Multi-state network meta-analysis of progression and survival data.

Area of Science:

• Biostatistics
• Clinical Epidemiology
• Health Economics

Background:

• Network meta-analysis synthesizes evidence from multiple randomized controlled trials.
• Overall survival (OS) and progression-free survival (PFS) are key cancer treatment efficacy endpoints.
• Existing methods often analyze OS and PFS separately.

Purpose of the Study:

• To introduce a joint network meta-analysis method for time-to-event outcomes (PFS and OS).
• To enable robust comparison of multiple cancer interventions within a network.
• To facilitate more accurate decision and cost-effectiveness analyses.

Main Methods:

• Utilizes a time-inhomogeneous tri-state Markov model (stable, progression, death).
• Models time-varying transition rates and relative treatment effects using parametric survival functions or fractional polynomials.
• Extracts data directly from published survival curves.

Main Results:

• The method allows joint synthesis of PFS and OS data.
• It relaxes the proportional hazards assumption, accommodating time-varying effects.
• The approach is applicable to networks with more than two treatments.

Conclusions:

• The proposed joint network meta-analysis method offers a flexible and comprehensive approach to synthesizing cancer treatment effectiveness data.
• It improves upon existing methods by jointly analyzing OS and PFS and relaxing restrictive assumptions.
• This methodology can enhance the precision of treatment comparisons and inform healthcare decision-making.