A bias correction in testing treatment efficacy under informative dropout in clinical trials

Area of Science:

• Clinical Trials
• Biostatistics
• Longitudinal Data Analysis

Background:

• Patient dropout is a significant source of missing data in clinical trials.
• Informative patient dropout can introduce bias into treatment efficacy estimations.
• Traditional methods like Observed Case (OC) and Last Observation Carried Forward (LOCF) may not adequately handle informative dropout.

Purpose of the Study:

• To propose a novel grouping method for correcting bias in treatment effect estimation.
• To address informative dropout schemes dependent on unobserved outcomes via random coefficients.
• To provide an asymptotic variance estimator for statistical inference.

Main Methods:

• Development of a grouping method for bias correction in time-saturated treatment effect models.
• Incorporation of an informative dropout scheme dependent on random coefficients.
• Derivation of an asymptotic variance estimator for robust statistical inference.

Main Results:

• The proposed grouping method effectively corrects biases in treatment effect estimation.
• The new method demonstrates improved stability and accuracy compared to OC, LOCF, and Mixed Model Repeated Measurement (MMRM) approaches.
• Simulation studies confirm the superiority of the proposed method in handling informative dropout.

Conclusions:

• The proposed grouping method offers a significant advancement in analyzing clinical trial data with informative patient dropout.
• This approach enhances the reliability of treatment efficacy inferences, crucial for drug development.
• The method provides a more stable and accurate alternative to existing statistical techniques for longitudinal data analysis.