If I understand your dataset then you should be able to do this in 1 data step vs. 46 using arrays. Try this: data med_18; set med_psych_visits; array num_day_m18_p(45) num_day_m18_p1-num_day_m18_p45; array psych_date_(45) psych_date_: ; do _i = 1 to 45;      num_day_m18_p(_i) = med_date_18 - psych_date_(_i); end; keep mrn num_day_m18_p: ; run; ... View more

Looks like recalldate is a character. Read it in with a datetime. informat and get the date part of the variable. Revised code: recalldate2 = put( intnx( 'month', datepart( input( recalldate, datetime. )), recall, 's' ), yymmn6. ); ... View more

Here is the equivalent SAS code: recalldate1 = put( intnx( 'month', recalldate, recall, 's' ), yymmn6. ); ... View more

1. When constructing the filename, should a period be before HISTORY, e.g. sys90h.syse.history.120327? 2. It's been a while since I've worked on a mainframe. If these are os/390 files, don't the segments of the filename have to begin with an alpha character? So the segment with the date must begin with a-z. ... View more

I encountered a similar situation where the last field in a text file contained a long comment with embedded crlf's or lf's. This solution begins by reading the variables of the first record as you normally would except that I add a line pointer hold (@). At this point, the problem is that you do not know if the value of txt var is complete or if parts of it are in subsequent lines of the text file. You need a way to determine if the next line is a new record or a continuation of txt var. Since each record begins with a datetime field, you could easily test if the next line begins with a datetime. Here's the code: data test; infile "c:\data\EmbeddedLineBreak.txt"    firstobs=2 dlm='09'x dsd end=eof; informat dt ymddttm19. debet $20. credit $20. sum_rur comma17. txt $100.; format dt datetime.; if eof then stop; input @1 dt -- txt @; * this section reads the next line to determine if the first * variable is a datetime. if so, then the next record is found. if not, * then another part of the txt var is found.; length _temp $100; do while ( not eof );    input _temp :$100. @@; * <-- this is important. holds the line pointer.    * if the input function is not missing then you found the    * beginning of the next record;    if input( _temp, ymddttm19. ) ne . then leave; * exit the loop;    * if the input func returns mising, the _temp var must contain part of the txt var.    * concatenate _temp to txt;    txt = dequote( catx( ' ', txt, _temp ));    _error_ = 0; end; drop _:; run; I created a tab-delimited file in Excel. The records look like this: dt debit credit sum_rur txt 2012-03-20 09:30:30 client a client b " 10,000 " "This field has an embedded linefeed here causing a line break in the text file." 2012-03-21 09:30:30 client c client d " 20,000 " "This field has no embedded linefeeds." 2012-03-22 21:30:30 client e client f " 30,000 " "2 embedded linefeeds. The first is here and another one is here causing another line break." ... View more

Any messages in the log? ... View more