Regression tables with {gtsummary}

On to Table 2!

Univariate regressions

Fit a series of univariate regressions of income on other variables.

tbl_uvregression(
  nlsy, 
  y = income,
  include = c(sex_cat, race_eth_cat,
              eyesight_cat, income, age_bir),
  method = lm)
Characteristic N Beta 95% CI1 p-value
age_bir 4,773 595 538, 652 <0.001
sex_cat 10,195
    Male — —
    Female -358 -844, 128 0.15
race_eth_cat 10,195
    Hispanic — —
    Black -1,747 -2,507, -988 <0.001
    Non-Black, Non-Hispanic 3,863 3,195, 4,530 <0.001
eyesight_cat 6,789
    Excellent — —
    Very good -578 -1,319, 162 0.13
    Good -1,863 -2,719, -1,006 <0.001
    Fair -4,674 -5,910, -3,439 <0.001
    Poor -6,647 -9,154, -4,140 <0.001
1 CI = Confidence Interval

Can also do logistic regression

tbl_uvregression(
  nlsy, 
  y = glasses,
  include = c(sex_cat, race_eth_cat,
              eyesight_cat, glasses, age_bir),
  method = glm,
  method.args = list(family = binomial()),
  exponentiate = TRUE)
Characteristic N OR1 95% CI1 p-value
age_bir 5,813 1.02 1.01, 1.03 <0.001
sex_cat 8,450
    Male — —
    Female 1.97 1.81, 2.15 <0.001
race_eth_cat 8,450
    Hispanic — —
    Black 0.76 0.67, 0.86 <0.001
    Non-Black, Non-Hispanic 1.34 1.19, 1.50 <0.001
eyesight_cat 8,444
    Excellent — —
    Very good 0.93 0.84, 1.03 0.2
    Good 0.95 0.84, 1.07 0.4
    Fair 0.81 0.68, 0.96 0.016
    Poor 1.15 0.81, 1.63 0.4
1 OR = Odds Ratio, CI = Confidence Interval

We probably want to do some multivariable regressions

linear_model <- lm(income ~ sex_cat + age_bir + race_eth_cat, 
                   data = nlsy)
linear_model_int <- lm(income ~ sex_cat*age_bir + race_eth_cat, 
                   data = nlsy)
logistic_model <- glm(glasses ~ eyesight_cat + sex_cat + income, 
                      data = nlsy, family = binomial())

gtsummary::tbl_regression()

tbl_regression(
  linear_model, 
  intercept = TRUE,
  label = list(
    sex_cat ~ "Sex",
    race_eth_cat ~ "Race/ethnicity",
    age_bir ~ "Age at first birth"
  ))
Characteristic Beta 95% CI1 p-value
(Intercept) 2,147 493, 3,802 0.011
Sex
    Male — —
    Female 25 -654, 705 >0.9
Age at first birth 438 381, 495 <0.001
Race/ethnicity
    Hispanic — —
    Black -772 -1,714, 171 0.11
    Non-Black, Non-Hispanic 7,559 6,676, 8,442 <0.001
1 CI = Confidence Interval

gtsummary::tbl_regression()

tbl_regression(
  logistic_model, 
  exponentiate = TRUE,
  label = list(
    sex_cat ~ "Sex",
    eyesight_cat ~ "Eyesight",
    income ~ "Income"
  ))
Characteristic OR1 95% CI1 p-value
Eyesight
    Excellent — —
    Very good 0.92 0.82, 1.03 0.2
    Good 0.92 0.80, 1.05 0.2
    Fair 0.80 0.66, 0.98 0.028
    Poor 1.03 0.69, 1.53 0.9
Sex
    Male — —
    Female 2.04 1.85, 2.25 <0.001
Income 1.00 1.00, 1.00 <0.001
1 OR = Odds Ratio, CI = Confidence Interval

Arguments

Argument Description

label=

 modify variable labels in table

exponentiate=

 exponentiate model coefficients

include=

 names of variables to include in output. Default is all variables

show_single_row=

 By default, categorical variables are printed on multiple rows. If a variable is dichotomous and you wish to print the regression coefficient on a single row, include the variable name(s) here.

conf.level=

 confidence level of confidence interval

intercept=

 indicates whether to include the intercept

estimate_fun=

 function to round and format coefficient estimates

pvalue_fun=

 function to round and format p-values

tidy_fun=

 function to specify/customize tidier function

You could put several together

tbl_no_int <- tbl_regression(
  linear_model, 
  intercept = TRUE,
  label = list(
    sex_cat ~ "Sex",
    race_eth_cat ~ "Race/ethnicity",
    age_bir ~ "Age at first birth"
  ))

tbl_int <- tbl_regression(
  linear_model_int, 
  intercept = TRUE,
  label = list(
    sex_cat ~ "Sex",
    race_eth_cat ~ "Race/ethnicity",
    age_bir ~ "Age at first birth",
    `sex_cat:age_bir` ~ "Sex/age interaction"
  ))

You could put several together

tbl_merge(list(tbl_no_int, tbl_int), 
          tab_spanner = c("**Model 1**", "**Model 2**"))
Characteristic Model 1 Model 2
Beta 95% CI1 p-value Beta 95% CI1 p-value
(Intercept) 2,147 493, 3,802 0.011 4,064 1,884, 6,245 <0.001
Sex
    Male — — — —
    Female 25 -654, 705 >0.9 -3,635 -6,432, -838 0.011
Age at first birth 438 381, 495 <0.001 364 285, 443 <0.001
Race/ethnicity
    Hispanic — — — —
    Black -772 -1,714, 171 0.11 -759 -1,701, 183 0.11
    Non-Black, Non-Hispanic 7,559 6,676, 8,442 <0.001 7,550 6,668, 8,433 <0.001
Sex/age interaction
    Female * Age at first birth 149 39, 260 0.008
1 CI = Confidence Interval

Exercises

  1. Download the script with some examples and save in your in-class project directory.

  2. Run the examples.

3-6. You’re on your own again!

Extra time? Start a table using the data you downloaded for your final project! Make sure you switch to that R project!

15:00