in that case this should do the job: data A; subj = "A"; input VISIT $ AVAL BASE CHANGE; cards; V1 40 40 0 V3 49 40 9 V5 49 40 9 V6 50 40 10 Y V8 60 40 20 V9 65 40 25 V10 66 40 26 ; run; data B; subj = "B"; input VISIT $ AVAL BASE CHANGE; cards; V1 40 40 0 V3 49 40 9 V5 49 40 9 V6 50 40 10 N V8 60 40 20 V9 65 40 25 V10 39 40 -1 ; run; data C; subj = "C"; input VISIT $ AVAL BASE CHANGE; cards; V1 40 40 0 V3 49 40 9 V5 49 40 9 V6 50 40 10 N because the last change is smaller than previous (21<25) V8 60 40 20 V9 65 40 25 V10 61 40 21 Y ; run; data D; subj = "D"; input VISIT $ AVAL BASE CHANGE; cards; V1 50 40 10 Y V8 60 40 20 V9 65 40 25 V10 70 40 30 ; run; data E; subj = "E"; input VISIT $ AVAL BASE CHANGE; cards; V1 8 8 0 V3 25 8 16 Y V5 16 8 8 V6 16 8 8 V8 25 8 16 Y V9 25 8 16 Y V10 25 8 16 Y ; run; data have; set A B C D E; run; data want; m = 0; do _N_=1 by 1 until(last.subj); set have; by subj; if m <= 0 and CHANGE >= 10 then m = _N_; /* makrk first occurence of change>=10 */ if m > 0 and AVAL < BASE then m = -1 ; /* check AVAL < BASE for records over CHANGE >= 10 */ if CHANGE < 10 then m = 0; end; do _N_=1 to _N_; set have; if m = _N_ then FLAG = "Y"; else FLAG = " "; output; end; drop m k; run; proc print; run;   Bart ... View more

Thanks guys ! both methods are working 🙂 ... View more

Another approach,   data test; input adverse $ 1-12 count1; cards; Diarhoea [1] 170 Diarhoea      100 Rash [2]       174  Rash            145  Nausea [3]   123  Nausea         67 Anemia [4]    156  Cardiac         167  Fatigue          174  ; run; data want; set test; by adverse notsorted; adverse1=scan(adverse,1,'[ '); run; data want2; length adverse2 $200; set want; retain adverse2; by adverse1 notsorted; if first.adverse1 then adverse2=adverse; run; data want3(rename=(adverse2=adverse)); set want2; by adverse2 notsorted; if last.adverse2; drop adverse1 adverse; run; ... View more