If it helps, yesterday I managed to implement the algorithm. proc sort data=input_ds out=input_ds_sort; by SubjectId diff_days; run; data input_ds_pairs_sel; set input_ds_sort; by SubjectId; retain prev_day  prev_flag1  prev_is_prior best_dt1 best_dt2 best_gap  pa_dt1 pa_dt2;  if first.SubjectId then do; prev_day = .; prev_flag1 = 0; prev_is_prior = .; best_dt1 = .; best_dt2 = .; best_gap = .; pa_dt1 = .; pa_dt2 = .; end; is_prior = (-21 <= diff_days < 0); is_after = (0 <= diff_days <= 7); curr_flag1 = (flag_KPI = 1); if not first.SubjectId then do; if prev_flag1 = 1 and curr_flag1 = 1 then do; /* PRIOR–PRIOR */ if prev_is_prior = 1 and is_prior = 1 then do; gap = diff_days - prev_day; if missing(best_gap) or (gap < best_gap) or (gap = best_gap and diff_days > best_dt2) then do; best_dt1 = prev_day; best_dt2 = diff_days; best_gap = gap; end; end; /* PRIOR→AFTER */ else if prev_is_prior = 1 and is_after = 1 then do; if missing(pa_dt1) then do; pa_dt1 = prev_day; pa_dt2 = diff_days; end; end; end; end; prev_day = diff_days; prev_flag1 = curr_flag1; prev_is_prior = (is_prior = 1); if last.SubjectId then do; length source $12; if not missing(best_dt1) then do; source = 'prior-prior'; dt1 = best_dt1; dt2 = best_dt2; gap = best_gap; output; end; else if not missing(pa_dt1) then do;  source = 'prior-after'; dt1 = pa_dt1; dt2 = pa_dt2; gap = dt2 - dt1; output; end; end; keep SubjectId dt1 dt2 gap source ; run;   Have a nice time Regards ... View more

Thnks for your reply   Hereafter: data kpi_data;     infile datalines dsd truncover;     input SubjectId :8.           differences_in_days :8.           flag_KPI :8.; datalines; 1,-21,1 1,-14,1 1,-7,1 2,-16,1 2,0,1 2,6,1 3,-16, 3,-9,1 3,5, 4,1, 5,-21,1 5,0,1 6,0, 7,-11, 7,1, 7,6, 8,-21,1 8,-11,1 8,0,1 9,-14, 9,-9, 9,-6, 9,1, 9,2, 10,-19, 10,-14, 10,-8,1 10,-5, 10,0, 11,-10, 11,-3,1 11,0,1 11,7, 12,-21,1 12,-4,1 12,1,1 12,2,1 12,2,1 12,7, 13,-15,1 13,2, 14,-15,1 14,-8, 14,3,1 15,-8, 15,1,1 16,-18,1 16,-15,1 16,-11, 16,-8,1 16,-4, 16,-3, 16,-2, 16,-1,1 16,5,1 16,6,1 17,-20,1 17,-14,1 17,-7,1 17,0,1 17,7,1 18,-21,1 18,-20,1 18,-13,1 18,-4,1 18,0,1 18,4,1 18,7,1 13,-17,1 13,-1,1 14,-15,1 14,-8,1 14,0,1 14,7,1 15,-21,1 15,-13,1 15,0,1 16,-21,1 16,-14,1 16,-7,1 16,0,1 16,1,1 17,-16, 17,-9, 17,-2, 17,0, 17,3, 17,4, 18,-17,1 18,-5,1 18,-2, 18,2, 18,5, 19,-15, 19,-1,1 20,-6, 20,-2, 20,6, 21,-21, 21,-20, 21,-18,1 21,-7,1 22,-20,1 22,-15,1 22,-6,1 22,4, 23,-18, 23,-14, 23,-11, 23,1,1 24,-14,1 24,-10,1 24,5,1 25,-1,1 25,-18, 25,-3, 26,-15, 26,0,1 27,-3,1 27,-4,1 28,-14,1 28,3,1 29,-10,1 29,6,1 30,-14,1 30,-11, 30,-8, 30,-1,1 30,3,1 ; run;   I tried using retain by creating two counters (one for days < 0 and one for days ≥ 0), but I wasn’t successful.   Thanks for any help       ... View more

Thank you very much for your help. I’ve found a solution for the survival analysis. In the coming days, I’ll look into the failure analysis as well, and I hope I’ll be able to make it work. ... View more

Dear, Thank you for your helpful note. I’ve read and tried to apply the SAS guidance you shared. Unfortunately, I’m unable to execute the %grtitle(...) command due to insufficient authorization. As an alternative, I proceeded to modify the Kaplan-Meier template directly. However, I’m still encountering an issue: the default title “Number of Subjects at Risk” continues to appear in the graph        despite having updated the relevant section of the template as follows: %if &ntitles gt 1 %then %do;    %if not &outside %then if (PLOTATRISK=1);       entrytitle "" /                  textattrs=GRAPHVALUETEXT;    %if not &outside %then %do; endif; %end; %end; The title still appears as shown in the screenshot I posted. Do you have any suggestions on how to fully suppress or remove it? Thanks again for your support! Best regards, ... View more

Many thanks for your valuable input ... View more

Thanks, it works! ... View more

Thanks, everything is working as expected now! ... View more

Dear, I confirm that the proposed solution works and I’m getting the correct analysis. I have several failure functions to process; in many cases the graph is fully complete (graph1 attached) while in others some X-axis labels are not displayed (graph2 attached) I assume because some are very close together). Is there any solution for this?   thanks in advance Kind regards   ... View more

Thank you very much for your clear and detailed explanation. Tomorrow morning I’ll test it in my SAS programming, but I’ve understood the reason why the SAS code wasn’t working. Have a nice time! ... View more