Survminer - ggforest, how to change variable names?

Question

Hello to everybody and thank you for your help. I'm managing to build a forest plot of a Cox model with the Survminer package, however, I've got stuck in trying to change the labels of the variables displayed. Is there a way to change the labels displayed without changing variable names? How could I update my string in order to change the names?

ggforest<-ggforest(coxph, data=data)

This image is obtained

If I try to use the legend.labs function I obtain this error

ggforest(coxph, data=data, legend.title = c("Prova"),legend.labs=c("Carried AED","Daytime", "Distance 1","Distance 2", "Distance 3"))
Error in ggforest(coxph, data = data, legend.title = c("Prova"), legend.labs = c("Carried AED",  : 
  unused arguments (legend.title = c("Prova"), legend.labs = c("Carried AED", "Daytime", "Distance 1", "Distance 2", "Distance 3"))

Can you help me with changing these labels?

Thank you again

Could you please share some reproducible data using `dput`? – Quinten Jul 31 '22 at 18:09 — Quinten, Jul 31 '22 at 18:09

score 3 · Accepted Answer · answered Jan 24 '23 at 14:29

Here is a workaround based on @Quintens answer:

library(forestmodel)
library(survival)

labelled::var_label(colon) <- list(
    sex = "Sex", #this variable is a numeric -> label works
    rx = "RX", # factor -> label doesn't work!
    adhere = "ADHERE" # numeric -> label works...
)

model <- coxph(Surv(time, status) ~ sex + rx + adhere, data = colon)

print(forest_model(model))

# this does the trick
model$xlevels <- unname(model$xlevels)

print(forest_model(model))

^{Created on 2023-01-24 with reprex v2.0.2}

score 2 · Answer 2 · answered Jul 31 '22 at 18:56

Based on this GitHub issue it seems not possible to change the printed variables' names in a ggforest without changing the variable names. They suggest to use the labelled package to label the variable names with the forestmodel. One problem here is that changing the names only works for numerical variables. I don't have your data, so I will give a reproducible example using the colon dataset:

library(forestmodel)
library(survival)
labelled::var_label(colon) <- list(
  sex = "Sex", #this variable is a numeric -> label works
  rx = "RX", # factor -> label doesn't work!
  adhere = "ADHERE" # numeric -> label works...
)

model <- coxph(Surv(time, status) ~ sex + rx + adhere, data = colon)

print(forest_model(model)) 
#> Warning in recalculate_width_panels(panel_positions, mapped_text =
#> mapped_text, : Unable to resize forest panel to be smaller than its heading;
#> consider a smaller text size

^{Created on 2022-07-31 by the reprex package (v2.0.1)}

Check this conversation for extra info.

I wonder if giving the factor entry a list with level names might succeed? It would be more in the classical R style. — IRTFM, Jul 31 '22 at 19:54

Survminer - ggforest, how to change variable names?

2 Answers2