I am interested in combining point-wise survival estimates from Kaplan-Meier curves, e.g. I want to combine the survival estimates at age 70. Which R package/functions should I look into? I see that the metaprop
function in meta
combines single proportions. Can I really treat these survival estimates as "proportions"? I guess they are a proportion of # survive past time t/# at risk. Which method of calculating the confidence interval should I use? The default is an exact binomial CI, which I don't think is appropriate.
Or would the rma.uni
function in the metafor
pacakge be more appropriate?