Relative risk, one sided

[julie_AF]

Hi all,

As part of a superiority trial, the hypotheses are as follows:

H0:PT = P0 ,
H1: PT > P0.
where PT and P0 denote the proportion of patients with alleviation in the treatment group (test medication, T) and in the placebo group (P), respectively.

I need to provide RRs for a 2x2 table with 95% confidence intervals (α = 0.05, one-sided).

I used this program:

	Proc Freq data= table;
		Tables treat * outcome /  relrisk(sup) alpha = 0.025 cmh;
		output out=rout relrisk cmh;
	run;

Firstly, I'd like confirmation that my code is correct, and secondly, how to define 'margin':

RELRISK(sup margin=)

Indeed, by not defining a margin, it automatically sets it for me

Also, does the p-value obtained in the 'Superiority Analysis for the Relative Risk' table correspond to the RR p-value?

Finally, with the output, I can't obtain the values for the Superiority Analysis for the Relative Risk in a data file. Do you have a solution?

Thank you for your help.

Julie

[Ksharp]

You are doing superiority/noninferiority test which indeed are one-sided test.

If you want more detail info check the attachment I posted.

About defining/selecting a margin,once you know the distribution of the difference between risk, you know how to do.

Check @Rick_SAS  very good blog .

https://blogs.sas.com/content/iml/2023/03/01/distribution-difference-beta.html