Selecting Observations from Repeated Measures Data

[tmcwill]

I'm trying to work with the bladder dataset obtained here http://support.sas.com/documentation/cdl/en/statug/66859/HTML/default/viewer.htm#statug_phreg_exampl...

As provided, the data are repeated measures with 4 observations for each subject. Each observation has a visit number from 1-4 and a binary status variable. I would like to have each subject only appear once with either the first visit where status=1 or if a subject never has status=1, then the last visit (visit=4). I'm thinking maybe I need to somehow use a do loop iterating over each ID, but I'm not sure how to implement. Any help is greatly appreciated! 

[unison · Accepted solution]

There is probably a cleaner way to do this but I think this is what you're looking for!

data want;
	set have(where=(status=1) in=a) have(in=b);
	by id;
	retain _outobs 0;

	if not (_outobs) and ((a and first.id) or (b and last.id)) then
		do;
			output;
			_outobs=1;
		end;

	if last.id then
		_outobs=0;
	drop _:;
run;

Checking result:

proc sort data=have(where=(status=1)) 
nodupkey out=check_sts;
by id;
run;

data check;
set check_sts have(where=(visit=4));
run;

proc sort data=check 
nodupkey out=check_out; 
by id; 
run;

proc compare base=want compare=check_out; run;

-unison

[mkeintz]

Although the interleaving of two streams of HAVE data works in this case, this is not really a situation that demands two streams.  Here's an approach that uses a single stream.  The logic, of course, is effectively identical to the solution provided by @unison :

data want (drop=n_status1);
  set have;
  by id;
  if first.id then n_status1=0;
  n_status1+(status=1);
  if (n_status1=1 and status=1) or
     (n_status1=0 and last.id=1) then output;
run;