Current practice in analysing and reporting binary outcome data-a review of randomised controlled trial reports

Area of Science:

• Clinical Trials
• Biostatistics
• Medical Research Reporting

Background:

• Randomized controlled trials (RCTs) with binary outcomes present unique reporting challenges.
• Different analytical methods can lead to varied interpretations of treatment effects.
• Sensitivity to assumptions about missing data can impact results.

Purpose of the Study:

• To review methods used for analyzing binary primary outcomes in RCTs.
• To assess how missing data were handled in these trials.
• To evaluate the reporting of results from RCTs with binary outcomes.

Main Methods:

• Systematic review of RCT reports published in January 2019 with a binary primary outcome.
• Inclusion of English language papers from PubMed across various medical research areas.
• Focus on individually randomized, parallel-group RCTs.

Main Results:

• 64% of 200 RCT reports used chi-squared-style tests for primary analysis.
• Only 55% reported a treatment effect measure; 38% reported only a p-value.
• Missing data handling was often inadequate, with available case analysis being most common (69%).
• 12% of articles reported appropriate sensitivity analyses for missing data.

Conclusions:

• Substantial improvement is needed in statistical analysis and reporting of treatment effects for RCTs with binary endpoints.
• Incomplete reporting of treatment effects hinders understanding and dissemination of findings.
• Published trials often lack clear descriptions of missing data and sensitivity analyses.