Survival Analysis
Survival analysis models time-to-event data — time to death, relapse, discharge, or any defined endpoint. Licklider enforces mandatory at-risk tables, censoring disclosure, median survival with 95% CI reporting, and proportional hazards assumption checks for Cox regression.
Section covers: Kaplan-Meier estimator · Log-rank test · Gehan-Breslow test ·
Median survival time (95% CI) · At-risk table · Cox proportional hazards regression ·
Hazard Ratio (HR, 95% CI) · Schoenfeld residuals (PH assumption check) · Restricted mean survival time (RMST)
When to Use This Section
Use Survival Analysis workflows when:
- Your outcome is time to an event (death, relapse, hospitalization, etc.)
- Some subjects have not yet experienced the event at the time of analysis (censoring)
- You need to compare survival between two or more groups
- You want to adjust the survival estimate for covariates (Cox regression)
Available Scenarios
Kaplan-Meier Workflow
Two-arm survival comparison — oncologist comparing OS between treatment arms (n=88). At-risk table, log-rank test, median survival CI.
Cox Regression Workflow
Adjusted survival analysis — adjusting HR for age, stage, and performance status with PH assumption check via Schoenfeld residuals.
Key Concepts in Survival Analysis
- Event: The defined endpoint (death, relapse, etc.) — must be specified before analysis
- Censoring: Subjects who have not experienced the event at the end of follow-up — their data is retained but their event time is treated as a lower bound
- At-risk table: Shows how many subjects remain under observation at each time point — mandatory for any published KM curve
- Median survival time: The time at which 50% of subjects have experienced the event — reported with 95% CI
- Hazard Ratio: The instantaneous event rate in one group relative to another — reported with 95% CI from Cox regression
- PH assumption: Cox regression assumes proportional hazards — that the HR is constant over time. Schoenfeld residuals test this assumption