How to get the survival time in long format

[nwang5]

Hi SAS community,

I used a long format of data and included four waves. And every two years had a wave. I plan to use proc phreg to do survival analysis and to calculate the survival time. How do I get survival time? I also did not know if ID 02 who did not get the disease should have 0 or 8 years of survival time.

Thanks for all your help! 

ID	wave	Disease	expected results (survival time)
01	1	0	6
01	2	0	6
01	3	0	6
01	4	1	6
02	1	0	8 or 0?
02	2	0	8 or 0?
02	3	0	8 or 0?
02	4	0	8 or 0?
03	1	0	2
03	2	1	2
03	3	1	2
03	4	1	2

[Tom · Accepted solution]

So it sounds like you are accessing them at the BEGINNING of the 2 year period.

So the survivors should have a time of 6 years.  At WAVE 5 you would know whether or not they survived 8 years. 

You need to also keep the Disease variable (or something) to serve as your censoring variable for the survival analysis.

data want;
  set have ;
  by id wave;
  retain censor time ;
  if first.id then call missing(of censor time);
  if missing(time) and Disease then do;
    censor=0; time=2*(wave-1);
  end;
  if last.id then do;
    if missing(time) then do; censor=1; time=2*(wave-1); end;
    output;
  end;
  keep id censor time;
run;

Results:

Obs    ID    censor    time

 1     01       0        6
 2     02       1        6
 3     03       0        2

[nwang5]

Thank you so much, Tom! You helped me a lot. Thanks for answering my previous questions. Thanks for making my research and life easier! I really appreciate it.

[nwang5]

I am sorry that I have a following up question. I found some participants who did no finish all these for waves. How could I deal with these people who did not participate in four waves?

ID	wave	Depression	expected results (survival time)
01	1	0	6
01	2	0	6
01	3	0	6
01	4	1	6
02	1	0	8
02	2	0	8
02	3	0	8
02	4	0	8
03	1	0	2
03	2	1	2
03	3	1	2
03	4	1	2
04	1	0	2
04	2	0	2
05	1	0	2
05	2	1	2
05	3	1	2
06	1	0	8
06	3	0	8
06	4	0	8