Public Health Data Tabulation with {gtsummary}

Abhijit P Pakhare

Introduction

Acknowledgements

drawing

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License (CC BY-SA4.0).

Creators of original presentation

Daniel D. Sjoberg

danieldsjoberg.com

@statistishdan

linkedin.com/in/ddsjoberg

github.com/ddsjoberg

Shannon Pileggi

pipinghotdata.com

@pipinghotdata

linkedin.com/in/shannon-m-pileggi

github.com/shannonpileggi

Karissa Whiting

karissawhiting.com

@karissawhiting

linkedin.com/in/karissa-whiting-48877a52

github.com/karissawhiting

Checklist


Install recent R release

Current version 4.2.3

Install RStudio

I am on version 2022.07.1+554

Install packages

install.packages(c("gtsummary", "tidyverse"))

Ensure you can knit Rmarkdown files

{gtsummary} overview

  • Create tabular summaries with sensible defaults but highly customizable
  • Types of summaries:
    • “Table 1”-types
    • Cross-tabulation
    • Regression models
    • Survival data
    • Survey data
    • Custom tables
  • Report statistics from {gtsummary} tables inline in R Markdown
  • Stack and/or merge any table type
  • Use themes to standardize across tables
  • Choose from different print engines

Example Dataset

  • The “steps.xlsx” data set is available for downloading under Dataset Tab

  • This is a randomly sampled subset of WHO STEPS Survey conducted for NCD risk factor surveillance

  • It consists of selected variables from all three steps of the survey.

library(gtsummary)
library(tidyverse)
library(rio)
library(here)
df <- import(here("datasets", "steps.xlsx"))
head(df) |> gt::gt()
Gender Age Age_Group Education SBP DBP BMI LDL Smoking Physical_Inactivity Raised_BP District WSTEP3
Women 60 59-69 Illiterate 116.6667 77.00000 16.10928 49.2 No No No Balaghat 8398.779
Men 67 59-69 Higher Secondary + 166.6667 97.33333 25.29558 74.6 No No Yes Satna 6269.137
Men 18 18-30 Higher Secondary + 127.6667 79.33333 22.17560 53.8 No No No Ratlam 30503.785
Women 67 59-69 Illiterate 106.3333 67.00000 17.28791 42.0 No Yes No Dindori 8898.991
Men 38 30-44 Secondary 114.0000 74.00000 18.14509 28.0 Yes No No Damoh 16746.247
Women 58 44-59 Illiterate 158.3333 87.33333 26.18594 95.6 No No Yes Ratlam 15545.019

Example Dataset

This presentation will use a variables from the dataset to demonstrate different tables.

Variable
Gender
Age
Age_Group
Education
SBP
DBP
BMI
LDL
Smoking
Physical_Inactivity
Raised_BP
District
WSTEP3 
sm_trial <-
  trial |> 
  select(trt, age, grade, response)

tbl_summary()

Basic tbl_summary()

df |> 
  tbl_summary()
Characteristic N = 3,0001
Gender
    Men 1,039 (35%)
    Women 1,961 (65%)
Age 40 (30, 51)
Age_Group
    18-30 691 (23%)
    30-44 1,097 (37%)
    44-59 799 (27%)
    59-69 413 (14%)
Education
    Higher Secondary + 832 (28%)
    Illiterate 1,162 (39%)
    Primary 493 (16%)
    Secondary 513 (17%)
SBP 125 (114, 138)
    Unknown 26
DBP 78 (71, 85)
    Unknown 26
BMI 20.2 (18.0, 23.3)
    Unknown 78
LDL 57 (42, 75)
Smoking 275 (9.2%)
Physical_Inactivity 594 (20%)
Raised_BP 825 (28%)
District
    Balaghat 357 (12%)
    Bhopal 438 (15%)
    Damoh 218 (7.3%)
    Dhar 429 (14%)
    Dindori 141 (4.7%)
    Hoshangabad 235 (7.8%)
    Ratlam 281 (9.4%)
    Satna 452 (15%)
    Sheopur 115 (3.8%)
    Shivpuri 334 (11%)
WSTEP3 16,031 (14,218, 17,604)
1 n (%); Median (IQR)

  • Four types of summaries: continuous, continuous2, categorical, and dichotomous

  • Statistics are median (IQR) for continuous, n (%) for categorical/dichotomous

  • Variables coded 0/1, TRUE/FALSE, Yes/No treated as dichotomous

  • Lists NA values under “Unknown”

  • Label attributes are printed automatically

Customize tbl_summary() output

tbl_summary(
  df,
  by = Raised_BP,
  include = c(Gender, Age_Group, BMI)
)
Characteristic No, N = 2,1751 Yes, N = 8251
Gender
    Men 744 (34%) 295 (36%)
    Women 1,431 (66%) 530 (64%)
Age_Group
    18-30 636 (29%) 55 (6.7%)
    30-44 868 (40%) 229 (28%)
    44-59 495 (23%) 304 (37%)
    59-69 176 (8.1%) 237 (29%)
BMI 19.7 (17.7, 22.4) 21.8 (18.9, 25.3)
    Unknown 70 8
1 n (%); Median (IQR)
  • by: specify a column variable for cross-tabulation

