Optimal survival analyses with prevalent and incident patients

Area of Science:

• Epidemiology
• Biostatistics

Background:

• Period-prevalent cohorts offer cost-effective survival outcome analysis.
• Existing methods lack rigorous quantification of prevalent/incident patient mix impact on statistical inference.
• Current study designs often overlook the statistical implications of relative patient frequencies.

Purpose of the Study:

• To develop an approach for identifying the optimal mix of prevalent and incident patients.
• To maximize precision across the entire estimated survival curve using a flexible weighting scheme.
• To address the gap in methods for quantifying and incorporating patient mix into study design.

Main Methods:

• Development of an optimization approach for patient mix.
• Theoretical derivation of formulas for optimal cohort composition.
• Utilizing weighted log-rank test and Cox proportional hazards models for inference.

Main Results:

• Inference is most powerful with entirely prevalent or incident cohorts.
• Substantial efficiency gains are achievable with the optimal mix of patients.
• Simulations validate the proposed optimization criteria.

Conclusions:

• The proposed methods provide a rigorous framework for optimizing patient recruitment in period-prevalent cohort studies.
• Application to kidney transplant waitlist outcomes demonstrates practical utility.
• This approach enhances precision and efficiency in survival outcome research.