Case control matching

kjmathes03 — Fri, 11 Mar 2022 18:36:06 GMT

Hello,

I have a dataset with 87 cases and 1000's of controls. I want to match 2:1 with out replacement for the controls. I want to match on year of procedure (+/- 2 years), diagnosis (which can have 6 different values) and sex. I have searched through many papers online which did not remove duplicate controls.

As well as the above paper which also did not remove duplicates.

I am trying the following code

DATA cases controls;
SET cohort3;
IF fvii_inclusion = 1 THEN OUTPUT cases;
ELSE OUTPUT controls;
RUN;
data control2;
set controls;
time_low=timet-2;
time_high=timet+2;
run;
proc sql;
create table controls_id
as select
one.uniqueid as case_id,
two.uniqueid as control_id,
one.timet as case_time,
two.timet as control_time ,
one.proc as case_proc,
two.proc as control_proc,
one.sex as case_sex,
two.sex as control_sex,

from cases one, control2 two
where ((one.timet between two.time_low and two.time_high) and (one.sex=two.sex) and one.proc=two.proc) ;

I also reviewed the suggestions on hash tags for a similar problem posted but it changed nothing.

HELP much appreciated!

Re: Case control matching

benjamin_2018 — Mon, 14 Mar 2022 14:46:32 GMT

If you want to remove duplicate controls then could you try adding the distinct option in the select statement in your proc sql code?

Then, if you want 2 controls per case, you could try proc surveyselect, something like:

proc surveyselect data=controls_id method=srs sampsize=2;

strata case_id;

run;

topic Re: Case control matching in SAS Programming

Case control matching

Re: Case control matching