Sorry. I accidentally deleted my post. Had to repost again.   I've a data set listing an ID's visits to each facility (fac_code) with start date and endate date and visit type, and the facility woud assign an EPID (episode ID) like chart number to the persono, but the faciltiy might give people a different EP ID upont their return.  So now I'd like to bundle such cases, by creating a fac seq key and an meta episode ID so that 'm able to group them together for my analysis later. For a simpler scenario (#1) I had the code written, as below. But there is another layer or complexity (#2) that I'm not able to resolve and hope you guys have a solution (ideally just expanding the existent code.) 1. if the person returned to the home consecutively (e.g. ID 3, A > A), then I'd like to assign a facility sequal ID (starting with 1) to buncle them together and if the person has differnet EPIDs for the same facility (if returning to the same one) , I'd retain the first EPID (e.g. ID3 to facilty A). But if the person goes from facilty A, B and then A, the sequence would be 1,2,3, not 1,2,1 (ID 4). I've the code as below. The output values for fac_seq and EPID2 are the same as the second set of columns from the right in my data input (out1_facseq out1_epID2) . 2. But I now need to take into account of the visit type (vst_type). Basically if the first visits are SH (short, before it becomes longstay), I need to assign fac_seq as 0 (eg ID 2) and EPID stay unchanged (so EPID2=EPID) until there is a LS visit, then fac_Seq becomes 1 and EPID follows the same rules as #1 above. If SH takes place after LS, then it follows the same rule as LS (e.g. ID=4). The last set of columns in the data input (out2_facseq out2_epID2) are my "want" values. I tried to revise my code in step 1 with the following changes, but it didn't assign ID 3 (record 3) and ID 4 (record 4) right. (Again it's best the solution is just expanding the existnet code, not entirely new code.).      from        by ID fac_code notsorted;        if first.ID then fac_seq=0;        if first.fac_code then fac_seq+1;   to             by ID fac_code notsorted;        if first.ID then fac_seq=0;        if first.fac_code and vst_type in ("LS") then fac_seq+1; data have; input ID 1  fac_code $ 3 epID 6-8 entry_dt 10-13 disch_dt 15-19 vst_type $ 21-22 out1_facseq 27 out1_epID2  30-32 out2_facseq 36 out2_epID2 39-41; datalines; 1 A  110 2011 2012  LS    1  110   1  110 1 A  110 2012 2014  LS    1  110   1  110 1 A  110 2014 2015  LS    1  110   1  110 1 B  120 2015 2016  LS    2  120   2  120 2 A  300 2011 2013  SH    1  300   0  300 2 A  305 2013 2013  SH    1  300   0  305 2 B  410 2014 2014  LS    2  410   1  410 3 A  220 2011 2013  SH    1  220   0  220 3 A  220 2013 2013  SH    1  220   0  220 3 A  250 2014 2015  LS    1  220   1  250 3 B  300 2015 2016  LS    2  300   2  300 4 A  505 2011 2013  SH    1  505   0  505 4 B  620 2013 2013  LS    2  620   1  620 4 B  620 2014 2015  LS    2  620   1  620 4 A  505 2015 2016  SH    3  505   2  505 ; proc print; run; *step 1 - assign fac seq; proc sort data=have out=have;     by ID entry_dt disch_dt; run; data have; format ID fac_code fac_seq;     set have;     by ID fac_code notsorted;     if first.ID then fac_seq=0;     if first.fac_code then fac_seq+1; run; proc print; run; *2 -create EPID2; proc sort data=have; by ID fac_code fac_seq entry_dt; run; data want; format ID fac_code fac_seq epID epID2;     set have;     by ID fac_code fac_seq entry_dt;     retain epID2;     if first.fac_seq then epID2=epID; run; proc sort data=want; by ID entry_dt; run; proc print; run; ... View more

I have vertical data recording a person 's in and out of hospital and reason for leaving. I'm trying to fix people who have multiple deaths. The data is vertical: A person (PID) moved from one hospital to another (HospID), and within the same hospital, multiple records to show people's leaving and returning. See below for a sample of 5 cases. Also the actual data is date, not year, and there is no date comflict).    Naturally, 1. a person can only die once (so records for ID 3 & 4 are wrong) 2. if there is subsequent records after death, death should reset as unknown (DK). (so record for #5 is wrong).   So ID 1 and 2 are OK. The ^1st record of ID 3^, the ^2nd record of ID 4^ and the ^2nd record of ID 5^ should be set as DK. I was thinking to use lag function but it doesn't do the work (or can be messy).  It seems I should use retain to retain value from the previous record (aftert sorting the data and set discharge year missing to be say 2019) and then reset the value as I move from one record to the next. But how do I execute?   Thanks in advance for any help you can give me.   data have; input PID 1 HospID $ 3 admission_year 5-8 discharge_year 10-13 reason_to_leave $ 15-21; datalines; 1 A 2012 2013 tempout 1 A 2013 2014 moved 1 B 2015 2016 death 2 A 2012 2013 tempout 2 A 2013 2014 left 2 B 2015 3 A 2012 2013 death 3 A 2013 2014 moved 3 B 2015 2016 death 4 A 2012 2013 moved 4 B 2013 2014 death 4 B 2014 2015 tempout 4 B 2015 2016 death 5 A 2012 2014 tempout 5 A 2015 2016 death 5 B 2013 2014 moved 5 C 2015 ; run; proc print noobs; run;   ... View more

Hello,   I've a dataset with 20 variables; I'd like to have each variable as a dependent variable and the other 19 as independent variables. So in total I'd have 20 regressions. Here is a sample data of 10 cases with just 4 variables x1-x4.   data have; input  x1 x2 x3 x4; datalines; x1 x2 x3 x4 815 545 40 8699 110 574 22 583 571 604 36 5889 469 594 5 3579 246 622 57 2478 745 652 46 5300 508 525 25 4919 1322 667 33 11576 1576 605 32 7845 1684 661 47 10225 ;   Here is what I would do if using copying and pasting.   proc  reg data=have;  model x1 = x2 x3 x4; run; proc  reg data=have;  model x2 = x1 x3 x4; run; proc  reg data=have;  model x3 = x1 x2 x4; run; proc  reg data=have;  model x4 = x1 x2 x3; run;    I could just do this way but since I need to expand the SAS code to use ODS to output parametric estimates or to output predicted values, etc., I'd very much like to have the code efficient if possible. Someone told me to use %scan and tranwrd. I tried but I guess something was missing, so it didn't work.   Any help is appreciated. ... View more