Conditional power and information fraction calculations at an interim analysis for random coefficient models.

Area of Science:

• Biostatistics
• Clinical Trial Design
• Nephrology

Background:

• Random coefficient (RC) models are vital for analyzing longitudinal data in clinical trials, particularly for estimating change over time.
• Accelerated approval strategies often rely on surrogate endpoints, necessitating robust confirmatory longitudinal endpoints to demonstrate clinical benefit, as seen in immunoglobulin A nephropathy trials.

Purpose of the Study:

• To provide practical methods for calculating conditional power (CP) in random coefficient models during interim analyses.
• To demonstrate the utility of CP and information fraction calculations for optimizing clinical trial design and validating surrogate endpoints for accelerated approval.

Main Methods:

• The paper details the calculation of conditional power (CP) for random coefficient (RC) models using longitudinal data.
• Illustrates methods with practical examples, focusing on the rate of change in estimated glomerular filtration rate (eGFR) slope.

Main Results:

• Demonstrates how conditional power calculations can inform decisions during interim trial analyses.
• Highlights the importance of these calculations for supporting confirmatory endpoints at the time of accelerated approval.

Conclusions:

• Understanding conditional power and information fraction in RC models is crucial for effective clinical trial design.
• These statistical methods offer valuable support for the confirmatory longitudinal endpoint in accelerated approval pathways, particularly in conditions like immunoglobulin A nephropathy.