Questions on matching controls to cases?

[gizat]

Dear SAS experts,

For my matched case-control study, I’m preparing my data for a conditional logistic regression.  Using the matching variables, I’ve got a dataset containing the details for cohort (cases and controls identified). At this stage I know my cases and the controls. However, I do not know which controls were matched to which case. I need to make risk sets or strata so that I can perform conditional logistic regression. 

Each case and control have unique identifying variable, I want to identify the controls matched to each case. Could you please help me how I can link the cases to their matched controls and assign them a specific id for each? My matching ration is 1:3.

Looking forward to hearing from you. 

Thank you,

[smantha]

Do you know if the match variables overlap between cases. If not you can use the following logic. Group is the variable that has case or control values.
Proc sort data= have; by match1 match2 match3 group
Data want;
Set have;
By match1 match2 match3;
Retain caseno 0;
If first.match1 then caseno=caseno+1;
Run;
This works well only if match1 match2 and match3 are all discrete values and matches are exact. If you are doing a match on ranges you can split your have dataset into two datasets one containing all the records of cases only, and the other containing all records. Then do a sql join on both datasets.

[gizat]

Hi Smantha,

Yes, they do. I've three matching variables; age (+-2 years), gender (match same gender) and year of hospital admission (+-2 years). What I want is to choose three controls for per case among possible controls and want to identify the controls matched to each case. 

More help is appreciated!

Kind regards,