Is it possible to make test for survival between groups at a defined specific time?

[Tom Hsiung]

-sts list- command could compute the confidence intervals for survival function at specific time interval. However, no comparison was made and so no P values.

by -sts ci-



by -sts list-

[Carlo Lazzaro]

Tom:
you may want to consider -logrank-:

. sts test drug if age==67, logrank

         failure _d:  died
   analysis time _t:  studytime


Log-rank test for equality of survivor functions

      |   Events         Events
drug  |  observed       expected
------+-------------------------
0     |         1           0.33
1     |         2           2.67
------+-------------------------
Total |         3           3.00

            chi2(1) =       2.00
            Pr>chi2 =     0.1573

[Tom Hsiung]

Hi, Carlo

Thank you for your help. SO the key point is the "age==" parameter which defines the timer interval (from time zero to time t). For instance, if the age== is 5 (days), the survival comparison is the 5 day-survival, and if 7 (days) is the 7 day-survival. Did the way I understand correct?

Tom

[Tom Hsiung]

Oh, it seems that it was the log-rank test of that particular time interval, like the 5th day, the 10th day. I am trying to know the overall log-rank test results for a duration (from day 0 to a particular day later). I got errors from my command.

Tom

[Tom Hsiung]

Figured it out! Thank you!

[Steve Samuels]

Your screen shots are unreadable (blurred & small on my monitor), which is why FAQ 12 asks that you not post them. Upload graphs as .png documents and show results, code, and data listings posted from your results window or log file between CODE delimiters, described in FAQ 12.

[Steve Samuels]

While I can't read any of your screen shots, I will correct Carlo's typo: If you wanted to do the log rank test on survival through age 67, the correct syntax would be:

sts test drug if age<=67, logrank

[Carlo Lazzaro]

Steve is correct.