Modelling batched Gaussian longitudinal weight data in mice subject to informative dropout

Area of Science:

• Statistics
• Biostatistics
• Longitudinal Data Analysis

Background:

• Longitudinal data analysis is crucial for understanding changes over time.
• Informative dropout, where data loss is related to the outcome, complicates analysis.
• Batch data collection, instead of individual, presents unique modeling challenges.

Purpose of the Study:

• To develop statistical models for longitudinal data with informative dropout when outcomes are collected in batches.
• To compare shared parameter and pattern mixture models for this specific scenario.
• To analyze a mouse weight study comparing control and treatment groups.

Main Methods:

• Development of shared parameter and pattern mixture models.
• Application to a mouse weight study with batch measurements.
• Simulation studies to evaluate model performance and robustness.

Main Results:

• Treatment group mice exhibited lower mid-life weights.
• Treatment group mice demonstrated a slower rate of weight decline in later life.
• Both models performed well under correct specification; pattern mixture models were more robust to misspecification.

Conclusions:

• The developed shared parameter and pattern mixture models effectively handle longitudinal data with informative dropout and batch collection.
• Pattern mixture models offer greater robustness against model misspecification.
• Shared parameter models provide more direct interpretation of results.