Multiple imputation methods for nonparametric inference on cumulative incidence with missing cause of failure

Area of Science:

• Biostatistics
• Survival Analysis
• Epidemiology

Background:

• Accurate estimation of cumulative incidence is crucial in survival analysis, especially when dealing with competing risks.
• Missing cause-of-failure data presents a significant challenge in analyzing time-to-event data.
• Existing methods may be limited when cause of failure is unknown for a portion of subjects.

Purpose of the Study:

• To propose a novel nonparametric method for estimating cumulative incidence functions with missing cause-of-failure data.
• To provide a robust statistical framework for handling situations where failure causes are unknown or missing.
• To develop methods for confidence interval construction and hypothesis testing in such scenarios.

Main Methods:

• Utilized multiple imputation techniques under the missing at random assumption to estimate the cumulative incidence function.
• Developed asymptotic theory to support the statistical properties of the proposed estimators.
• Incorporated methods for constructing confidence intervals and performing comparative tests between groups.

Main Results:

• Simulation studies demonstrated the strong performance and reliability of the proposed nonparametric imputation-based methods.
• The methods provide accurate cumulative incidence estimates even with substantial missing cause-of-failure data.
• The approach was successfully illustrated using data from a breast cancer clinical trial.

Conclusions:

• The proposed nonparametric approach effectively addresses cumulative incidence estimation with missing cause-of-failure data.
• Multiple imputation offers a viable and statistically sound strategy for handling such data complexities.
• The developed methods are applicable to real-world clinical trial data and survival analysis research.