Kaplan–Meier & log-rank
§5.1
Rev 4.6 · reviewed Feb 2026Kaplan–Meier & log-rank
For time-to-event outcomes with censoring. One line per patient: time then status (1 = event, 0 = censored), or add “+” for censored. Full Cox adjustment lives in Consult.
Source formula: Kaplan EL, Meier P. J Am Stat Assoc 1958;53:457–81.
Group A — time & status
Group B — time & status
Table. Events observed vs expected
| Group | n | Observed | Expected |
|---|---|---|---|
| A | 8 | 5 | 3.6 |
| B | 8 | 5 | 6.4 |
* Median survival — A: 22, B: 33. Kaplan EL, Meier P. J Am Stat Assoc 1958;53:457–81.
† Log-rank χ² on 1 df. Mantel N. Cancer Chemother Rep 1966;50:163–70.
‡ HR estimated from observed/expected. Tierney JF, et al. Trials 2007;8:16 (log-rank HR method).
Log-rank χ² (1 df)
0.91
copyp
0.340
copyHazard ratio A vs B (95% CI)*
1.76 (0.49–6.40)
copyNo significant difference in survival (p = 0.340). The hazard ratio of 1.76 is an approximation from the log-rank statistic; for an adjusted HR with covariates, use a Cox model — plan it in Consult.
Understand what this is measuring