Dose-effect meta-analysis for psychopharmacological interventions using randomised data

Area of Science:

• Psychopharmacology
• Biostatistics
• Clinical Research Methodology

Background:

• Current meta-analyses of psychopharmacological interventions often neglect or limit dose-range analysis, leading to uncertainty and heterogeneity.
• Integrating dose-effect models into meta-analysis can address these limitations.

Purpose of the Study:

• To describe and illustrate one-stage and two-stage dose-effect meta-analysis models.
• To determine the dose-response relationship for selective serotonin reuptake inhibitor (SSRI) antidepressants.

Main Methods:

• Utilized one-stage and two-stage meta-analysis models with common or random effects.
• Applied restricted cubic splines to model dose-effect on an odds ratio scale using data from 60 randomized controlled trials of SSRIs.
• Investigated dose-response linearity.

Main Results:

• A dose-effect curve was estimated, showing response probability increases up to 30 mg/day of fluoxetine-equivalent, achieving a 50% response rate.
• No additional response increase was observed beyond 40 mg/day.
• The one-stage model, incorporating all studies, yielded slightly less uncertainty than the two-stage model.

Conclusions:

• Dose-effect meta-analysis provides clinicians with crucial information on how drug effects vary with dosage.
• The one-stage model is recommended for practical application due to its ability to incorporate all available evidence and reduce uncertainty.