Search code examples
rcox-regression

How do I add covariates to a cox model via a function input

I have a cox model as such

cntrl_reduced <- coxph(Surv(time, status) ~ A + B + C, data=df)

When I run this I get no errors and the model works. I am writing this into a function and would like the predictor/covariables to be modified.

So I have the input to the function as

covariates= c('A', 'B', 'C')

and within the function I unlist this

covariates <- noquote(paste(covariates, collapse =' + '))

covariates
output: A + B + C

However, when I try to run

cntrl_reduced <- coxph(Surv(time, status) ~ covariates, data=df)

I get the error

Error in model.frame.default(formula = Surv(time, status) ~ covariates,  : 
  variable lengths differ (found for 'covariates')

How can I set it up so I can make my covariate input into a variable, rather than hard coded?

Solution

  • You can't put that kind of object after the ~ symbol. Instead you need to use as.formula. For example

    covariates <- c("A", "B", "C")
formulaString <- paste("Surv(time, status) ~", paste(covariates, collapse="+"))
coxph(as.formula(formulaString), data=df)