Given data visits; input ID clin_date:date10. surg_date:date9. clin_wt; format clin_date date10. surg_date date10.; datalines; 001 01APR2017 03APR2017 85 001 02APR2017 03APR2017 84 001 03APR2017 03APR2017 85 002 10APR2017 01MAY2017 92 003 10APR2017 05MAY2017 54 003 12APR2017 05MAY2017 52 004 12APR2017 05MAY2017 62 004 13APR2017 05MAY2017 . 005 10MAY2017 10MAY2017 . ; run;   Maybe something like: proc sql; select * from visits where clin_wt is not null group by id having surg_date = max(surg_date) and clin_date = max(clin_date) union select * from visits where clin_wt is null and ID not in ( select id from visits where clin_wt is not null group by id having surg_date = max(surg_date) and clin_date = max(clin_date) ) group by id having surg_date = max(surg_date) and clin_date = max(clin_date); quit; which gives me  ID clin_date surg_date clin_wt 1 3-Apr-17 3-Apr-17 85 2 10-Apr-17 1-May-17 92 3 12-Apr-17 5-May-17 52 4 12-Apr-17 5-May-17 62 5 10-May-17 10-May-17     ... View more

There are multiple ways to do this:   Try: proc sql; select * from have where id in (select distinct id from have where type = 2); quit; ... View more

If all else fails, combine the images into one image (bell-vdm.png) and insert that. ... View more

Try this:   data abuse; infile cards dlm=" "; input case o_ssn:$11. v_ssn:$11. date:$10.; datalines; 1 000-00-0001 000-00-0002 01/01/2000 2 000-00-0002 000-00-0001 01/01/2000 3 000-00-0003 000-00-0004 01/01/2000 4 000-00-0005 000-00-0006 01/01/2000 5 000-00-0007 000-00-0008 01/01/2000 6 000-00-0008 000-00-0007 01/01/2000 ; run; proc sql; create table no_reverse_match as select a.case, a.o_ssn, a.v_ssn from abuse a inner join abuse b on a.o_ssn = b.v_ssn and a.v_ssn = b.o_ssn and a.o_ssn < a.v_ssn where a.date=b.date; quit;     ... View more