of several antipsychotics in clinical practice were high, in some cases reaching up to three times the doses associated with near-maximal efficacy, such as for aripiprazole (ED95, 10.2 mg/d; maximum dose, 30 mg/d).
• Low to very low certainty of evidence was found regarding dose-response relationships of amisulpride, blonanserin, cariprazine, clozapine, haloperidol, lumateperone, and ziprasidone, with studies being insufficient for conducting the meta-analysis of olanzapine/samidorphan, xanomeline/trospium chloride, and zotepine.

IN PRACTICE:

"Our analysis provides estimates of the plateau doses, which can serve as general guidance in clinical practice and should be considered alongside the side-effect profiles to help balance the trade-offs involved in treatment decisions," the authors wrote.

"Although dose-response relationships were broadly similar between children/adolescents and adults, considerable interindividual variation can be expected in clinical practice, and multiple participant characteristics may influence the required dose," they added.

SOURCE:

This study was led by Yuki Furukawa, Technical University of Munich, Munich, Germany. It was published online on March 26, 2026, in eClinicalMedicine.

LIMITATIONS:

The one-stage dose-response meta-analytic approach might have introduced heterogeneity due to differences between trials comparing single doses with placebo and those comparing multiple doses. Dose-response relationships for some antipsychotics had low or very low confidence due to the risk for bias and imprecise estimates for several antipsychotics in adults and most individual antipsychotics in children/adolescents. Data included trial averages, and substantial interindividual variability might be present. Data were too sparse to conduct separate dose-response meta-analyses for key subgroups.

DISCLOSURES:

This meta-analysis was funded by grants from the German Research Foundation; the German Center for Mental Health; and the Federal Ministry of Research, Technology and Space. Several authors reported receiving grants and honoraria from the SENSHIN Medical Research Foundation and multiple pharmaceutical companies including but not limited to Angelini, Bristol Myers Squibb, Boehringer Ingelheim, Gedeon Richter, Sun Pharma, and others. One author reported being an employee of Elsevier in the role unrelated to this study, working as a senior editor for The Lancet Public Health.

This article was created using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication.