topic Re: How to get the missmatch values for timepoints that for the same formseq for a subject in SAS Programming

How to get the missmatch values for timepoints that for the same formseq for a subject

Ravindra_ — Tue, 06 Apr 2021 10:40:46 GMT

I have a dataset where it is having multiple timepoints and multiple form sequence for the same subject id and i need to output those records where the form sequence is same for the subject but the timepoint is different, here as per the requirement if the form sequence is same for a subject the timepoint as as well must be same, if it is different it must be in the output. I have been trying with lag function and also using first. but i found that jot useful and not able to achieve the result. any help please.

example datasets as below

data ndsn;
infile datalines;
input timepoint formseq  usubjid $15.;
datalines;
1 1 111A116
1 1 111A116
1 1 111A116
5 5 111A116
4 5 111A116
5 5 111A116
8 8 111A116
8 8 111A116
6 6 111A116
6 6 111A116
3 6 111A116
;
run;

expected output

data ndsn;
infile datalines;
input timepoint formseq  usubjid $15.;
datalines;
4 5 111A116
3 6 111A116
;
run;

I had tried lag function and also using first. but they didn't provide expected result, i thought of using if then but that is taking long time, can i know a solution for this

Re: How to get the missmatch values for timepoints that for the same formseq for a subject

Kurt_Bremser — Tue, 06 Apr 2021 10:53:12 GMT

LAG and FIRST. are OK as an idea, it's how you implement them:

data want;
set ndsn;
by usubjid formseq notsorted;
retain flag;
if first.formseq then flag = 1;
if not first.formseq and timepoint ne lag(timepoint)
then do;
  if flag then output;
  flag = 0;
end;
drop flag;
run;

Re: How to get the missmatch values for timepoints that for the same formseq for a subject

Ravindra_ — Tue, 06 Apr 2021 11:07:28 GMT

Thanks a lot Kurt for a quick reply, it worked for me, i have got expected output now. Just for the purpose of knowledge, can i know the purpose of using notsorted

Re: How to get the missmatch values for timepoints that for the same formseq for a subject

Kurt_Bremser — Tue, 06 Apr 2021 11:15:10 GMT

It is because in your example formseq 6 follows formseq 8. Which means that I either have to use the NOTSORTED option or use PROC SORT first to get the values in order.

Re: How to get the missmatch values for timepoints that for the same formseq for a subject

Ksharp — Tue, 06 Apr 2021 12:21:47 GMT

data ndsn;
infile datalines;
input timepoint formseq  usubjid $15.;
datalines;
1 1 111A116
1 1 111A116
1 1 111A116
5 5 111A116
4 5 111A116
5 5 111A116
8 8 111A116
8 8 111A116
6 6 111A116
6 6 111A116
3 6 111A116
;
run;
data want;
 do until(last.formseq);
  set ndsn;
  by  formseq notsorted;
  if first.formseq then _timepoint=timepoint;
  if _timepoint ne timepoint  then output;
 end;
 drop _timepoint;
run;