The network meta-analytic-predictive approach to non-inferiority trials

Area of Science:

• Clinical Trials Methodology
• Biostatistics
• Evidence Synthesis

Background:

• Non-inferiority clinical trials compare a test treatment against an active-control, not placebo.
• These designs are employed when placebo use is ethically problematic or impractical.
• A key challenge is assessing the test treatment's potential efficacy versus placebo.

Purpose of the Study:

• To introduce and illustrate a network meta-analytic-predictive approach for non-inferiority trials.
• To indirectly estimate the test treatment's effect against placebo using historical data.
• To enhance the interpretation of non-inferiority trial results.

Main Methods:

• A network meta-analysis integrates data from historical and current non-inferiority trials.
• The approach predicts the hypothetical test treatment versus placebo effect.
• It incorporates between-trial variability for all parameters, focusing on trial-specific effects.

Main Results:

• The network meta-analytic-predictive approach provides an indirect estimation of test treatment efficacy against placebo.
• The method accounts for variability across different trials.
• Illustrative examples with binary outcomes demonstrate the approach's application.

Conclusions:

• The network meta-analytic-predictive approach offers a robust method for evaluating test treatments in non-inferiority trials.
• This statistical framework addresses the critical question of potential superiority over placebo.
• It enhances the evidential basis for clinical decisions when placebo controls are not feasible.