Selecting first record based on two groups?

[Tpham]

I have been trying to use the first/last commands for the last hour and trying to figure out how to get this to work.

I have the following sample lab data:

id
date
count
year


1
12/5/2007
< 75
2007


1
2/27/2008
500
2008


1
6/30/2008
< 75
2008


1
10/27/2008
< 75
2008


1
2/23/2009
900
2009


1
6/1/2009
< 75
2009


1
8/24/2009
< 75
2009


1
12/7/2009
< 75
2009


2
4/25/2007
77
2007


2
7/6/2007
< 75
2007


2
8/27/2007
< 75
2007


2
11/26/2007
< 75
2007


2
1/28/2008
< 75
2008


2
6/30/2008
< 75
2008


2
11/5/2008
< 75
2008


2
2/9/2009
< 75
2009


2
5/11/2009
< 75
2009


2
9/23/2009
< 75
2009


2
3/1/2010
< 75
2010

So as you noticed, each person may have multiple labs per year. I want to grab the first lab (by date) for each year for each person. So basically I want a table that looks like this:

id
date
count
year


1
12/5/2007
< 75
2007


1
2/27/2008
500
2008


1
2/23/2009
900
2009


2
4/25/2007
77
2007


2
1/28/2008
< 75
2008


2
2/9/2009
< 75
2009


2
3/1/2010
< 75
2010

So far, I have tried sorting it by ID and date created a dataset using first.date; but that didn't work.

I was wondering if someone can provide me with some help? I am pretty lost on how to do this!

Thanks!

[PGStats · Accepted solution]

It's FIRST.YEAR that must be used, you want the first record (date) within each year :

proc sort data=sample; by id year date; run;

data want;
set sample; by id year;
if first.year;
run;

PG