Using SAS® to Perform Individual Matching in Design of Case-Control Studies - setnumber?

kfwright23 — Tue, 17 Oct 2017 21:21:08 GMT

I attempted to replicate the code listed in the appendix of this paper:

http://support.sas.com/resources/papers/proceedings10/061-2010.pdf

I do not think that this step is working for me. I would greatly appreciate any suggestions to correct my code.

************Select the first case- the step is repeated for all cases**************;

data active;

setnumber = 1;

set cases point=setnumber ;

output;

stop;

run;

Below is the code that I used with my variables, not using the macro:

/****************** Code Generated From Call of Macro *************
Assume a working data set named cases and controls
************************************************************************/
* Sort control dataset by a random number;
proc sql;
create table random_controls as
SELECT *,
ranuni(12345) as random from controls order by random;
quit;

* Rename control variable names - put c_ at beginning (code not shown);
* Select the first case- the step is repeated for all cases;

data active;
setnumber = 1;
set TalTen point=setnumber;
output;
stop;
run;

********************************************************************
Main section of the program. Create dataset for matches, non-matches
********************************************************************;
data match (keep = bannerid pell act_maxcomp hs_gpa3 setnumber ccstat)
nomatch (keep = bannerid pell act_maxcomp hs_gpa3)
used (keep= c_bannerid);
set active ; * This has the current case;

* Read through control dataset with random access and pointer option;
do i = 1 to totobs;
set random_controls point=i nobs=totobs;

* Check if case and control data match;

if abs (pell - c_pell) = 0 and
abs (act_maxcomp - c_act_maxcomp) <= 1 and
abs (hs_gpa3 - c_hs_gpa3) <= .2 then do;
* We have a match!;
ccstat= 1;
output match; * This is the case data, aadding variable ccstat = 1;

* Store control values in variables with same name as case;
* Then output again to same dataset;
pell = c_pell;
act_maxcomp = c_act_maxcomp;
hs_gpa3 = c_hs_gpa3;
bannerid = c_bannerid;
ccstat= 2;
output match;
* Output control data values to USED dataset;
output used;
stop; * Need to end DATA step since we have a match;
end;
end; * ends I loop;
output nomatch; * If I loop is exhausted then no match;
run;

/*SAS Global Forum 2010 Coders' Corner
9*/

proc append data=match
base=matchall;

proc append data=nomatch
base=nomatchall;

* Need to re-sort control dataset by subject id;
proc sort data=random_controls;
by c_bannerid;

* Remove used control from control dataset;
data random_controls;
merge random_controls
used (in=used);
by c_bannerid;
if used ne 1; *aka ^= ;
run;

* Need to resort control dataset by random number for next iteration;;
proc sort data=random_controls;
by random;

Re: Using SAS® to Perform Individual Matching in Design of Case-Control Studies - setnumber?

kfwright23 — Fri, 27 Oct 2017 17:54:50 GMT

see attached PDF

topic Re: Using SAS® to Perform Individual Matching in Design of Case-Control Studies - setnumber? in Statistical Procedures

Using SAS® to Perform Individual Matching in Design of Case-Control Studies - setnumber?

Re: Using SAS® to Perform Individual Matching in Design of Case-Control Studies - setnumber?