Customize tbl_summary() output

tbl_summary(
  df,
  by = Raised_BP,
  include = c(Gender, Age_Group, BMI),
  type = BMI ~ "continuous2",
)
Characteristic No, N = 2,1751 Yes, N = 8251
Gender
    Men 744 (34%) 295 (36%)
    Women 1,431 (66%) 530 (64%)
Age_Group
    18-30 636 (29%) 55 (6.7%)
    30-44 868 (40%) 229 (28%)
    44-59 495 (23%) 304 (37%)
    59-69 176 (8.1%) 237 (29%)
BMI
    Median (IQR) 19.7 (17.7, 22.4) 21.8 (18.9, 25.3)
    Unknown 70 8
1 n (%)

  • by: specify a column variable for cross-tabulation

  • type: specify the summary type

Customize tbl_summary() output

tbl_summary(
  df,
  by = Raised_BP,
  include = c(Gender, Age_Group, BMI),
  type = BMI ~ "continuous2",
  statistic = 
    list(
      BMI ~ c("{mean} ({sd})", 
              "{min}, {max}"), 
      Gender ~ "{n} / {N} ({p}%)"
    ),
)
Characteristic No, N = 2,1751 Yes, N = 8251
Gender
    Men 744 / 2,175 (34%) 295 / 825 (36%)
    Women 1,431 / 2,175 (66%) 530 / 825 (64%)
Age_Group
    18-30 636 (29%) 55 (6.7%)
    30-44 868 (40%) 229 (28%)
    44-59 495 (23%) 304 (37%)
    59-69 176 (8.1%) 237 (29%)
BMI
    Mean (SD) 20.5 (3.9) 22.5 (4.8)
    Range 11.9, 47.8 13.5, 47.8
    Unknown 70 8
1 n / N (%); n (%)

  • by: specify a column variable for cross-tabulation

  • type: specify the summary type

  • statistic: customize the reported statistics

Customize tbl_summary() output

tbl_summary(
  df,
  by = Raised_BP,
  type = BMI ~ "continuous2",
  include = c(Gender, Age_Group, BMI),
  statistic = 
    list(
      BMI ~ c("{mean} ({sd})", 
              "{min}, {max}"), 
      Gender ~ "{n} / {N} ({p}%)"
    ),
  label = 
    Age_Group ~ "Age Group",
)
Characteristic No, N = 2,1751 Yes, N = 8251
Gender
    Men 744 / 2,175 (34%) 295 / 825 (36%)
    Women 1,431 / 2,175 (66%) 530 / 825 (64%)
Age Group
    18-30 636 (29%) 55 (6.7%)
    30-44 868 (40%) 229 (28%)
    44-59 495 (23%) 304 (37%)
    59-69 176 (8.1%) 237 (29%)
BMI
    Mean (SD) 20.5 (3.9) 22.5 (4.8)
    Range 11.9, 47.8 13.5, 47.8
    Unknown 70 8
1 n / N (%); n (%)

  • by: specify a column variable for cross-tabulation

  • type: specify the summary type

  • statistic: customize the reported statistics

  • label: change or customize variable labels

Customize tbl_summary() output

tbl_summary(
  df,
  by = Raised_BP,
  type = BMI ~ "continuous2",
  include = c(Gender, Age_Group, BMI),
  statistic = 
    list(
      BMI ~ c("{mean} ({sd})", 
              "{min}, {max}"), 
      Gender ~ "{n} / {N} ({p}%)"
    ),
  label = 
    Age_Group ~ "Age Group",
  digits = list(all_continuous()~1,
                all_categorical()~ c(0,1))
)
Characteristic No, N = 2,1751 Yes, N = 8251
Gender
    Men 744 / 2,175.0 (34%) 295 / 825.0 (36%)
    Women 1,431 / 2,175.0 (66%) 530 / 825.0 (64%)
Age Group
    18-30 636 (29.2%) 55 (6.7%)
    30-44 868 (39.9%) 229 (27.8%)
    44-59 495 (22.8%) 304 (36.8%)
    59-69 176 (8.1%) 237 (28.7%)
BMI
    Mean (SD) 20.5 (3.9) 22.5 (4.8)
    Range 11.9, 47.8 13.5, 47.8
    Unknown 70 8
1 n / N (%); n (%)

  • by: specify a column variable for cross-tabulation

  • type: specify the summary type

  • statistic: customize the reported statistics

  • label: change or customize variable labels

  • digits: specify the number of decimal places for rounding

{gtsummary} + formulas

Named list are OK too! label = list(age = "Patient Age")

Add-on functions in {gtsummary}

tbl_summary() objects can also be updated using related functions.

