get the results of several regressions in a single dataframe

April 25, 2022

I have a dataframe like this one, with 10 columns Alcohol (from alcohol1 to alcohol10)

ID  Status  matching  Alcohol1  Alcohol2
 1     1       1           1      0
 2     0       1           0      1
 3     0       2           0      1
 4     1       2           0      0

I have many logistic regression models to run using the alcohol columns as explanatory variables. I have created a function that performs this.

And now I would like to summarize the results of all my models in one data frame.
I need my estimated coefficient, the lower bound of my 95% confidence interval, the upper bound and the pvalue. I managed to get my estimated coefficient and the pvalue but I can’t get the confidence interval (or the variance which would also fit) (which is normally obtained with a summary)

Here are the functions I used:

library(broom)
library(purrr)

f1 <- function(column) {
  tidy(clogit(
    as.formula(
     paste("status ~", column, "+ strata(matching)")),
    data = table
  ))
 }

model_results <- map_dfr(
  set_names(names(table)[4:14]), 
  f1
)

The expected result would be something like this

term      estimate   lower.95   upper.95    pvalue
Alcohol1                
Alcohol2                
…

thank you in advance for the help

>Solution :

You haven’t really given us a reproducible example, but I’m going to guess that adding conf.int = TRUE to your tidy() call will do what you want (the default names are conf.low and conf.high, you can add a call to rename if you want).

I made the workflow "pipier" for fun (but your way is perfectly fine). There might be a little bit of select stuff required, or possibly adding a .id = argument to map_dfr (can’t tell without a MCVE).

model_results <- (
 names(table)[4:14]
   %>% set_names()
   %>% map(~ reformulate(c(., "strata(matching)"), response = "status"))
   %>% map(clogit, data = table)
   %>% map_dfr(tidy, conf.int = TRUE, .id = "column")
)