topic Re: Kaplan meier estimated risk ratio in SAS Programming

Kaplan meier estimated risk ratio

CathyVI — Thu, 20 Jan 2022 20:36:09 GMT

I am stocked with a KM estimated risk ratio, hazard ratio, adjusted hazard ratio. Is there a code to analyze this estimates in sas?

Re: Kaplan meier estimated risk ratio

Reeza — Thu, 20 Jan 2022 20:47:15 GMT

Are you sure you're asking the correct question? Can you add more details as to what you're trying to do here?

Re: Kaplan meier estimated risk ratio

tarheel13 — Thu, 20 Jan 2022 21:35:12 GMT

I know you can get hazard ratio with proc phreg.

Re: Kaplan meier estimated risk ratio

CathyVI — Tue, 25 Jan 2022 23:08:58 GMT

@Reeza @tarheel13

I got the Kaplan Meier estimated risk ratio (0.377) from the Product-Limit Survival Estimates of Treatment & placebo by dividing the failure rate of the treatment group by the failure rate of the placebo group and subtract the result from 1 i.e., (1-failure rate treatment / failure rate Placebo).

So my first questions is cant SAS do it for me without me manually doing the division and subtraction for the proportion? This is my code that produces the Product-Limit Survival Estimates.

proc lifetest data=a ;
time specimen*censor6(0);
strata pp/TEST;
run;

Second question is how do I calculate the p-value and confidence Interval just like in the table table, I have tried to add (TEST function for p-value of a homogeneity test specified in the STRATA statement but I didn't get it.

Re: Kaplan meier estimated risk ratio

Reeza — Tue, 25 Jan 2022 23:15:38 GMT

https://gist.github.com/statgeek/d3bce2a9e2ef0523db9d

Do you have raw data and you're trying to calculate your ratios as in the first post?

@CathyVI wrote:

@Reeza @tarheel13

I got the Kaplan Meier estimated risk ratio (0.377) from the Product-Limit Survival Estimates of Treatment & placebo by dividing the failure rate of the treatment group by the failure rate of the placebo group and subtract the result from 1 i.e., (1-failure rate treatment / failure rate Placebo).

So my first questions is cant SAS do it for me without me manually doing the division and subtraction for the proportion? This is my code that produces the Product-Limit Survival Estimates.

proc lifetest data=a ;
time specimen*censor6(0);
strata pp/TEST;
run;

Second question is how do I calculate the p-value and confidence Interval just like in the table table, I have tried to add (TEST function for p-value of a homogeneity test specified in the STRATA statement but I didn't get it.

Re: Kaplan meier estimated risk ratio

CathyVI — Wed, 26 Jan 2022 00:17:00 GMT

@Reeza Yes, I have a raw data but the GitHub link seems complicated for me.

Re: Kaplan meier estimated risk ratio

Reeza — Wed, 26 Jan 2022 01:02:42 GMT

Did you run it? You would need to modify where it saves the output, last parameter.
Run it, see if something in there is similar to what you need then figure out how you can call it for your data.
You're just changing the last line of code.

Re: Kaplan meier estimated risk ratio

CathyVI — Wed, 26 Jan 2022 16:13:03 GMT

@Reeza I don't want the CI in percent. I want the Confidence interval and p-value of the failure rate.

Re: Kaplan meier estimated risk ratio

Reeza — Wed, 26 Jan 2022 16:26:13 GMT

"I got the Kaplan Meier estimated risk ratio (0.377) from the Product-Limit Survival Estimates of Treatment & placebo by dividing the failure rate of the treatment group by the failure rate of the placebo group and subtract the result from 1 "

I'm not familiar with that calculation so not sure how you'd go about getting a confidence interval for it to be honest.

Re: Kaplan meier estimated risk ratio

CathyVI — Wed, 26 Jan 2022 17:14:32 GMT

@Reeza I saw one of your previous comment online

Solved: How to get confidence interval for failure rate in... - SAS Support Communities and maybe you can use this comment to help me.

In the table above, I wanted to calculate the Vaccine Efficacy(VE) 0.377 and its CI and P-value. This is the same dataset so i should get the same result.

I used the follow code to the get the failure rate, see table below. When I divided the failure rates(yellow highlighted) from each other I got 0.3590/0.5778 = 0.6213222568. 1-0.6213222568 = 0.3786

which i think is the same VE result they got. How do i get this result in sas without manually calculating it.

proc lifetest data=a ;
time specimenint*censor6(0);
strata pp;
run;

Thanks

Re: Kaplan meier estimated risk ratio

Reeza — Wed, 26 Jan 2022 17:31:05 GMT

You can estimates for a specific time point. Was there something in the experimental design about the time point/end point? You would put that in the TIMELIST component to get estimates at that specific time point.