A comparative review of network meta-analysis models in longitudinal randomized controlled trial

Area of Science:

• Biostatistics
• Evidence Synthesis
• Longitudinal Data Analysis

Background:

• Standard meta-analysis and network meta-analysis (NMA) typically analyze single endpoints.
• Clinical trials often report data at multiple, non-standardized time points and vary in duration.
• Existing NMA models struggle to effectively incorporate this diverse longitudinal data.

Purpose of the Study:

• To review and compare methods for network meta-analysis that incorporate multiple time points from longitudinal studies.
• To evaluate the performance of different statistical approaches for indirect treatment effect comparisons across varying study durations and time points.

Main Methods:

• Review of advanced network meta-analysis techniques designed for longitudinal data.
• Focus on three key methods: Dakin et al.'s mixed treatment comparison, Ding et al.'s Bayesian evidence synthesis, and Jansen et al.'s fractional polynomial approach.
• Illustration of methods through simulations and a real-world data application.

Main Results:

• The reviewed methods extend traditional NMA to handle multiple time points and varying study durations.
• Fractional polynomial methods demonstrated flexibility and suitability for a wide range of applications involving longitudinal data in NMA.
• Comparisons highlighted the strengths and weaknesses of each approach in synthesizing evidence from diverse trial designs.

Conclusions:

• Methods incorporating multiple time points are crucial for robust network meta-analysis when head-to-head comparisons are absent.
• Fractional polynomial modeling provides a powerful and flexible strategy for analyzing longitudinal data in NMA.
• The choice of method should consider the specific characteristics of the available trial data and the research question.