Here's one way: data have; input site $ subject visit & $ eval $; datalines; 01 111 Screen Y 01 111 Visit 1 N 01 111 Visit 2 Y 01 222 Screen Y 01 222 Visit 1 Y 01 222 Visit 2 Y 02 333 Screen Y 02 333 Visit 1 Y 02 333 Visit 2 N 02 444 Screen Y 02 444 Visit 1 Y 02 444 Visit 2 Y ; run; /* Find the site/subject combinations for which EVAL should be N */ proc sql; create table visits_with_N as select distinct site ,subject from have where upcase(visit) like "VISIT%" and upcase(eval) = 'N'; quit; /* Join those combinations to HAVE, and set EVAL to N where there is a match */ proc sql; create table want as select h.site ,h.subject ,h.visit ,case when v.subject is null then h.eval else 'N' end as eval from have h left join visits_with_n v on h.site = v.site and h.subject = v.subject order by site, subject, visit, eval; quit; proc print data = want; By the way, thank you for the clear question and run-able sample data. ... View more

Here's a simple example, just to illustrate: data have; input name $ age; datalines; Allen 12 Bob 23 Charles 34 Doug 45 Edward 56 ; run; title 'Before:'; proc print data=have; run; data want; set have; if name = 'Bob' then delete; run; title 'After:'; proc print data=want; run; ... View more

It looks like you have this posted already. ... View more

Here's one way: data have; input id event $ date :mmddyy10.; format date mmddyy10.; datalines; 1 6_W 12/03/2020 1 8_W . 2 6_W 04/07/2012 2 8_W . 3 6_W 05/08/2012 3 8_W . 3 10_W . ; run; data want; set have; if event = '6_W' then do; event = 'S_1'; output; event = '6_W'; date = .; end; output; run; ... View more

Is this what you're looking for? data have; input string $32.; datalines; 4.9 x 10(6)/kg 2.51 x 10(6)/kg 7.54 x 10(6) / kg 1.90 x 10(6)/kg 2.23 x 10(6)/kg 2.29 x 10^6 2.51 x 10(6)/kg 2.55 x 10(6)/kg 2.58 x 10(6)/kg 2.96 x 10(6)/kg 2.98 x 10(6)/kg 3.06 x 10^(6) cells/kg 3.14 x 10(6)/kg 3.96 x 10 (6) 4.38 x 10(6)/kg 4.60 x 10(6)/kg 4.9 x 10(6)/kg 5.12 x 10^(6) cells/kg 5.13 x 10(6) ; run; data want; set have; number = input(substr(string, 1, find(string, 'x', 1)-1), best.); run; ... View more

Where are you seeing duplicated observations?    Using the exact code I supplied above, if you run this: proc print data=want; run; It results in this, which appears to match your desired output:   ... View more

The "B" outpatient dataset is further divided into type: "2" which are emergency room visits and "3" which are regular doctors visits. Since your sample data has types 2 and 3 for A, I am assuming you swapped "A" and "B" in your data (so that "A" is outpatient and "B" is inpatient).    With that assumption, your question and data seem to make sense. See if this is what you're looking for: data have; input source $ id $ type $ begin_date :date9. end_date :date9.; format begin_date end_date date9.; datalines; A 1 3 28JAN2019 . B 1 1 29JAN2019 04FEB2019 A 1 2 29JAN2019 . A 1 3 29JAN2019 . A 1 3 30JAN2019 . A 1 3 31JAN2019 . A 1 3 01FEB2019 . A 1 3 02FEB2019 . A 1 3 03FEB2019 . A 1 3 04FEB2019 . A 1 3 05FEB2019 . A 2 3 28JAN2019 . A 2 3 30JAN2019 . A 2 3 31JAN2019 . B 2 1 01FEB2019 03FEB2019 A 2 2 01FEB2019 . A 2 3 01FEB2019 . A 2 3 02FEB2019 . A 2 3 03FEB2019 . ; run; data inpatient; set have (where=(source='B')); run; proc sql; create table joined as select h.* ,o.begin_date as inpatient_begin_date ,o.end_date as inpatient_end_date from have h left join inpatient o on h.id = o.id order by h.id, h.begin_date, h.type; quit; data want (drop=inpatient:); set joined; if not(source = 'A' and type = '3' and inpatient_begin_date le begin_date le inpatient_end_date) then output; run; ... View more