Which method is used to assess the proportional hazards assumption?

• A. Kaplan-Meier curves.
• B. Schoenfeld residuals.
• C. Log-rank test.
• D. Chi-square test.

Answer: (B)

In Cox proportional hazards modeling, the crucial assumption is that hazard ratios stay constant over time. To evaluate this, the most direct approach is using Schoenfeld residuals. After fitting the model, examine these residuals for a relationship with time: a systematic trend or nonzero slope suggests that covariate effects change over time, meaning the proportional hazards assumption is violated. This can be assessed visually by plotting residuals against time or formally with a test for zero slope (often implemented as a Schoenfeld residuals test). Kaplan-Meier curves show survival over time but don’t test whether hazard ratios are constant. The log-rank test compares survival curves between groups but isn’t a check of the PH assumption itself. The chi-square test is a general test and doesn’t specifically assess time-varying effects in a Cox model.