Hello all, I'd like to do a 1:1 matching where I want to use control once (i.e, 1 case: 1 control). I found a great article/document showing how to match using proc SQL. The problem is that I want to do 1:1 matching, and I can't figure out how. My codes: ** split into two datasets; data case control; set sample_1; rand_num=uniform(0); if Reinfection_Status="yes" then output case; else if Reinfection_Status="no" then output control; run; ** Simple match: exact match on age, race, and month of primary infection; proc sql; create table controls_id as select one.PatientID as case_PatientID, two.PatientID as control_PatientID, one.Age as case_Age, two.Age as control_Age, one.Ethnicity as case_Ethnicity, two.Ethnicity as control_Ethnicity, one.Month_Pr as case_Month_Pr, two.Month_Pr as control_Month_Pr, one.rand_num as rand_num from case one, control two where (one.Age=two.Age and one.Ethnicity=two.Ethnicity and one.Month_Pr=two.Month_Pr); run; * remove duplicate control subjects; proc sort data=controls_id nodupkey; by control_PatientID rand_num; run; I have 1066 cases and 7088 controls, so I want 1066 cases and 1066 controls to run the analysis. When I ran the proc SQL, I had 1736 observations and when I tried to remove duplicate control subjects using proc sort nodupkey, I had 0 observations with duplicates. How should I go from here..? I hope I'm making sense. Please let me know if you need anything else from me. Thank you.
... View more
