topic Re: Identify subjects with a skipped dose in SAS Programming

Identify subjects with a skipped dose

molla — Fri, 07 Apr 2017 18:16:41 GMT

Hi,

Identify the subjects with skipped dose

ID Date Dose

001 01Jan2014 10

001 02jan2014 10

001 03jan2014 10

001 05jan2014 10

002 01jan2014 20

002 02jan2014 20

002 03jan2014 20

002 04jan2014 20

003 02jan2014 10

003 03jan2014 10

003 04jan2014 10

003 06jan2014 10

help me in solving this problem.

Re: Identify subjects with a skipped dose

art297 — Fri, 07 Apr 2017 18:41:09 GMT

Here is one way:

data want (drop=counter first_date);
  set have;
  retain first_date;
  by id;
  if not (first.id and last.id);
  if first.id then do;
    counter=0;
    first_date=date;
  end;
  else counter+1;
  if last.id then do;
    if date-first_date ne counter then output;
  end;
run;

Art, CEO, AnalystFinder.com

Re: Identify subjects with a skipped dose

Astounding — Fri, 07 Apr 2017 19:20:21 GMT

Does skipped mean they missed a day?

If the first dose is January 2, does that mean that they skipped January 1?

Here's one possibility:

data want;

set have;

by id;

days_skipped = dif(date) * (first.id=0);

run;

It assumes that your data is already sorted by ID, and that your DATE variable contains valid SAS dates.

Re: Identify subjects with a skipped dose

molla — Fri, 07 Apr 2017 19:58:18 GMT

Hi I have done this in following way,correct me if any thing is wrong

proc sort data=test;
	by id;
run;

data test_;
	set test;
	by id;
	x=lag(date);
	format x date9.;

	if first.id then
		x=.;

	if not first.id then
		y=date-x;

	if y ne 1 and y ne . then
		x1='unscheduled visit';
run;

data final;
	set test_;
	where x1='unscheduled visit';
	drop x y x1;
run;

Re: Identify subjects with a skipped dose

art297 — Fri, 07 Apr 2017 20:21:54 GMT

Both your and my code appear to be doing the same thing, but your method required two passes of the data, while mine only required one pass.

However, unless your dealing with big data, you probably won't care that it will take twice as long.

Regardless, you could have done yours in one pass if rather than using:

if y ne 1 and y ne . then x1='unscheduled visit';

simply using

if y ne 1 and y ne .;

Art, CEO, AnalystFinder.com

Re: Identify subjects with a skipped dose

molla — Fri, 07 Apr 2017 20:28:40 GMT

what if in this case if we need to find the unscheduled visits?

001 01jan2012 10

001 08jan2012 20

001 20jan2012 30

001 27jan2012 40

001 02jan2012 99

001 03jan2012 99

002 02jan2012 10

002 05jan2012 20

002 03jan2012 99

Re: Identify subjects with a skipped dose

molla — Fri, 07 Apr 2017 20:40:02 GMT

cud u pls explain me what does the following statement do

days_skipped = dif(date) * (first.id=0);

Re: Identify subjects with a skipped dose

art297 — Fri, 07 Apr 2017 20:44:39 GMT

I presume that you mean how do you identify the skipped dossages. Regardless, if I were you I'd wouldn't ask it in this thread but, rather, submit the question as a new thread/question. You will likely get more responses if people have a chance to be recognized as the solution provider.

Art, CEO, AnalystFinder.com

Re: Identify subjects with a skipped dose

art297 — Fri, 07 Apr 2017 21:34:01 GMT

dif(date) takes the difference between the value of date and the value of date on the previous record.

first.id will equal 1 for the first id and 0 for all other records within that id, so first.id=0 will have a value of 1 for each record except the first.id record where it will have a value of 0.

so multiplying the two will product the number of days since the last visit for each record except the first record for each id. The first record for each ID will get a value of 0.

Art, CEO, AnalystFinder.com

p.s. Have you posted your new question in a new thread yet?

Re: Identify subjects with a skipped dose

molla — Fri, 07 Apr 2017 21:38:45 GMT

Yes I have posted