  • add_*() add additional column of statistics or information, e.g. p-values, q-values, overall statistics, treatment differences, N obs., and more

  • modify_*() modify table headers, spanning headers, footnotes, and more

  • bold_*()/italicize_*() style labels, variable levels, significant p-values

Update tbl_summary() with add_*()

df |>
  tbl_summary(
    by = Raised_BP
  ) |> 
  add_p()
Characteristic No, N = 2,1751 Yes, N = 8251 p-value2
Gender 0.4
    Men 744 (34%) 295 (36%)
    Women 1,431 (66%) 530 (64%)
Age 35 (28, 46) 50 (40, 60) <0.001
Age_Group <0.001
    18-30 636 (29%) 55 (6.7%)
    30-44 868 (40%) 229 (28%)
    44-59 495 (23%) 304 (37%)
    59-69 176 (8.1%) 237 (29%)
Education <0.001
    Higher Secondary + 610 (28%) 222 (27%)
    Illiterate 790 (36%) 372 (45%)
    Primary 372 (17%) 121 (15%)
    Secondary 403 (19%) 110 (13%)
SBP 119 (111, 127) 149 (142, 161) <0.001
    Unknown 26 0
DBP 75 (69, 80) 91 (83, 97) <0.001
    Unknown 26 0
BMI 19.7 (17.7, 22.4) 21.8 (18.9, 25.3) <0.001
    Unknown 70 8
LDL 55 (42, 72) 61 (43, 82) <0.001
Smoking 187 (8.6%) 88 (11%) 0.080
Physical_Inactivity 426 (20%) 168 (20%) 0.6
District <0.001
    Balaghat 236 (11%) 121 (15%)
    Bhopal 303 (14%) 135 (16%)
    Damoh 168 (7.7%) 50 (6.1%)
    Dhar 292 (13%) 137 (17%)
    Dindori 103 (4.7%) 38 (4.6%)
    Hoshangabad 182 (8.4%) 53 (6.4%)
    Ratlam 218 (10%) 63 (7.6%)
    Satna 356 (16%) 96 (12%)
    Sheopur 82 (3.8%) 33 (4.0%)
    Shivpuri 235 (11%) 99 (12%)
WSTEP3 16,469 (14,982, 22,945) 14,644 (9,505, 16,230) <0.001
1 n (%); Median (IQR)
2 Pearson’s Chi-squared test; Wilcoxon rank sum test
  • add_p(): adds a column of p-values

Update tbl_summary() with add_*()

df |>
  tbl_summary(
    by = Raised_BP,
    missing = "no"
  ) |> 
  add_overall()
Characteristic Overall, N = 3,0001 No, N = 2,1751 Yes, N = 8251
Gender
    Men 1,039 (35%) 744 (34%) 295 (36%)
    Women 1,961 (65%) 1,431 (66%) 530 (64%)
Age 40 (30, 51) 35 (28, 46) 50 (40, 60)
Age_Group
    18-30 691 (23%) 636 (29%) 55 (6.7%)
    30-44 1,097 (37%) 868 (40%) 229 (28%)
    44-59 799 (27%) 495 (23%) 304 (37%)
    59-69 413 (14%) 176 (8.1%) 237 (29%)
Education
    Higher Secondary + 832 (28%) 610 (28%) 222 (27%)
    Illiterate 1,162 (39%) 790 (36%) 372 (45%)
    Primary 493 (16%) 372 (17%) 121 (15%)
    Secondary 513 (17%) 403 (19%) 110 (13%)
SBP 125 (114, 138) 119 (111, 127) 149 (142, 161)
DBP 78 (71, 85) 75 (69, 80) 91 (83, 97)
BMI 20.2 (18.0, 23.3) 19.7 (17.7, 22.4) 21.8 (18.9, 25.3)
LDL 57 (42, 75) 55 (42, 72) 61 (43, 82)
Smoking 275 (9.2%) 187 (8.6%) 88 (11%)
Physical_Inactivity 594 (20%) 426 (20%) 168 (20%)
District
    Balaghat 357 (12%) 236 (11%) 121 (15%)
    Bhopal 438 (15%) 303 (14%) 135 (16%)
    Damoh 218 (7.3%) 168 (7.7%) 50 (6.1%)
    Dhar 429 (14%) 292 (13%) 137 (17%)
    Dindori 141 (4.7%) 103 (4.7%) 38 (4.6%)
    Hoshangabad 235 (7.8%) 182 (8.4%) 53 (6.4%)
    Ratlam 281 (9.4%) 218 (10%) 63 (7.6%)
    Satna 452 (15%) 356 (16%) 96 (12%)
    Sheopur 115 (3.8%) 82 (3.8%) 33 (4.0%)
    Shivpuri 334 (11%) 235 (11%) 99 (12%)
WSTEP3 16,031 (14,218, 17,604) 16,469 (14,982, 22,945) 14,644 (9,505, 16,230)
1 n (%); Median (IQR)
  • add_overall(): adds a column of overall statistics

