When performing survival analysis (Kaplan Meier), especially when comparing survival of a cohort of patients against that of a matched general population, is it appropriate to display the respective curves with confidence intervals or survival standard error (as given in the resulting JMP table for the curves) or does it matter anyway? The matched survival in my general population is calculated according to the method of Dianne Finkelstein (Finkelstein, D. M., Muzikansky, A., & Schoenfeld, D. A. (2003) Journal of the National Cancer Institute, 95(19), 1434–1439.) for each year of follow-up and displayed with CI and the cohort survival curve displayed with JMPs pointwise CI. When reporting survival numerically in a summary table though, since SE is not symetrical about the mean, compared to CI which is, what convention should be followed, mean±CI or mean with upper and lower SE?
