timelist created by proc lifetest is different from the actually followup days?

[SeaMoon_168]

I want to find a KM rate at 90 days. so I used

proc lifetest data=testdata timlist=90;
time day*censor(0);
strata group;
run;

However, I found the timelist and the day are different. Could you help me figure out why there is a gap between these two? If I do really want to find the km rate at day 90, how to set up the options of proc lifetest?

Many thanks!

[FreelanceReinh]

Hello @SeaMoon_168,

If a value T in the TIMELIST= option is not one of the observed survival times, the second column ("day" in your example) will contain the most recent event time before T in the same stratum (group in your example) -- or zero if there is none. The values in columns "Survival", "Failure" and "Survival Standard Error" correspond to that most recent event time, whereas the values in columns "Number Failed" and "Number Left" correspond to the most recent survival time before T (again, in the same stratum), which includes censored survival times, if any.

---

@SeaMoon_168 wrote:

If I do really want to find the km rate at day 90, how to set up the options of proc lifetest?

---

You get the correct Kaplan-Meier estimate with this option because of the definition of the Kaplan-Meier estimator as a step function.

[Ksharp]

As Freelance said there are not K-M estimator between "timelist and the day".

Your original code could you right result.