Update tbl_summary() with add_*()

df |>
  tbl_summary(
    by = Raised_BP,
    missing = "no"
  ) |> 
  add_overall() |> 
  add_n()
Characteristic N Overall, N = 3,0001 No, N = 2,1751 Yes, N = 8251
Gender 3,000
    Men 1,039 (35%) 744 (34%) 295 (36%)
    Women 1,961 (65%) 1,431 (66%) 530 (64%)
Age 3,000 40 (30, 51) 35 (28, 46) 50 (40, 60)
Age_Group 3,000
    18-30 691 (23%) 636 (29%) 55 (6.7%)
    30-44 1,097 (37%) 868 (40%) 229 (28%)
    44-59 799 (27%) 495 (23%) 304 (37%)
    59-69 413 (14%) 176 (8.1%) 237 (29%)
Education 3,000
    Higher Secondary + 832 (28%) 610 (28%) 222 (27%)
    Illiterate 1,162 (39%) 790 (36%) 372 (45%)
    Primary 493 (16%) 372 (17%) 121 (15%)
    Secondary 513 (17%) 403 (19%) 110 (13%)
SBP 2,974 125 (114, 138) 119 (111, 127) 149 (142, 161)
DBP 2,974 78 (71, 85) 75 (69, 80) 91 (83, 97)
BMI 2,922 20.2 (18.0, 23.3) 19.7 (17.7, 22.4) 21.8 (18.9, 25.3)
LDL 3,000 57 (42, 75) 55 (42, 72) 61 (43, 82)
Smoking 3,000 275 (9.2%) 187 (8.6%) 88 (11%)
Physical_Inactivity 3,000 594 (20%) 426 (20%) 168 (20%)
District 3,000
    Balaghat 357 (12%) 236 (11%) 121 (15%)
    Bhopal 438 (15%) 303 (14%) 135 (16%)
    Damoh 218 (7.3%) 168 (7.7%) 50 (6.1%)
    Dhar 429 (14%) 292 (13%) 137 (17%)
    Dindori 141 (4.7%) 103 (4.7%) 38 (4.6%)
    Hoshangabad 235 (7.8%) 182 (8.4%) 53 (6.4%)
    Ratlam 281 (9.4%) 218 (10%) 63 (7.6%)
    Satna 452 (15%) 356 (16%) 96 (12%)
    Sheopur 115 (3.8%) 82 (3.8%) 33 (4.0%)
    Shivpuri 334 (11%) 235 (11%) 99 (12%)
WSTEP3 3,000 16,031 (14,218, 17,604) 16,469 (14,982, 22,945) 14,644 (9,505, 16,230)
1 n (%); Median (IQR)
  • add_overall(): adds a column of overall statistics
  • add_n(): adds a column with the sample size

Update tbl_summary() with add_*()

df |>
  tbl_summary(
    by = Raised_BP,
    missing = "no"
  ) |> 
  add_overall() |> 
  add_n() |> 
  add_stat_label(
    label = all_categorical() ~ "No. (%)"
  )
Characteristic N Overall, N = 3,000 No, N = 2,175 Yes, N = 825
Gender, No. (%) 3,000
    Men 1,039 (35%) 744 (34%) 295 (36%)
    Women 1,961 (65%) 1,431 (66%) 530 (64%)
Age, Median (IQR) 3,000 40 (30, 51) 35 (28, 46) 50 (40, 60)
Age_Group, No. (%) 3,000
    18-30 691 (23%) 636 (29%) 55 (6.7%)
    30-44 1,097 (37%) 868 (40%) 229 (28%)
    44-59 799 (27%) 495 (23%) 304 (37%)
    59-69 413 (14%) 176 (8.1%) 237 (29%)
Education, No. (%) 3,000
    Higher Secondary + 832 (28%) 610 (28%) 222 (27%)
    Illiterate 1,162 (39%) 790 (36%) 372 (45%)
    Primary 493 (16%) 372 (17%) 121 (15%)
    Secondary 513 (17%) 403 (19%) 110 (13%)
SBP, Median (IQR) 2,974 125 (114, 138) 119 (111, 127) 149 (142, 161)
DBP, Median (IQR) 2,974 78 (71, 85) 75 (69, 80) 91 (83, 97)
BMI, Median (IQR) 2,922 20.2 (18.0, 23.3) 19.7 (17.7, 22.4) 21.8 (18.9, 25.3)
LDL, Median (IQR) 3,000 57 (42, 75) 55 (42, 72) 61 (43, 82)
Smoking, No. (%) 3,000 275 (9.2%) 187 (8.6%) 88 (11%)
Physical_Inactivity, No. (%) 3,000 594 (20%) 426 (20%) 168 (20%)
District, No. (%) 3,000
    Balaghat 357 (12%) 236 (11%) 121 (15%)
    Bhopal 438 (15%) 303 (14%) 135 (16%)
    Damoh 218 (7.3%) 168 (7.7%) 50 (6.1%)
    Dhar 429 (14%) 292 (13%) 137 (17%)
    Dindori 141 (4.7%) 103 (4.7%) 38 (4.6%)
    Hoshangabad 235 (7.8%) 182 (8.4%) 53 (6.4%)
    Ratlam 281 (9.4%) 218 (10%) 63 (7.6%)
    Satna 452 (15%) 356 (16%) 96 (12%)
    Sheopur 115 (3.8%) 82 (3.8%) 33 (4.0%)
    Shivpuri 334 (11%) 235 (11%) 99 (12%)
WSTEP3, Median (IQR) 3,000 16,031 (14,218, 17,604) 16,469 (14,982, 22,945) 14,644 (9,505, 16,230)
  • add_overall(): adds a column of overall statistics
  • add_n(): adds a column with the sample size
  • add_stat_label(): adds a description of the reported statistic

