0

Kaplan-Meier curves have been shown with many different names in studies.

  1. Survival
  2. % survival
  3. Survival probability
  4. Survival proportion

Is it okay to use any of them or are there slight differences? Is it a proportion or a probability? Does it depend when CIs are also provided?

ethan282712
  • 73
  • 4
  • I would go with "survival probability" which seems the most comprehensive and concerns the overall probability of survival in the population. The term "survival proportion" is understable as well but "proportion" is somehow, I guess, tied with the notion of sample (the proportion of success in a sample). I found this subject about the distinction between both terms if you want to take a look: https://stats.stackexchange.com/questions/1525/whats-the-difference-between-a-probability-and-a-proportion – periwinkle Oct 21 '21 at 14:18

1 Answers1

1

You might want to distinguish between displays of raw survival data like Kaplan-Meier curves and survival functions predicted from a semiparametric or parametric model. For example, "proportion" might most directly represent raw data while "probability" might represent model predictions. Thanks to @periwinkle for the helpful link about that distinction. Even that distinction gets a bit fuzzy, however, as raw Kaplan-Meier curves are generally presented with modeled confidence limits.

Whichever you choose, make sure that the numeric axis labels agree with the text label. If you use "Percent survival" then the y-axis range should be 0 to 100. If it's "Survival probability" then the range should be 0 to 1.

EdM
  • 57,766
  • 7
  • 66
  • 187