Analysis of Randomised Trials Including Multiple Births When Birth Size Is Informative

Area of Science:

• Biostatistics
• Clinical Trials
• Perinatal Research

Background:

• Informative birth size, where outcomes depend on infant number per birth, is common in trials involving singletons and multiples.
• Existing methods for analyzing informative birth size lack comprehensive performance evaluation.
• Understanding these methods is crucial for accurate interpretation of treatment effects in diverse birth cohorts.

Purpose of the Study:

• To evaluate the performance of different statistical methods for handling informative birth size in randomized trials.
• To provide evidence-based recommendations for analyzing trials with both single and multiple births.
• To clarify the interpretation of treatment effects based on chosen analytical approaches.

Main Methods:

• Compared three generalized estimating equation (GEE) approaches: cluster-weighted GEEs, standard GEEs with independence correlation, and standard GEEs with exchangeable correlation.
• Utilized simulation studies to assess method performance under varying conditions, including treatment-by-multiple birth interactions.
• Analyzed a real-world example dataset to compare the strength of evidence for treatment effectiveness.

Main Results:

• Informative birth size significantly affected treatment effect estimates when treatment effects differed between singletons and multiples.
• The choice of GEE method influenced the strength of evidence for treatment effectiveness in the example dataset.
• Standard GEEs with independence correlation yielded an infant-level interpretation, while cluster-weighted GEEs provided a mother-level interpretation.

Conclusions:

• Informative birth size is a critical consideration in randomized trials involving single and multiple births.
• Pre-specification of analysis methods is essential to address potential biases from informative birth size.
• Recommends using standard GEEs with an independence working correlation structure for an infant-level interpretation of treatment effects.