Update with bold_*()/italicize_*()

df |>
  tbl_summary(
    by = Raised_BP
  ) |>
  add_p() |> 
  bold_labels() |> 
  italicize_levels() |> 
  bold_p(t = 0.8)
Characteristic No, N = 2,1751 Yes, N = 8251 p-value2
Gender 0.4
    Men 744 (34%) 295 (36%)
    Women 1,431 (66%) 530 (64%)
Age 35 (28, 46) 50 (40, 60) <0.001
Age_Group <0.001
    18-30 636 (29%) 55 (6.7%)
    30-44 868 (40%) 229 (28%)
    44-59 495 (23%) 304 (37%)
    59-69 176 (8.1%) 237 (29%)
Education <0.001
    Higher Secondary + 610 (28%) 222 (27%)
    Illiterate 790 (36%) 372 (45%)
    Primary 372 (17%) 121 (15%)
    Secondary 403 (19%) 110 (13%)
SBP 119 (111, 127) 149 (142, 161) <0.001
    Unknown 26 0
DBP 75 (69, 80) 91 (83, 97) <0.001
    Unknown 26 0
BMI 19.7 (17.7, 22.4) 21.8 (18.9, 25.3) <0.001
    Unknown 70 8
LDL 55 (42, 72) 61 (43, 82) <0.001
Smoking 187 (8.6%) 88 (11%) 0.080
Physical_Inactivity 426 (20%) 168 (20%) 0.6
District <0.001
    Balaghat 236 (11%) 121 (15%)
    Bhopal 303 (14%) 135 (16%)
    Damoh 168 (7.7%) 50 (6.1%)
    Dhar 292 (13%) 137 (17%)
    Dindori 103 (4.7%) 38 (4.6%)
    Hoshangabad 182 (8.4%) 53 (6.4%)
    Ratlam 218 (10%) 63 (7.6%)
    Satna 356 (16%) 96 (12%)
    Sheopur 82 (3.8%) 33 (4.0%)
    Shivpuri 235 (11%) 99 (12%)
WSTEP3 16,469 (14,982, 22,945) 14,644 (9,505, 16,230) <0.001
1 n (%); Median (IQR)
2 Pearson’s Chi-squared test; Wilcoxon rank sum test
  • bold_labels(): bold the variable labels
  • italicize_levels(): italicize the variable levels
  • bold_p(): bold p-values according a specified threshold

Update tbl_summary() with modify_*()

