### Saturday, December 31, 2005

## st: Survival analysis: Interpretation of Cox regression

Hi, I am relatively new to survival analysis. I was hoping I could get some informal advice about interpretation of Cox regression. Thank you in advance.

1) What is the proper interpretation of a significant result in Cox regression? 2) How can I verify if survivor function at a particular time (e.g. 5 years) are statistically different? 3) How do I perform the above comparison adjusting for other factors?

Hypothetical Example: ********************** If the failure is death and the disease is such that every patient is going to die eventually, The main interest would be how long life can be extended using medications v/s surgery.

Cox regression indicates "the probability of winning the race, and not the magnitude by which the race is won".

If treatment with medications is found to be better than surgery in Cox regression using the logrank test, then what does this indicate?

>>> "Patients given medications were more likely to survive as compared to those treated surgically at...<can any time be specified? If not, what is the point of this statement??>"

>>> "Patients given treatment1 survived longer than those given treatment2" (probably wrong interpretation?)

I can verify the second statement using: sts list, by(treatment) compare -thus comparing the survivor functions.

How do compare if the survivor functions were statistically significantly different? Can sts test be used for this?

Multivariate: *************** But this would be univariate analysis. The advantage of Cox regression for me is that multiple factors affecting survival in my disease: Nature of treatment, age, sex, and previous history of the disease.

Is there any way I can make the statement: "5 year survival for the patients treated with medications was better than those treated surgically, adjusting for age, sex and previous history of the disease."?

Will sts list, by(treatment) adjustfor(age,sex,prevhistory) compare -Take care of this?

Thanks and regards, Siddharth.

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Tag: statalist