PivoGraph
Language
Open PivoGraph

Free online graph maker

Kaplan–Meier survival curve maker

A Kaplan–Meier curve shows the proportion of subjects still free of an event, such as death, over time, and steps down at each event. PivoGraph draws the curves from one row per subject, with 1 for an event and 0 for a censored subject, and compares the groups with the log-rank test on the same table.

Make a survival curve Opens the editor with sample data you can replace. No sign-up; your data stays on your device.
Kaplan–Meier Survival Curves with 95% Confidence Bands
Example: Kaplan–Meier Survival Curves with 95% Confidence Bands

The data you need

A Survival table: the time in X and, in the column of the subject's group, 1 for an event or 0 for a censored subject. Name each column after a group, such as Vehicle and 5 mg/kg. A subject is censored when it was last seen without the event, for example alive at the end of the study or lost to follow-up. Several subjects can share the same time.

What the graph can show

  • Four layouts under Survival curves: Staircase with ticks, Staircase, points, Point to point and Point to point, error bars.
  • Probability of Survival or Probability of Death, as Fractions or Percents, set in Format Plot.
  • Show number at risk table prints the number of subjects still at risk below the graph.
  • Censor ticks and Confidence bands in Format Graph style the marks at censored times and add a 95% confidence band around each curve.

Compare the curves

Open the Analysis panel and choose Compare survival curves. The results include:

  • The Log-rank (Mantel-Cox) and Gehan-Breslow-Wilcoxon tests, plus the log-rank test for trend with three or more groups.
  • Each group's N, number of events and median survival with its 95% CI.
  • With two groups, the hazard ratio and the ratio of median survival, each with its 95% CI.

Add results text to graph places the P values, hazard ratio and medians on the graph. These calculations are compared with R's survival package on the validation page.

Export for a journal

Export the figure as TIFF at up to 1200 dpi, or as PDF, EPS or SVG vector artwork; see How to export a 300 dpi TIFF figure for a journal. Name the test and n per group in the legend, for example "log-rank (Mantel-Cox) test, n = 16 mice per group".

Frequently asked questions

Log-rank or Gehan-Breslow-Wilcoxon? Report the log-rank test unless you planned otherwise. Gehan-Breslow-Wilcoxon gives more weight to early events, when more subjects are still at risk.

Why is a median survival undefined? When fewer than half of a group had the event, the curve never falls to 50%, so the median cannot be read from it.