tbl <-
  df |> 
  tbl_summary(by = Raised_BP, 
              missing = "no") |>
  modify_header(
      stat_1 ~ "**No raised BP**",
      stat_2 ~ "**Raised BP**"
  ) |> 
  modify_spanning_header(
    all_stat_cols() ~ "**BP Status**") |> 
  modify_footnote(
    all_stat_cols() ~ 
      paste("median (IQR) for continuous;",
            "n (%) for categorical")
  )
tbl
Characteristic BP Status
No raised BP1 Raised BP1
Gender
    Men 744 (34%) 295 (36%)
    Women 1,431 (66%) 530 (64%)
Age 35 (28, 46) 50 (40, 60)
Age_Group
    18-30 636 (29%) 55 (6.7%)
    30-44 868 (40%) 229 (28%)
    44-59 495 (23%) 304 (37%)
    59-69 176 (8.1%) 237 (29%)
Education
    Higher Secondary + 610 (28%) 222 (27%)
    Illiterate 790 (36%) 372 (45%)
    Primary 372 (17%) 121 (15%)
    Secondary 403 (19%) 110 (13%)
SBP 119 (111, 127) 149 (142, 161)
DBP 75 (69, 80) 91 (83, 97)
BMI 19.7 (17.7, 22.4) 21.8 (18.9, 25.3)
LDL 55 (42, 72) 61 (43, 82)
Smoking 187 (8.6%) 88 (11%)
Physical_Inactivity 426 (20%) 168 (20%)
District
    Balaghat 236 (11%) 121 (15%)
    Bhopal 303 (14%) 135 (16%)
    Damoh 168 (7.7%) 50 (6.1%)
    Dhar 292 (13%) 137 (17%)
    Dindori 103 (4.7%) 38 (4.6%)
    Hoshangabad 182 (8.4%) 53 (6.4%)
    Ratlam 218 (10%) 63 (7.6%)
    Satna 356 (16%) 96 (12%)
    Sheopur 82 (3.8%) 33 (4.0%)
    Shivpuri 235 (11%) 99 (12%)
WSTEP3 16,469 (14,982, 22,945) 14,644 (9,505, 16,230)
1 median (IQR) for continuous; n (%) for categorical
  • Use show_header_names() to see the internal header names available for use in modify_header()

Column names

show_header_names(tbl)
Column Name Column Header
label Characteristic
stat_1 No raised BP
stat_2 Raised BP



all_stat_cols() selects columns "stat_1" and "stat_2"

Update tbl_summary() with add_*()

df |>
  select(Raised_BP, SBP, BMI, LDL, Smoking) |>
  tbl_summary(
    by = Raised_BP,
    statistic = list(c(SBP, BMI, LDL) ~ "{mean} ({sd})",
                     Smoking ~ "{p}%"),
    missing = "no"
  ) |> 
  add_difference()
Characteristic No, N = 2,1751 Yes, N = 8251 Difference2 95% CI2,3 p-value2
SBP 119 (11) 153 (21) -34 -35, -32 <0.001
BMI 20.5 (3.9) 22.5 (4.8) -2.0 -2.4, -1.7 <0.001
LDL 58 (28) 64 (31) -5.4 -7.9, -3.0 <0.001
Smoking 8.6% 11% -2.1% -4.6%, 0.43% 0.092
1 Mean (SD); %
2 Welch Two Sample t-test; Two sample test for equality of proportions
3 CI = Confidence Interval
  • add_difference(): mean and rate differences between two groups. Can also be adjusted differences

Update tbl_summary() with add_*()

df |>
  tbl_summary(
    by = Raised_BP,
    missing = "no"
  ) |> 
  add_stat(...)

  • Customize statistics presented with add_stat()

  • Added statistics can be placed on the label or the level rows

  • Added statistics may be a single column or multiple

Add-on functions in {gtsummary}

And many more!

See the documentation at http://www.danieldsjoberg.com/gtsummary/reference/index.html

And a detailed tbl_summary() vignette at http://www.danieldsjoberg.com/gtsummary/articles/tbl_summary.html

Cross-tabulation with tbl_cross()

tbl_cross() is a wrapper for tbl_summary() for n x m tables

df |>
  tbl_cross(
    row = District, 
    col = Raised_BP,
    percent = "row",
    margin = "row"
  ) |>
  add_p(source_note = TRUE) |>
  bold_labels()
Raised_BP
No Yes
District
    Balaghat 236 (66%) 121 (34%)
    Bhopal 303 (69%) 135 (31%)
    Damoh 168 (77%) 50 (23%)
    Dhar 292 (68%) 137 (32%)
    Dindori 103 (73%) 38 (27%)
    Hoshangabad 182 (77%) 53 (23%)
    Ratlam 218 (78%) 63 (22%)
    Satna 356 (79%) 96 (21%)
    Sheopur 82 (71%) 33 (29%)
    Shivpuri 235 (70%) 99 (30%)
Total 2,175 (72%) 825 (28%)
Pearson’s Chi-squared test, p<0.001

Continuous Summaries with tbl_continuous()

tbl_continuous() summarizes a continuous variable by 1, 2, or more categorical variables

df |>
  tbl_continuous(
    variable = SBP,
    by = Raised_BP,
    include = c(Age_Group, Gender)
  )
Characteristic No, N = 2,1751 Yes, N = 8251
Age_Group
    18-30 116 (109, 125) 141 (134, 144)
    30-44 119 (112, 127) 145 (140, 153)
    44-59 121 (115, 129) 150 (142, 161)
    59-69 125 (116, 132) 157 (147, 175)
Gender
    Men 122 (113, 129) 148 (142, 160)
    Women 118 (111, 126) 149 (142, 161)
1 SBP: Median (IQR)

