The Kaplan–Meier (KM) method estimates the survival curve: the probability of remaining event-free over time.
Why it matters
It uses everyone's information, including censored people, and shows the whole pattern.
Key ideas
- S(t) = product of (1 − events ÷ at risk) at each event time
- The curve steps down at each event
- Median survival: time when S(t) reaches 0.5
- The 95% CI band shows uncertainty
- A table of numbers at risk should accompany the plot
- Compare groups with the log-rank test
Real-life research example
Median survival is 18 months in group A. At 12 months, estimated survival is 70% (95% CI 58–80%).
When to use
- Describe survival of one or more groups
When NOT to use
- Don't read the tail of the curve when very few remain at risk
Mini example
Start at 1.0. After a death among 10 at risk: 1.0 × (1 − 1/10) = 0.90
Quick takeaway
KM draws the survival curve using all available follow-up.
Try it yourself
Enter your own times and events. Read the median survival and the survival probability at two time points.
Model answer
The median is where the curve first reaches 0.5 or below. If the curve never drops that low, median survival is "not reached".