hazard ratio of interaction Cox model
Giulia Barbati wrote:
Dear Forum, I have a question about interaction estimate in the Cox model: why the hazard ratio of the interaction is not produced in the summary of the model? (Instead, the estimate of the coefficient is given in the print of the model.)
The 'hazard ratio of the interaction' is not well defined. Decide what hazard ratio you want to estimate, then ask summary to compute that, e.g. summary(modINT, XX_PR=c(1,3), XX_DISF=2) will estimate the 3:1 XX_PR hazard ratio at XX_DISF=2 Frank Harrell
# Example:
modINT <-cph( Surv(T_BASE, T_FIN,STATUS)~ NYHA + ASINI + RFP + FE_REC + XX_PR*XX_DISF)
print(modINT)
coef se(coef) z p
NYHA=2 1.254 0.584 2.15 0.031767
ASINI 0.665 0.409 1.62 0.104247
RFP=2 0.725 0.704 1.03 0.302578
FE_REC=2 -1.637 0.810 -2.02 0.043331
XX_PR 2.189 0.649 3.37 0.000748
XX_DISF 3.233 1.000 3.23 0.001222
XX_PR * XX_DISF -2.852 1.280 -2.23 0.025852
summary(modINT)
Effects Response : Surv(T_BASE, T_FIN, STATUS)
Factor Low High Diff. Effect S.E. Lower 0.95 Upper 0.95
ASINI 2.0725 2.85 0.7775 0.52 0.32 -0.11 1.14
Hazard Ratio 2.0725 2.85 0.7775 1.68 NA 0.90 3.13
XX_PR 0.0000 1.00 1.0000 2.19 0.65 0.92 3.46
Hazard Ratio 0.0000 1.00 1.0000 8.92 NA 2.50 31.86
XX_DISF 0.0000 1.00 1.0000 3.23 1.00 1.27 5.19
Hazard Ratio 0.0000 1.00 1.0000 25.35 NA 3.57 179.88
NYHA - 2:1 1.0000 2.00 NA 1.25 0.58 0.11 2.40
Hazard Ratio 1.0000 2.00 NA 3.50 NA 1.12 11.00
RFP - 2:1 1.0000 2.00 NA 0.73 0.70 -0.65 2.10
Hazard Ratio 1.0000 2.00 NA 2.07 NA 0.52 8.20
FE_REC - 2:1 1.0000 2.00 NA -1.64 0.81 -3.23 -0.05
Hazard Ratio 1.0000 2.00 NA 0.19 NA 0.04 0.95
Adjusted to: XX_PR=0 XX_DISF=0
____________________________________________________________________________________ Be a better friend, newshound, and [[alternative HTML version deleted]] ______________________________________________ R-help at 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.
Frank E Harrell Jr Professor and Chair School of Medicine
Department of Biostatistics Vanderbilt University