Sample size determination for the association between longitudinal and time-to-event outcomes using the joint...

Area of Science:

• Biostatistics
• Clinical Trials
• Longitudinal Data Analysis

Background:

• Joint models (JMs) integrate longitudinal and time-to-event data, offering improved bias reduction and efficiency for clinical trial analysis.
• Wider adoption of JMs necessitates further development in design considerations, particularly for sample size calculations.

Purpose of the Study:

• To extend existing joint model sample size formulas by incorporating a time-dependent slopes parameterization.
• To provide a more accurate sample size formula for testing the association between longitudinal outcomes and time-to-event data when slopes influence hazard.

Main Methods:

• Developed an extended closed-form sample size formula for joint models using the time-dependent slopes parameterization.
• Compared the proposed formula's accuracy against the existing method by Chen et al. in scenarios where longitudinal slopes affect event hazard.
• Applied the proposed method to calculate sample size for a biomarker qualification study in Hutchinson-Gilford progeria syndrome.

Main Results:

• The proposed formula, utilizing time-dependent slopes, provides a more accurate estimate of statistical power when both the current value and the slope of the longitudinal outcome influence the time-to-event hazard.
• The study demonstrates the practical application of the extended formula in a real-world scenario involving an ultra-rare disease.

Conclusions:

• The developed sample size formula enhances the precision of power calculations for joint models, particularly when longitudinal changes (slopes) impact event risk.
• This advancement supports more robust study designs in clinical trials, especially for rare diseases where efficient resource allocation is critical.