Dear community, I have a question regarding the standard error of "marginal hazard ratio" estimates from a cox proportional hazard model with an interaction term.
I have to dichotomous variables ADMA and ALA (both low/high) where I have divided the continuous variables at a certain level. I am aware about the contra arguments categorizing a continous variable and I could as well use the variables continuously. Just to keep my question simple. This is the function call and output: > coxph( ami ~ ADMA2 * ALA2 , data = d2) Call: coxph(formula = ami ~ ADMA2 * ALA2, data = d2) coef exp(coef) se(coef) z p ADMA2high 1.137 3.118 0.302 3.759 0.00017 ALA2high 0.148 1.160 0.196 0.756 0.45000 ADMA2high:ALA2high -0.671 0.511 0.448 -1.496 0.13000 Likelihood ratio test=13.1 on 3 df, p=0.00437 n= 1364, number of events= 129 For publication I would like to present the marginal estimates like | | ADMA low | ADMA high | |----------+------------------+------------------| | ALA low | Reference | 3.12 (1.72,5.63) | |----------+------------------+------------------| | ALA high | 1.16 (0.79,1.70) | 1.85 ( ? , ? ) | Now, the HR for high AMDA, high ALA would be exp(1.137 + 0.148 - 0.671) = 1.85 adding up the betas from output, but how would I calculate a 95% CI for this estimate? Regards, Reinhard [[alternative HTML version deleted]] ______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.