A simulation approach for power calculation in large cohort studies based on multistate models.

Area of Science:

• Epidemiology
• Biostatistics
• Medical Research

Background:

• Accurate power calculations are vital for the success of large epidemiological and clinical studies.
• Existing methods may not adequately address complexities like staggered entry, competing risks, and covariate misclassification.

Purpose of the Study:

• To develop and implement a flexible simulation-based framework for realistic power calculations.
• To compare power between full cohort and nested case-control designs.
• To provide practical sample size recommendations for epidemiological research.

Main Methods:

• A simulation framework using R for power calculations.
• Modeling time-to-event data under competing risks using a six-state Markov model.
• Incorporation of staggered recruitment, individual hazard rates, interaction effects, and covariate misclassification.
• Use of Cox models for data analysis, accounting for left-truncation and right-censoring.
• Simulation of unobserved heterogeneity using frailty terms in Cox models.

Main Results:

• The framework successfully generates realistic sample paths for time-inhomogeneous Markov processes.
• Nested case-control analyses demonstrate comparable power to full cohort analyses under certain conditions.
• Validation against theoretical concepts and real-world scenarios, including the German National Cohort.

Conclusions:

• The proposed simulation framework offers a robust method for power calculations in complex study designs.
• It facilitates accurate sample size determination, enhancing the reliability of epidemiological research findings.
• The R-based template provides a practical tool for researchers to address specific study design challenges.