FOR THE NEXT GENERATION OF PHYSICIANS
← All USMLE guides
BIOSTATISTICS, EPIDEMIOLOGY AND ETHICS

Survival analysis: Kaplan-Meier curves and hazard ratios.

Reading Kaplan-Meier curves, median survival, censoring, hazard ratios and log-rank tests in trial questions.

Survival analysis handles time-to-event data. Know how to read the curve and interpret a hazard ratio.

Kaplan-Meier

The curve drops at each event. Censored patients (lost, withdrew or still alive at study end) are shown as tick marks and removed from the risk set. The median survival is where the curve crosses 50 percent. The log-rank test compares curves.

Hazard ratio

A hazard ratio below 1 favors the treatment group, above 1 the control; 1 means no difference. Cox proportional hazards models adjust for covariates.

Trial analysis

Intention-to-treat analysis keeps patients in their assigned groups and preserves randomization; per-protocol analysis includes only those who adhered. Adverse event reports and surrogate endpoints (blood pressure rather than strokes) must be interpreted with caution.

Keep going in biostatistics, epidemiology and ethics

Related on USMLE WEB

Frequently asked questions

What does censoring mean?

A patient is lost to follow-up or the event has not occurred by study end.

What does intention-to-treat analysis preserve?

The benefits of randomization.