Statistical Methods for Cohort Studies of CKD: Survival Analysis in the Setting of Competing Risks

Area of Science:

• Nephrology
• Biostatistics
• Clinical Epidemiology

Background:

• Survival analysis is crucial for evaluating outcomes like end-stage renal disease (ESRD) and mortality in chronic kidney disease (CKD) populations.
• Traditional methods assume independent censoring, which is often violated in clinical settings due to competing risks such as death.
• Dependent censoring, where a competing event influences the time to the event of interest, is common in CKD research.

Purpose of the Study:

• To describe clinical scenarios in renal research involving competing risks.
• To review statistical approaches for analyzing survival data with competing risks.
• To compare cause-specific hazards models and the Fine and Gray approach for competing risks analysis.

Main Methods:

• Review of statistical methodologies for handling competing risks in survival analysis.
• Comparison of cause-specific hazards models and subdistribution hazards models (Fine and Gray approach).
• Application of these methods using data from the Chronic Renal Insufficiency Cohort Study, examining fibroblast growth factor 23 and risks of mortality and ESRD.

Main Results:

• Dependent censoring due to competing risks is prevalent in CKD populations.
• Cause-specific hazards models and Fine and Gray models offer different perspectives on risk estimation.
• The choice of method impacts the interpretation of survival data in the presence of competing risks.

Conclusions:

• Accurate survival analysis in CKD requires explicit consideration of competing risks.
• Understanding the assumptions and interpretations of different competing risks models is essential for clinical research.
• Practical recommendations are provided for analyzing and interpreting survival data incorporating competing risks in CKD studies.