Informative censoring in relative survival.

Area of Science:

• Epidemiology
• Biostatistics
• Cancer Research

Background:

• Administrative censoring in relative survival analysis can introduce bias when age distributions change during a study.
• Existing methods for relative survival estimation may not adequately address this informative censoring problem.
• The impact of informative censoring on cancer survival data is a recognized challenge in epidemiological studies.

Purpose of the Study:

• To review and demonstrate deficiencies in current relative survival estimation methods.
• To propose a novel weighting approach to correct for informative censoring.
• To evaluate the proposed methods using simulations and real-world cancer registry data.

Main Methods:

• Review of existing relative survival estimators, including the Ederer I estimator.
• Development and application of a weighting strategy to adjust for informative censoring.
• Simulation studies and analysis of actual cancer registry data to assess method performance.

Main Results:

• Informative censoring due to administrative study end can significantly bias relative survival estimates.
• The proposed weighting method effectively corrects for informative censoring in both net survival and Ederer I estimators.
• The magnitude of the informative censoring problem was quantified using empirical data.

Conclusions:

• The proposed weighting method provides a robust solution to the informative censoring problem in relative survival analysis.
• Understanding and correcting for informative censoring is crucial for accurate cancer survival estimation.
• Guidance is provided on the practical application and underlying assumptions of reviewed relative survival methods.