如何确定此Emmeans（）输出

Question

library(survival)
library(emmeans)

# Fit model with covariate-covariate interaction
fit <- coxph(Surv(time, status) ~ trt * karno, data = veteran)
summary(fit)

# Estimate trt HR at different values of karno
emm_df <-  emmeans(fit, ~ trt + karno, at = list(karno = seq(0, 100, by = 5)), type = "unlink") |> pairs(rev = T) |> summary(infer = T)

head(emm_df)
 contrast                  ratio     SE  df asymp.LCL asymp.UCL null z.ratio p.value
 trt2 karno0 / trt1 karno0 2.984 1.8095 Inf    0.2785    31.970    1   1.803  0.9996
 trt1 karno5 / trt1 karno0 0.885 0.0346 Inf    0.7592     1.031    1  -3.139  0.4244
 trt1 karno5 / trt2 karno0 0.296 0.1715 Inf    0.0309     2.848    1  -2.102  0.9922
 trt2 karno5 / trt1 karno0 2.438 1.4338 Inf    0.2445    24.314    1   1.516  1.0000
 trt2 karno5 / trt2 karno0 0.817 0.0263 Inf    0.7206     0.927    1  -6.285  <.0001
 trt2 karno5 / trt1 karno5 2.756 1.5413 Inf    0.3095    24.550    1   1.813  0.9996

我如何仅在相同值的相同值下仅筛选

trt2/trt1

的对比度？例如我只想保留：

karno

trt2 karno0 / trt1 karno0

trt2 karno5 / trt1 karno5

等 如果我可以即时将此变量转换为数字（准备绘制），例如将上面的类别转换为

trt2 karno10 / trt1 karno10

thanks

使用

0,5,10,etc

符号：

|

Answer 1