Survey data with tbl_svysummary()

survey::svydesign(
  ids = ~1, 
  data = df, 
  weights = ~WSTEP3
) |>
  tbl_svysummary(
    by = Raised_BP,
    include = c(Age_Group, Gender, District),
    percent = "row"
  ) |>
  add_p() |>
  modify_spanning_header(
    all_stat_cols() ~ "**Raised BP**")
Characteristic Raised BP p-value2
No, N = 39,432,9561 Yes, N = 11,491,2761
Age_Group <0.001
    18-30 16,646,634 (92%) 1,445,876 (8.0%)
    30-44 14,073,391 (79%) 3,635,210 (21%)
    44-59 7,322,516 (62%) 4,524,442 (38%)
    59-69 1,390,415 (42%) 1,885,747 (58%)
Gender 0.9
    Men 13,472,904 (78%) 3,889,460 (22%)
    Women 25,960,052 (77%) 7,601,815 (23%)
District 0.001
    Balaghat 3,785,643 (71%) 1,533,116 (29%)
    Bhopal 5,737,817 (75%) 1,935,086 (25%)
    Damoh 3,123,656 (81%) 717,595 (19%)
    Dhar 4,812,133 (72%) 1,829,268 (28%)
    Dindori 1,843,049 (78%) 505,601 (22%)
    Hoshangabad 3,468,386 (82%) 750,387 (18%)
    Ratlam 4,065,569 (81%) 931,610 (19%)
    Satna 6,266,329 (82%) 1,414,737 (18%)
    Sheopur 1,814,049 (77%) 536,462 (23%)
    Shivpuri 4,516,325 (77%) 1,337,415 (23%)
1 n (%)
2 chi-squared test with Rao & Scott’s second-order correction

Survey data with tbl_svysummary()

survey::svydesign(
  ids = ~1, 
  data = df, 
  weights = ~WSTEP3
) |>
  tbl_svysummary(
    by = Raised_BP,
    include = c(Age_Group, Gender, District),
    percent = "row",
    digits = list(all_categorical() ~ 1),
    statistic = list(all_categorical() ~ "{p}")
  ) |>
  add_p() |>
  modify_spanning_header(
    all_stat_cols() ~ "**Raised BP**")
Characteristic Raised BP p-value2
No, N = 39,432,9561 Yes, N = 11,491,2761
Age_Group <0.001
    18-30 92.0 8.0
    30-44 79.5 20.5
    44-59 61.8 38.2
    59-69 42.4 57.6
Gender 0.9
    Men 77.6 22.4
    Women 77.3 22.7
District 0.001
    Balaghat 71.2 28.8
    Bhopal 74.8 25.2
    Damoh 81.3 18.7
    Dhar 72.5 27.5
    Dindori 78.5 21.5
    Hoshangabad 82.2 17.8
    Ratlam 81.4 18.6
    Satna 81.6 18.4
    Sheopur 77.2 22.8
    Shivpuri 77.2 22.8
1 %
2 chi-squared test with Rao & Scott’s second-order correction

tbl_regression()

Traditional model summary()

df <- df %>% 
  mutate(Raised_BP = if_else(Raised_BP=="Yes", 1,0))
m1 <- 
  glm(
    Raised_BP ~ Age + Smoking + BMI,
    data = df,
    family = binomial(link = "logit")
  )
summary(m1)

Call: glm(formula = Raised_BP ~ Age + Smoking + BMI, family = binomial(link = “logit”), data = df)

Deviance Residuals: Min 1Q Median 3Q Max
-2.4395 -0.7723 -0.4914 0.9125 2.4430

Coefficients: Estimate Std. Error z value Pr(>|z|)
(Intercept) -6.154933 0.298385 -20.628 <2e-16 Age 0.067720 0.003669 18.460 <2e-16 SmokingYes -0.008596 0.148722 -0.058 0.954
BMI 0.104944 0.010593 9.907 <2e-16 ***

Signif. codes: 0 ‘’ 0.001 ’’ 0.01 ’’ 0.05 ‘.’ 0.1 ’ ’ 1

(Dispersion parameter for binomial family taken to be 1)

Null deviance: 3463.0  on 2921  degrees of freedom

Residual deviance: 2930.8 on 2918 degrees of freedom (78 observations deleted due to missingness) AIC: 2938.8

Number of Fisher Scoring iterations: 4

Looks messy and it’s not easy to digest

Basic tbl_regression()

::: {.cell output-location=‘column’}

tbl_regression(m1)
Characteristic log(OR)1 95% CI1 p-value
Age 0.07 0.06, 0.07 <0.001
Smoking
    No — —
    Yes -0.01 -0.30, 0.28
0.9
BMI 0.10 0.08, 0.13 <0.001
1 OR = Odds Ratio, CI = Confidence Interval

:::

  • Displays p-values for covariates

  • Shows reference levels for categorical variables

  • Model type recognized as logistic regression with odds ratio appearing in header

