topic Re: Proc Freq for Risk Ratio, 95% CI of Risk Ratio and p-value with methon Chan and Zhang(1999) in Statistical Procedures

Proc Freq for Risk Ratio, 95% CI of Risk Ratio and p-value with method of Chan and Zhang(1999)

DmytroYermak — Tue, 02 Oct 2018 09:28:11 GMT

Hi all,

Could you please help with the statistics evaluation.

I need to calculate: Risk Ratio, 95% CI of Risk Ratio and p-value with method of Chan and Zhang(1999).

Could you please say the syntax of Proc Freq for the case?

Thank you!

Re: Proc Freq for Risk Ratio, 95% CI of Risk Ratio and p-value with methon Chan and Zhang(1999)

FreelanceReinh — Tue, 02 Oct 2018 09:31:25 GMT

Hi @DmytroYermak,

According to the SAS/STAT 14.3 documentation, that method is the new default, so the RELRISK option of the EXACT statement is sufficient:

exact relrisk;

In earlier SAS/STAT versions (up to 14.2) you had to specify

exact relrisk(method=score);

Re: Proc Freq for Risk Ratio, 95% CI of Risk Ratio and p-value with methon Chan and Zhang(1999)

DmytroYermak — Tue, 02 Oct 2018 09:36:59 GMT

Thank you! Would you please prompt how to calculate p-value within the procedure as well.

Re: Proc Freq for Risk Ratio, 95% CI of Risk Ratio and p-value with methon Chan and Zhang(1999)

FreelanceReinh — Tue, 02 Oct 2018 10:29:24 GMT

There doesn't seem to be an option to compute a corresponding test (although the CI is based on one-sided exact tests). At least I can't find it, sorry. Maybe you can

iteratively modify the ALPHA= option of the TABLES statement until a pre-specified value equals a confidence limit
or compute the test outside of PROC FREQ (i.e. manually in a data step -- I don't have code for this)
or use one of the available tests for the risk difference instead if appropriate.