ggflexsurvplot removing groups from KM plot

[andersbergren]

ggflexsurvplot() should show the KM curve with model fit curves for all strata. Instead the strata for the KM curve have been relabelled "All".

May be related to the length(summary(fit)) check that should only relabel strata to "All" if if the summary list length equals 1.

Here you can see my model fit curves and and the broken red "All" KM curve which should be broken out by strata.

[andersbergren]

Seems related to this code in the function:
isfac <- .is_all_covariate_factor(fit) if(!all(isfac)){ .formula <- .build_formula(surv.obj, "1") n.strata <- 1 }
It creates a vector that's not a factor, then checks if that vector is a factor. Always returning false I believe.

[andersbergren]

Result when check is removed

[kassambara]

ggflexsurvplot() expected the grouping variable to be strictly a factor for showing grouped KM.

fixes: now, the grouping variable can be either a factor or a character vector (in your example).

Should work now, thanks!

You can install the latest dev version as follow:

devtools::install_github("kassambara/survminer", build_vignettes = FALSE, dependencies = FALSE)

[jumbanhounc]

I'm still getting this problem. I think the problem is in the function is_factor_or_character has:
'is.facet(x) | is.character(x) and should be 'is.factor(x) | is.character(x)

[igutermes]

The issue sting lingers.

Just in case anyone is still stuck in this problem, just use a as.character() in your model and the function will work as expected.