Power and sample size calculations for interval-censored survival analysis

Area of Science:

• Biostatistics
• Survival Analysis
• Longitudinal Data Analysis

Background:

• Interval-censored failure time data presents unique challenges for sample size and power calculations.
• Traditional methods may require specialized software or complex statistical approaches.
• Longitudinal studies often generate data where exact failure times are unknown, only intervals.

Purpose of the Study:

• To propose a novel, accessible method for determining sample size and statistical power in studies with interval-censored failure time data.
• To integrate power calculations within the standard framework of parametric survival model fitting.
• To provide a practical tool for researchers designing longitudinal studies with periodic assessments.

Main Methods:

• The proposed method utilizes a parametric survival model fitted to an expanded dataset.
• Easily computed weights are employed within the survival model framework.
• The approach is demonstrated using a Weibull survival model for a two-group comparison, allowing specification of hazard ratios.

Main Results:

• Simulation results confirm the validity and utility of the proposed power calculation method.
• The study quantifies the impact of the number of assessments (visits) and failure interval lengths on statistical power.
• The method is shown to be robust and applicable to various survival and censoring distributions.

Conclusions:

• The developed method offers a straightforward and efficient approach to sample size and power calculations for interval-censored data.
• It leverages standard statistical software, making it widely accessible to researchers.
• The findings provide valuable insights for optimizing study design in longitudinal research involving time-to-event outcomes.