A dose-effect network meta-analysis model with application in antidepressants using restricted cubic splines

Area of Science:

• Pharmacology
• Biostatistics
• Clinical Research

Background:

• Network meta-analysis (NMA) is crucial for comparing multiple interventions but often overlooks drug dosage.
• Ignoring dose-effect relationships in NMA increases statistical heterogeneity and can lead to suboptimal clinical conclusions.
• Understanding how dosage influences treatment effectiveness is vital for personalized medicine and evidence-based practice.

Purpose of the Study:

• To develop and validate novel network meta-analysis models that explicitly incorporate dose-effect relationships using restricted cubic splines.
• To extend existing NMA models to a dose-effect network meta-regression framework, accounting for covariates and class effects.
• To apply these advanced models to a real-world dataset of antidepressant efficacy in depression.

Main Methods:

• Development of a suite of network meta-analysis models incorporating restricted cubic splines to model dose-effect relationships.
• Extension to a dose-effect network meta-regression model to handle study-level covariates.
• Application to aggregate data from 21 antidepressants and placebo for depression, analyzing efficacy across different dosages.

Main Results:

• All antidepressants demonstrated greater efficacy than placebo beyond a specific dose threshold.
• Identified the precise dose at which each antidepressant's efficacy surpasses placebo.
• Estimated the saturation point beyond which increasing antidepressant dosage yields diminishing returns in efficacy.
• Revealed that smaller study sample sizes may inflate the perceived efficacy of certain antidepressants.

Conclusions:

• The proposed dose-effect network meta-analysis models offer a flexible and robust method for analyzing dose-response in multiple interventions.
• These models can help decision-makers identify optimal drug dosages and improve treatment selection for conditions like depression.
• Accounting for dose-effect relationships and covariates enhances the precision and clinical relevance of network meta-analyses.