topic Re: survival analysis (Kaplan-Meyer) in Statistical Procedures

survival analysis (Kaplan-Meyer)

michtka — Tue, 30 Apr 2013 22:36:41 GMT

Hi all, I got the next result from survival analysis (Kaplan-Meyer method) on time of failure of a treatment before complete week 9.

I have got the next result using PROC LIFETEST:

Here you are the part corresponding to the drug...placebo part is not important for my questions:

treatment week censor survival number failed number left

drug 0 0 1 0 23

drug 3.0 0 2 25

drug 3.0 0 0.8 3 24

drug 9.0 1 0.62 10 17

drug 9.1 1 0.55 12 15

drug 9.2 1 0.53 11 14

And I want to create the final table:

week survival failed number left treatment

3.0

6.0

9.0

I've got the next questions, I would like you to help me with:

1) Which value I need get to the final table for duration=3? The last one, that is populated?

2) Which value I need to get to the final table for durantion=6.0?...blank data, or I need to imputed it with the LOCF from the duration=3.0?

3) Which value I need to get to the final table for duration=9.0?...9.0, or can be possible 9.1 and 9.2 too?

I hope you can help me with this.

Thanks in advance.

Re: survival analysis (Kaplan-Meyer)

Reeza — Wed, 01 May 2013 00:31:56 GMT

in proc lifetest, specify timelist=(3, 6, 9).

proc lifetest data=have timelist=(3,6,9);

There might be another option you can use as well to supress the list.

Also, the KM curve is stepped, so if you're missing a value you tend to take the lower or higher value depending on what you've specified.

Re: survival analysis (Kaplan-Meyer)

michtka — Wed, 01 May 2013 00:56:35 GMT

Thanks Reza, for example, because KM curve is stepped, if I had to include week 4 that is not in the result, I would consider the survival value of week3?, because the function will be flat between [week3 and week9) with the value =0.8.

Is it right considerer Survival(Week4)=Survival(Week3)=0.8?

Thanks in advance.

Re: survival analysis (Kaplan-Meyer)

michtka — Wed, 01 May 2013 01:02:12 GMT

Yes, timelist opcion is doing it...thank you very much for your help Reeza.

Re: survival analysis (Kaplan-Meyer)

Reeza — Wed, 01 May 2013 02:57:50 GMT

Just something to keep in mind, SAS will fill in the values appropriately

Re: survival analysis (Kaplan-Meyer)

SteveDenham — Wed, 01 May 2013 13:03:26 GMT

SAS will fill in the values. Sometimes it fills them in somewhat differently than you might want

Steve Denham