[R] median survival time from survfit

[Marc Schwartz]

On Mar 18, 2011, at 6:06 AM, Dunia Scheid wrote:

> Hello,
> 
> I am trying to compute the mdeian of the survival time from the function
> survfit:
> 
>> fit <- survfit(Surv(time, status) ~ 1)
>> fit
> Call: survfit(formula = Surv(time, status) ~ 1)
> 
> records   n.max n.start  events  median 0.95LCL 0.95UCL
>    111     111     111      20      NA      NA      NA
> 
> The results is NA? the fit$surv gives values between 1 and 0.749! Am I doing
> this correct? Is there a correct way to do this?
> 
> Many thanks for your help,
> Dunia


Median survival is the time at which the survivorship function equals 0.5. Since your minimum value appears to be 0.749, you never get there, thus the output shows NA.

If you did not have any censored observations, median survival would also be the point at which 50% of your sample has not yet observed the event of interest.

See ?print.survfit for additional information specific to this function.

HTH,

Marc Schwartz