ggcompetingrisks following multistate Cox PH

[adamssv]

Expected behavior

Should it be possible to fit a multi-state Cox PH model, then generate a ggcompetingrisks() plot given data to fix model covariates...?

Actual behavior

error:
!Names repair functions can't return 'NA' values

Steps to reproduce the problem

Follow example from Thernau et al, with survival package.
(edit, citation: https://cran.r-project.org/web/packages/survival/vignettes/compete.pdf)

library(survival)
mgus2 <- mgus2
cfit1 <- coxph(Surv(etime, event) ~ age + sex + mspike, mgus2, id=id)
print(cfit1, digits=2)

dummy <- expand.grid(sex=c("F","M"), age=60, mspike=1.2)
dummy
csurv <- survfit(cfit1, newdata=dummy)

#this plot works showing probability of death or PCM, the two outcomes
plot(csurv, xmax=25*12, xscale=12, xlab="Years", ylab="Pr(Outcome)", col=1:2, lty=c(1,1,2,2), lwd=2)
legend(10, .14 , outer(c("female", "male "), c("pcm","death"), paste, sep=", ") ,
       col=1:2, lty=c(1,1,2,2), bty='n', lwd=2 )

#should this be able to make a similar plot with ggcompetingrisks() -style "stacked" plot instead of the lines?
ggcompetingrisks(csurv)
# returns error about names being NA

session_info()

# please paste here the result of
devtools::session_info()

survminer v 0.4.9
survival v 3.5-0
R 4.1.2