topic Re: Competing risk in multivariable Cox regression model in Discussions

Competing risk in multivariable Cox regression model

jmpquestions123 — Fri, 09 Jun 2023 01:00:28 GMT

Does JMP have this capability to calculate competing risk for a multivariable Cox regression model? Such as the Fine-Gray hazard model? I found a potential method using the fit parametric survival that allows for competing causes but I don't see where to generate hazard ratios, but I'm also not sure if this is the correct test to begin with. Thank you

Re: Competing risk in multivariable Cox regression model

jmpquestions123 — Sat, 08 Apr 2023 13:57:50 GMT

Similarly, for cumulative incidence curves with competing risks, how can I compare between groups? Is there a Gray's test?

Re: Competing risk in multivariable Cox regression model

peng_liu — Sat, 15 Apr 2023 04:34:07 GMT

Seems these are related but separate questions. Here are some answers:

JMP's Cox Proportional Hazard platform does not offer competing risk analysis at the moment.
Parametric Survival conducts parametric modeling, it is different from the Cox model above. The platform does not produce hazard ratio . You may be able to calculate ratio of two hazard rates from two groups, under comparable conditions, after fitting a model. After fitting a model, you can calculate hazard at any given time. And by definition of hazard ratio, you can calculate a ratio.
JMP offers a Cumulative Incidence Curves in Competing Cause platform. And it is named as "Subdistributions", which is name in the reliability data analysis context. Take a look at the JMP sample data table "Appliance.jmp". Run the embedded script "Life Distribution", and look for the menu item "Show Subdistributions".
I, however, don't think this is what you are looking for, because you mention Cox model, so you have covariates in your data. For the "Subdistributions" feature, the platform does not support covariates.

If you would like to submit a request, please contact Tech Support at support@jmp.com and we would like to collect necessary information.