Data have; input ID$ PARAMCD$ RESULT BASERSLT CRITFL$; datelines; 004 ALT 0.03 0.03 004 ALT 0.06 0.03 004 AST 0.03 0.03 004 CHOL 0.03 0.03 004 4HCHOL 0.03 0.03 004 ALT 0.12 0.03 Y 004 LIPA 0.09 0.09 005 ALT 0.03 0.03 005 AST 0.06 0.04 005 CHOL 0.03 0.03 005 4HCHOL 0.03 0.03 005 ALT 0.06 0.03 N 005 LIPA 0.12 0.12 ; Data want; input ID$ PARAMCD$ RESULT BASERSLT CRITFL$; datelines; 004 ALT 0.03 0.03 Y 004 ALT 0.06 0.03 Y 004 AST 0.03 0.03 Y 004 CHOL 0.03 0.03 Y 004 4HCHOL 0.03 0.03 Y 004 ALT 0.12 0.03 Y 004 LIPA 0.09 0.09 Y 005 ALT 0.03 0.03 N 005 AST 0.06 0.04 N 005 CHOL 0.03 0.03 N 005 4HCHOL 0.03 0.03 N 005 ALT 0.06 0.03 N 005 LIPA 0.12 0.12 N; 1.CRITFL for the ID will have either "Y" or "N" and not a combination and hence if at least one record of ID has a "Y" or "N" or even "" i want to see that being retained to all the other records of the ID. Hope this is a bit helpful and apologies for the attachments. Thanks ... View more

/* OK. Test the following code. If there was something wrong,post an example. */ data have; Input Id visitn testcd $ result baseresult basefl $ pstbsfl $; datalines; 1 1 ALT 12 3 . Y 1 2 AST 3 1 . Y 2 1 AST 3 1 Y . 2 2 AST 2 1 . Y ; run; proc sql; create table want as select *, case when max(pstbsfl='Y' and testcd='ALT') and max(pstbsfl='Y' and testcd='ALT' and result>=4*baseresult) then 'Y' when max(pstbsfl='Y' and testcd='ALT') and max(pstbsfl='Y' and testcd='ALT' and result>=4*baseresult)=0 then 'N' when max(pstbsfl='Y' and testcd='ALT')=0 then ' ' else 'X' end as critfl from have group by id; quit;   EDITED. ... View more

data want; set have; by id; if last.id then flg='Y'; else flg='.'; run;     If I may be so bold, setting FLG to 'Y' or '.' is not such a good idea; setting FLG to numeric 0 or 1 is not only easier to type, but has other benefits as well (such as computing proportions that are flagged can be done by computing a mean). ... View more

Data want; Input ID $ test$ visitnum basefl$ result$ adt randdt basec$ 1 CHOL 1 Y HIGH 17JUN2018 18JUN2018 HIGH 1 CHOL 1 Y LOW 17JUN2018 18JUN2018 HIGH 1 CHOL 1 Y HIGH 18JUN2018 18JUN2018 HIGH 1 CHOL 2 . LOW 24JUN2018 18JUN2018 HIGH; run; @ballardw  Thanks for your response and apologies for so many confusions.   Here below is what i meant regarding equally apart and the data is how you have shown in your answer and here above is the data look that i want finally.     when I meant equally apart i mean to say that when we calculate the difference between adt and randdt if we get the same value for any two records.   Logic that I want here is that: 1. If a ID is having 2 or more records with BASEFL eq ‘Y’ and are equally apart (difference between adt and randdt is same )from randdt then consider worst case (HIGH) as BASEC and retain the value across the ID. 2. If a ID is having 2 or more records with BASEFL eq ‘Y’ and are not equally apart from RANDDT then choose the BASEC from the closest record to RANDDT and retain its value across the ID.   TIA again ... View more

Hi,   thanks for looking at my data   here is what i wanted it to look alike   id test  visit  dy targetdy result  baselineflg 1  tst1   1      1        1            R      Y 1  tst1   1     -29     1            R 1  tst1   1       1       1           S   sorry for the text above please consider the closest dy to targetdy.     Thanks               ... View more

By   "difference between dp at number 1 and rp at value at next number". I think you want the rp test value minus the dp test value for each number.  If so, then:   data want; merge have (where=(tst='dp') rename=test_rt=dp_test) have (where=(tst='rp') rename=test_rt=rp_test); by number; rp_minus_dp=rp_test-dp_test; drop tst ; if last.number; run; If that's not what you mean, please show what the results would look like for the data you provided. ... View more