Improve error message when user passes survfit.coxphms model

[steliosbl]

Hello,
I am attempting to plot "probability in state" curves from a multi-state object created with survfit2() from a Cox proportional hazards model. Unfortunately, when calling ggcuminc, I encounter the following error:

Error in data.frame(time = x$time, n.risk = c(x$n.risk), n.event = c(x$n.event),  : 
 arguments imply differing number of rows: 286, 858, 3432, 0

This error occurs when following the standard steps for fitting a multi-state Cox model per the survival package's multi-state vignette. I have written a minimal, reproducible example of the error below following the vignette's steps with a stock dataset.

library(survival)
library(ggsurvfit)

data(cancer, package="survival")

# First, we prepare the multi-state dataset.
# This code is copied directly from the survival package's multi-state vignette, section 2.2
ptemp <- with(mgus2, ifelse(ptime==futime & pstat==1, ptime-.1, ptime))
data3 <- tmerge(mgus2, mgus2,  id=id, death=event(futime, death),
                pcm = event(ptemp, pstat))
data3 <- tmerge(data3, data3, id, enum=cumtdc(tstart))
temp <- with(data3, ifelse(death==1, 2, pcm))
data3$event <- factor(temp, 0:2, labels=c("censor", "pcm", "death"))

# Finally, we fit the Cox PH model and run survfit2
msfit <- coxph(Surv(tstart, tstop, event) ~ sex + age, data=data3, id=id)
ndata <- expand.grid(sex=c("F", "M"), age=c(60, 80))
mssurv <- survfit2(msfit, newdata=ndata, p0=c(1,0,0))

# Run ggcuminc
ggcuminc(mssurv, outcome=c("censor", "pcm", "death"))

> Error in data.frame(time = x$time, n.risk = c(x$n.risk), n.event = c(x$n.event),  : 
  arguments imply differing number of rows: 286, 858, 3432, 0

Does ggcuminc support survfit.coxphms objects, i.e. survfit objects generated from a multi-state Cox model? Or am I making a mistake when using it?

Thank you for your assistance.

[ddsjoberg]

Dear @steliosbl , thanks for the post.

The package doesn't support visualizations from the regression models. But we can certainly improve the error message in this setting.