Graphing survival curve estimates for time-dependent covariates.

Area of Science:

• Biostatistics
• Statistical analysis

Background:

• Graphical representations enhance interpretation of statistical findings, particularly in survival analysis.
• Kaplan-Meier estimates are standard for fixed categorical covariates but may be inadequate for time-dependent covariates.

Purpose of the Study:

• To highlight the limitations of the Kaplan-Meier method for time-dependent covariates.
• To introduce and explain the Simon and Makuch method for handling time-dependent covariates in survival analysis.

Main Methods:

• The Simon and Makuch method evaluates covariate status at each event time for individuals at risk.
• Survival computations mirror the Kaplan-Meier method, using ratios of events to those at risk.
• Unlike Kaplan-Meier, the Simon and Makuch method does not fix covariate levels at time 0.

Main Results:

• The Simon and Makuch method accounts for dynamic changes in individual covariate status over time.
• Differences in survival curve representation emerge when applying both methods to time-dependent covariates.
• Examples illustrate discrepancies between Kaplan-Meier and Simon and Makuch methods in Cox proportional hazards regression.

Conclusions:

• The Simon and Makuch method provides a more accurate approach for survival analysis with time-dependent covariates.
• Graphical representation of survival data is crucial for understanding covariate effects over time.
• The study underscores the importance of selecting appropriate statistical methods for complex covariate scenarios.