topic Re: Kaplan Meier stats in Statistical Procedures

Kaplan Meier stats

rajredy7172012 — Wed, 25 May 2022 16:40:30 GMT

Hi:

I'm looking to get KM stats for 25th percentile, median & 75th percentile, but can't understand what's the issue. I have 2 treatment groups instead of 3. 1 group with missing data. Below is the data and the program.

Program:

ods output Quartiles=quartiles;
proc lifetest data=adtte1 method=KM alpha=0.05 conftype=loglog;
where trtpn ^= 2;
time AVAL*CNSR(1);
strata trtpn/notest;
run;

Re: Kaplan Meier stats

Reeza — Wed, 25 May 2022 16:56:35 GMT

What is the issue? What's happening that is unexpected?

Re: Kaplan Meier stats

ballardw — Wed, 25 May 2022 16:58:20 GMT

Are you sure that where Trtpn ^= 2 leaves three treatment groups in the data? From that output it looks like the WHERE statement removed the level 2 values.

Re: Kaplan Meier stats

rajredy7172012 — Wed, 25 May 2022 17:01:03 GMT

Yes, for trtpn=2 i removed the values since they don't have any records.

Re: Kaplan Meier stats

rajredy7172012 — Wed, 25 May 2022 17:02:02 GMT

Issue is I don't get the stats for each treatment group. Each treatment group has 2 records.

Re: Kaplan Meier stats

Reeza — Wed, 25 May 2022 17:05:54 GMT

The data shown is all your input data? 2 records per each group and you want to calculate quartiles off this data?

Re: Kaplan Meier stats

rajredy7172012 — Wed, 25 May 2022 17:06:53 GMT

Yes, that's right. The data is in a developing stage, but keeping my program ready. Just want to make sure the syntax is right.

Re: Kaplan Meier stats

Reeza — Wed, 25 May 2022 17:24:10 GMT

Your syntax is correct, but you cannot calculate quartiles off 2 data points per group.

Re: Kaplan Meier stats

OsoGris — Wed, 25 May 2022 17:30:09 GMT

Your syntax looks OK to me but with only 2 observations per strata I'm not very surprised you are not getting the Quartiles to be computed. For this to happen the survival curve has to drop below the points of S(t) = .75, .50 and .25 for the first, second and third quartiles respectively.

Re: Kaplan Meier stats

Reeza — Wed, 25 May 2022 17:32:52 GMT

The documentation has fake data you can use to ensure you code will report the correct stats, assuming you have data that supports the analysis.

Re: Kaplan Meier stats

ballardw — Wed, 25 May 2022 17:50:15 GMT

@rajredy7172012 wrote:

Yes, for trtpn=2 i removed the values since they don't have any records.

So if you have 3 levels of the variable trtpn and you remove one why would you expect any data in the output for it? And missing records means that you only had two levels to begin with. So really, where is the expectation of getting output for non-existent values?