An illness-death multistate model to implement delta adjustment and reference-based imputation with time-to-event...

Area of Science:

• Biostatistics
• Survival Analysis
• Clinical Trial Methodology

Background:

• Treatment policy strategies evaluate therapies irrespective of intercurrent events (ICEs).
• Estimating treatment effects with ICEs is complex when follow-up is interrupted.
• Current methods like multiple imputation (MI) for delta adjustment (DA) and reference-based (RB) imputation have limitations.

Purpose of the Study:

• To present a novel, fully analytical solution for treatment policy estimands in the presence of ICEs.
• To overcome the challenges of subject follow-up after ICEs using DA or RB imputation.
• To provide a more efficient and accurate alternative to existing imputation methods.

Main Methods:

• Utilized an illness-death multistate model with transitions to the event of interest, ICEs, and from ICEs to the event.
• Estimated transition intensity functions using flexible parametric survival models.
• Employed numerical integration and the delta method for calculating marginal survival curves and standard errors (SEs).

Main Results:

• The proposed analytical solution provides a direct estimation of the treatment policy estimand.
• Simulations demonstrate that the analytical approach is more efficient than multiple imputation (MI).
• The analytical solution avoids the standard error misestimation issues associated with Rubin's variance equation in MI.

Conclusions:

• A fully analytical solution for treatment policy estimands with intercurrent events is feasible and effective.
• This novel method offers improved efficiency and statistical accuracy compared to multiple imputation.
• The approach provides a robust framework for handling unobserved transitions in survival analysis.