Inverse Probability Weighting for Recurrent Event Models

Area of Science:

• Biostatistics
• Clinical Trials
• Epidemiology

Background:

• Recurrent events are key clinical trial endpoints across many diseases.
• Intercurrent events complicate the interpretation of treatment effects on recurrent outcomes.
• Estimating hypothetical treatment effects, assuming no intercurrent events, is clinically relevant.

Purpose of the Study:

• To propose novel statistical estimators for hypothetical treatment effects in recurrent event data.
• To address confounding from baseline and time-varying covariates influenced by intercurrent events.

Main Methods:

• Utilized inverse probability weighting (IPW) for adjustment.
• Applied IPW to established Lin-Wei-Yang-Ying (LWYY) and negative binomial (NB) models.
• Incorporated adjustments for baseline and internal time-varying covariates.

Main Results:

• Proposed IPW estimators demonstrated superior performance in simulation studies.
• The new methods showed reduced bias and increased power compared to alternatives.
• Accurate estimation of hypothetical treatment effects was achieved.

Conclusions:

• The developed estimators provide a robust method for analyzing recurrent events in the presence of intercurrent events.
• This approach enhances the validity and precision of treatment effect estimation in clinical trials.
• The findings support the use of these methods for more reliable clinical trial data interpretation.