A framework for prospective, adaptive meta-analysis (FAME) of aggregate data from randomised trials

Area of Science:

• Biostatistics
• Clinical Epidemiology
• Health Research Methodology

Background:

• Most systematic reviews are planned retrospectively, increasing bias risk from prior knowledge and unpublished data.
• Aggregate data from published trials are typically used, potentially leading to reporting biases.
• Prior knowledge of trial results can compromise systematic review and meta-analysis integrity.

Purpose of the Study:

• To present a collaborative framework for prospective, adaptive meta-analysis (FAME) of aggregate data.
• To reduce bias in systematic reviews and meta-analyses.
• To monitor accumulating evidence for timely, definitive meta-analyses.

Main Methods:

• Developed and piloted the Framework for Adaptive Meta-analysis of aggregate data (FAME).
• Refined FAME principles through 4 systematic reviews in prostate cancer.
• Key principles include early review initiation, investigator liaison, prospective timing assessment, protocol registration, data interpretation considering all evidence, and updating assessment.

Main Results:

• The Framework for Adaptive Meta-analysis of aggregate data (FAME) was piloted alongside 4 systematic reviews.
• Principles were refined, demonstrating application via a hypothetical review and 3 published reviews.
• FAME facilitates early identification of optimal timing for definitive meta-analyses.

Conclusions:

• The Framework for Adaptive Meta-analysis of aggregate data (FAME) reduces potential bias.
• FAME enhances the timeliness, thoroughness, and reliability of systematic reviews using aggregate data.
• Prospective and adaptive approaches improve the quality of evidence synthesis.