Interaction Tips Summary

How to extract est and CIs from r object

# Example R code to extract estimates and confidence intervals from a model object
# install.packages("broom")
library(broom)

# Assuming you have a model object called 'model'
model <- lm(mpg ~ hp * cyl, data = mtcars)

# Extracting estimates and confidence intervals
tidy_model <- tidy(model, conf.int = TRUE)

# Display the results
tidy_model

# A tibble: 4 × 7
  term        estimate std.error statistic      p.value conf.low conf.high
  <chr>          <dbl>     <dbl>     <dbl>        <dbl>    <dbl>     <dbl>
1 (Intercept)  50.8      6.51         7.79 0.0000000172 37.4       64.1   
2 hp           -0.171    0.0691      -2.47 0.0199       -0.312     -0.0291
3 cyl          -4.12     0.988       -4.17 0.000267     -6.14      -2.09  
4 hp:cyl        0.0197   0.00881      2.24 0.0332        0.00169    0.0378

# Assuming you have a model object called 'model'
model <- lm(mpg ~ hp * cyl, data = mtcars)

# Extracting estimates and confidence intervals
tidy_model <- tidy(model, conf.int = TRUE)

# Display the results
tidy_model

# A tibble: 4 × 7
  term        estimate std.error statistic      p.value conf.low conf.high
  <chr>          <dbl>     <dbl>     <dbl>        <dbl>    <dbl>     <dbl>
1 (Intercept)  50.8      6.51         7.79 0.0000000172 37.4       64.1   
2 hp           -0.171    0.0691      -2.47 0.0199       -0.312     -0.0291
3 cyl          -4.12     0.988       -4.17 0.000267     -6.14      -2.09  
4 hp:cyl        0.0197   0.00881      2.24 0.0332        0.00169    0.0378

Extraction of data with gtsummary

Now we will use the gtsummary package to extract the estimates and confidence intervals from a model object.

# Example R code to extract estimates and confidence intervals from a model object using gtsummary
# install.packages("gtsummary", dependencies = TRUE)
library(gtsummary)

# Assuming you have a model object called 'model'
model <- lm(mpg ~ hp * cyl, data = mtcars)

# Extracting estimates and confidence intervals
tidy_model <- tidy(model, conf.int = TRUE)

# Display the results
tidy_model

# A tibble: 4 × 7
  term        estimate std.error statistic      p.value conf.low conf.high
  <chr>          <dbl>     <dbl>     <dbl>        <dbl>    <dbl>     <dbl>
1 (Intercept)  50.8      6.51         7.79 0.0000000172 37.4       64.1   
2 hp           -0.171    0.0691      -2.47 0.0199       -0.312     -0.0291
3 cyl          -4.12     0.988       -4.17 0.000267     -6.14      -2.09  
4 hp:cyl        0.0197   0.00881      2.24 0.0332        0.00169    0.0378

# Extracting estimates and confidence intervals
tbl_regression(model)

Characteristic	Beta	95% CI	p-value
hp	-0.17	-0.31, -0.03	0.020
cyl	-4.1	-6.1, -2.1	<0.001
hp * cyl	0.02	0.00, 0.04	0.033
Abbreviation: CI = Confidence Interval

Options of gtsummary

The gtsummary package provides several options for customizing the output of tbl_regression. Here are a few examples:

Exponentiating coefficients (eform)

To exponentiate the coefficients (useful for logistic regression models), you can use the exponentiate = TRUE argument:

logistic_model <- glm(am ~ hp + cyl, data = mtcars, family = binomial)

# Exponentiating coefficients
tbl_regression(logistic_model, exponentiate = TRUE)

Characteristic	OR	95% CI	p-value
hp	1.03	1.00, 1.06	0.042
cyl	0.18	0.04, 0.50	0.005
Abbreviations: CI = Confidence Interval, OR = Odds Ratio

Inlind estimate of gtsummary

To extract inline estimates and 95% confidence intervals for the first term (hp) using gtsummary, you can use the inline_text function. Here is how it looks:

# Example R code to extract inline estimates and 95% confidence intervals using gtsummary
# install.packages("gtsummary", dependencies = TRUE)
library(gtsummary)

# Assuming you have a model object called 'model'
model <- lm(mpg ~ hp * cyl, data = mtcars)

# Extracting inline estimates and confidence intervals for the first term (hp)
inline_text(tbl_regression(model), variable = "hp")

[1] "-0.17 (95% CI -0.31, -0.03; p=0.020)"

# or have only est and CI in without pvalue
inline_text(tbl_regression(model), variable = "hp", pattern = "{estimate} [{conf.low}, {conf.high}]")

[1] "-0.17 [-0.31, -0.03]"

Now let’s see how that could be used in Quarto as an inline command within formatted text, so the estimates of -0.17 (95% CI -0.31, -0.03; p=0.020) can be displayed directly in the document. To do so we are writing text as follows:

Now let's see how that could be used in Quarto as an inline 
command within formatted text, so the estimates of 
-0.17 (95% CI -0.31, -0.03; p=0.020) 
can be displayed directly in the document.