topic Re: Identifying outcome status with identification numbers in SAS Procedures

Identifying outcome status with identification numbers

Mdormond — Fri, 21 Mar 2014 01:47:36 GMT

I'm working on a case control study and the cases and controls are identified using a letter at the end of the identification number. Some cases are matched 1:1 with controls and some are matched 1:2, or 1:3. A case/control pair may be identified as follows: GI111A, GI111B, GI111X with A and B denoting control status and X denoting case status. I want to perform analyses comparing cases and controls and therefore want to make a new variable "case status" defined by the ending letter. Is there a procedure I can use to scan the identification numbers and then identify cases and control, respectively? I haven't tried any code yet, I was just wondering if there was a quick and easy way to do this?

Re: Identifying outcome status with identification numbers

PGStats — Fri, 21 Mar 2014 02:15:27 GMT

You could use:

case = char(id,length(id)) = "X";

in a datastep, where id is your identifier. The value of case will be 1 for cases and 0 for controls.

Re: Identifying outcome status with identification numbers

Mdormond — Fri, 21 Mar 2014 17:06:15 GMT

Would this create a new variable? Would I code it like creating a new variable?

Re: Identifying outcome status with identification numbers

PGStats — Fri, 21 Mar 2014 18:51:39 GMT

Yes, the variable case would be created by the statement.You would use that variable in your analysis/models.

Re: Identifying outcome status with identification numbers

Mdormond — Fri, 21 Mar 2014 19:14:49 GMT

What if all my preceding numbers are not the same length? I'm still confused on exactly how to code this?

Re: Identifying outcome status with identification numbers

PGStats — Fri, 21 Mar 2014 19:24:47 GMT

Please post the SAS code you are using now to prepare your data for analysis and we'll take it from there.