Choose Language Hide Translation Bar

Community Member

Joined:

Mar 15, 2016

## Cox proportional hazards

Hi!

I am very new to this forum and to JMP&statistics in general. I would appreciate if somebody could advise me on a following issue:

I am analyzing whether certain factors (A, B, C etc - all of them are continuous variables) are predictors of a disease (X) incidence. I have 2 independent groups of patients (group 1 - healthy, and group 2 - those ended up with a disease). I need to adjust those factors for age, gender, race etc. When I go to Analyze->Reliability and Survival->Fit Proportional hazards I am a little bit puzzled what goes where. Could you please advise?

I apologize those of you who will see this question as a very basic one. I was trying to look it up online but could not fins a definite answer anywhere

I will greatly appreciate your help!

Casandra

1 ACCEPTED SOLUTION

Accepted Solutions

Staff

Joined:

Jul 7, 2014

Solution

## Re: Cox proportional hazards

In this Mastering JMP WebEx, I showed how to fit a Cox's PH model to the data similar to yours. This is the link to the recordings: Advanced Mastering JMP: Analyzing Survival Data | JMP

Hope it helps.

3 REPLIES 3

Staff

Joined:

Jul 7, 2014

Solution

## Re: Cox proportional hazards

In this Mastering JMP WebEx, I showed how to fit a Cox's PH model to the data similar to yours. This is the link to the recordings: Advanced Mastering JMP: Analyzing Survival Data | JMP

Hope it helps.

Community Trekker

Joined:

Nov 20, 2012

## Re: Cox proportional hazards

Hi Jian

Does Jmp pro 13.2 have the option for time-dependent or time-varying covariates. I tried to find some info but I could not.

Hope you can shed more light on my question

Thank you

Chris

Staff

Joined:

Jul 22, 2014

## Re: Cox proportional hazards

It may be that the proportional hazards is not the appropriate approach here. You should learn from the Mastering webex that this method is for when you have a time-to-event outcome, such as time-to-death. These methods also deal with censoring. For example, where you have subjects at the end of the observed time period who have still not experienced the event.

It sounds like in your situation all you know about the outcome for the subject is a binary: healthy or with-disease (1 / 2). So this would be handled with a logistic regression.