Estimation of intervention effect using paired interval-censored data with clumping below lower detection limit

Area of Science:

• Biostatistics
• Immunology
• Clinical Trials

Background:

• Biomedical research frequently encounters semicontinuous outcome variables with zero-inflated data.
• Interval censoring and lower detection limits (LDL) complicate the analysis of such data, as seen with antibody levels measured by hemagglutination inhibition assays.
• Defining binary 'fold-increase' endpoints from paired measurements is crucial for assessing vaccine efficacy in clinical trials.

Purpose of the Study:

• To develop and validate a statistical model for paired, interval-censored data with clumping below the lower detection limit (LDL).
• To accurately estimate vaccine immunogenicity and intervention effects using a novel two-part random effects model.
• To provide a robust method for analyzing antibody response data in influenza vaccine trials.

Main Methods:

• Introduction of a two-part random effects model tailored for paired interval-censored data with clumping below LDL.
• Utilization of Monte Carlo approximation for estimating the 'fold-increase' endpoint and intervention effects.
• Application of bootstrapping for reliable variance estimation.

Main Results:

• The proposed statistical model effectively handles interval-censored data with clumping below the LDL.
• Monte Carlo approximation provides accurate estimation of key endpoints and intervention effects.
• Simulation studies demonstrate the method's strong performance.

Conclusions:

• The developed two-part random effects model offers a powerful approach for analyzing complex antibody data in vaccine research.
• The method enhances the assessment of vaccine immunogenicity, particularly when dealing with challenging data characteristics.
• The approach is illustrated effectively using real-world influenza vaccine trial data.