Here is an obnoxious SQL approach with the undocumented feature: data have; input ID     (Enter_Date     Exit_Date) (:mmddyy10.)   Sex$     Age; format     Enter_Date     Exit_Date mmddyy10.; cards; 1        1/1/2000       1/1/2002       m       24 1        5/5/2006        5/5/2009      m       30 1        8/8/2011        8/8/2013      m       35 2        2/2/2002        2/2/2005       f        42 2        6/6/2006        6/6/2010       f        46 3        3/3/2003        3/3/2004      m       19 3        4/4/2008        4/4/2012      m       24 ; proc sql; select a.*,b.enter_date-a.exit_date as new_date from (select *, monotonic() as obs from have) a left join (select *, monotonic() as obs from have) b on a.id=b.id and b.obs-a.obs=1 order by id, enter_date; quit; Haikuo ... View more