SAS day 8: Proc Freq Risk Difference
Goal :
- What is Risk Difference?
- How to interpret Risk Difference Confidence Interval.
Background:
Risk Difference is used to compare the proportions of successes between two independent groups with respect to a binary outcome. (cancer remission).
If the outcome is continuous, we compare the difference in means instead.
Risk Difference:
![\[RD= p1- p2\]](https://i1.wp.com/fangya18.com/wp-content/ql-cache/quicklatex.com-05085f2ffa5a3b0354298ca3bae922b7_l3.png?resize=109%2C16&ssl=1 "Rendered by QuickLaTeX.com")
where 
(
is the successes in group 1) ,
(
is the successes in group 2).
ColiN00B / Pixabay
Example:
H0 (Null Hypotheses) : 
: there is no difference in outcomes between Treatment A and Placebo.
Outcome:
| Treatment A | Treatment B | Placebo | |
|---|---|---|---|
| n(%) | 21(29.3) | 25(25.1) | 6(9.1) |
| Difference From Placebo | 18 | 11.3 | |
| p-value for difference | 0.0004 | 0.067 | |
| 97.5% CI for Difference | (9.2,35,6) | (-2.2,22.3) |
Interpretation:
We are 95% confident that the difference in proportion in Treatment A and Placebo is (9.2, 35.6), therefore we conclude the treatment outcome difference in success is significant.
Note: If we use the WALD option, we might get negative values for CI, which means, the difference is not significant.
SAS Code:
PROC FREQ DATA=adtte;
where (trt01pn in (1 2 ));
BY avisitn ;
TABLES trt01pn*aval / riskdiff(cl=wald) alpha=0.025;
exact riskdiff;
RUN;
data riskci;
set riskci;
col1 = put(-riskdifference*100, 4.1);
keep avisitn col1;
run;
Happy SAS Coding 😍!
Thanks very much to Renee 5, Shared the Risk Diff knowledge with me ! 🍉