Meta-Analyses Proved Inconsistent in How Missing Data Were Handled Across Their Included Primary Trials: A...

Area of Science:

• Medical Research Methodology
• Biostatistics
• Evidence Synthesis

Background:

• Handling missing data in systematic reviews is a critical but inconsistently applied aspect of research methodology.
• Uncertainty exists regarding how systematic review authors address missing data in primary studies.

Purpose of the Study:

• To evaluate the consistency of systematic review authors in handling missing data within meta-analyses.
• To assess the alignment between reported and actual methods for addressing missing data in systematic reviews.

Main Methods:

• Systematic review of 100 reviews with significant meta-analyses of dichotomous outcomes.
• Extraction and comparison of statistical data from 638 primary trials against meta-analysis data.
• Classification of "analytical method actually used" for missing data and assessment of reported methods.

Main Results:

• "Analytical methods actually used" for missing data could not be determined for 397 trials.
• Complete case analysis (53%) and "none had the event" assumption (24%) were common.
• Only 8% of reviews were consistent across trials; none reported methods. Of 7 reviews reporting methods, only 1 was consistent, and its approach contradicted its reported method.

Conclusions:

• Most systematic review authors demonstrate inconsistency in handling and reporting missing outcome data.
• Authors should clearly identify, specify, and consistently apply methods for addressing missing data in primary trials.