A proposed framework to guide evidence synthesis practice for meta-analysis with zero-events studies

Area of Science:

• Biostatistics
• Medical Research Methodology
• Evidence Synthesis

Background:

• Zero-events are a common challenge in meta-analysis.
• Improper handling of zero-events can lead to research waste and flawed healthcare recommendations.

Purpose of the Study:

• To propose a novel framework for classifying meta-analyses involving studies with zero-events.
• To guide researchers in selecting appropriate statistical methods for meta-analysis when zero-events are present.

Main Methods:

• A two-dimensional classification system was developed based on total event counts and the presence of zero-events in study arms.
• The framework was validated using a dataset from Cochrane systematic reviews.

Main Results:

• The proposed framework successfully categorizes meta-analyses with zero-events into six distinct subtypes.
• The classification demonstrated good concordance with a large real-world dataset.
• Existing methods for handling zero-events were mapped to each subtype, along with a 5-step decision principle.

Conclusions:

• The framework offers a structured approach for researchers to select synthesis methods in meta-analyses with zero-events.
• It provides a foundation for developing standardized methodological guidelines for addressing zero-events in meta-analysis.