Customize tbl_regression() output

::: {.cell output-location=‘column’}

tbl_regression(
  m1,
  exponentiate = TRUE
) |> 
  add_global_p() |>
  add_glance_table(
    include = c(nobs,
                logLik,
                AIC,
                BIC)
  )
Characteristic OR1 95% CI1 p-value
Age 1.07 1.06, 1.08 <0.001
Smoking
0.9
    No — —
    Yes 0.99 0.74, 1.32
BMI 1.11 1.09, 1.13 <0.001
No. Obs. 2,922
Log-likelihood -1,465
AIC 2,939
BIC 2,963
1 OR = Odds Ratio, CI = Confidence Interval

:::

  • Display odds ratio estimates and confidence intervals

  • Add global p-values

  • Add various model statistics

Supported models in tbl_regression()

  • biglm::bigglm()
  • biglmm::bigglm()
  • brms::brm()
  • cmprsk::crr()
  • fixest::feglm()
  • fixest::femlm()
  • fixest::feNmlm()
  • fixest::feols()
  • gam::gam()
  • geepack::geeglm()
  • glmmTMB::glmmTMB()
  • lavaan::lavaan()
  • lfe::felm()
  • lme4::glmer.nb()
  • lme4::glmer()
  • lme4::lmer()
  • logitr::logitr()
  • MASS::glm.nb()
  • MASS::polr()
  • mgcv::gam()
  • mice::mira
  • multgee::nomLORgee()
  • multgee::ordLORgee()
  • nnet::multinom()
  • ordinal::clm()
  • ordinal::clmm()
  • parsnip::model_fit
  • plm::plm()
  • rstanarm::stan_glm()
  • stats::aov()
  • stats::glm()
  • stats::lm()
  • stats::nls()
  • survey::svycoxph()
  • survey::svyglm()
  • survey::svyolr()
  • survival::clogit()
  • survival::coxph()
  • survival::survreg()
  • tidycmprsk::crr()
  • VGAM::vglm()

Custom tidiers can be written and passed to tbl_regression() using the tidy_fun= argument.

Univariate models with tbl_uvregression()

tbl_uvreg <- 
  df |> 
  select(-District, -WSTEP3, - SBP, -DBP) |> 
  tbl_uvregression(
    method = glm,
    y = Raised_BP,
    method.args = 
      list(family = binomial),
    exponentiate = TRUE
  )
tbl_uvreg
Characteristic N OR1 95% CI1 p-value
Gender 3,000
    Men — —
    Women 0.93 0.79, 1.11 0.4
Age 3,000 1.07 1.06, 1.08 <0.001
Age_Group 3,000
    18-30 — —
    30-44 3.05 2.25, 4.20 <0.001
    44-59 7.10 5.25, 9.77 <0.001
    59-69 15.6 11.2, 22.0 <0.001
Education 3,000
    Higher Secondary + — —
    Illiterate 1.29 1.06, 1.58 0.010
    Primary 0.89 0.69, 1.15 0.4
    Secondary 0.75 0.58, 0.97 0.031
BMI 2,922 1.12 1.09, 1.14 <0.001
LDL 3,000 1.01 1.00, 1.01 <0.001
Smoking 3,000
    No — —
    Yes 1.27 0.97, 1.65 0.080
Physical_Inactivity 3,000
    No — —
    Yes 1.05 0.86, 1.28 0.6
1 OR = Odds Ratio, CI = Confidence Interval

  • Specify model method, method.args, and the response variable

  • Arguments and helper functions like exponentiate, bold_*(), add_global_p() can also be used with tbl_uvregression()

tbl_merge()/tbl_stack()

tbl_merge() for side-by-side tables

A univariable table:

tbl_uvreg <- 
  df |> 
  select(-District, -WSTEP3, - SBP, -DBP) |> 
  tbl_uvregression(
    method = glm,
    y = Raised_BP,
    method.args = 
      list(family = binomial),
    exponentiate = TRUE
  )
tbl_uvreg
Characteristic N OR1 95% CI1 p-value
Gender 3,000
    Men — —
    Women 0.93 0.79, 1.11 0.4
Age 3,000 1.07 1.06, 1.08 <0.001
Age_Group 3,000
    18-30 — —
    30-44 3.05 2.25, 4.20 <0.001
    44-59 7.10 5.25, 9.77 <0.001
    59-69 15.6 11.2, 22.0 <0.001
Education 3,000
    Higher Secondary + — —
    Illiterate 1.29 1.06, 1.58 0.010
    Primary 0.89 0.69, 1.15 0.4
    Secondary 0.75 0.58, 0.97 0.031
BMI 2,922 1.12 1.09, 1.14 <0.001
LDL 3,000 1.01 1.00, 1.01 <0.001
Smoking 3,000
    No — —
    Yes 1.27 0.97, 1.65 0.080
Physical_Inactivity 3,000
    No — —
    Yes 1.05 0.86, 1.28 0.6
1 OR = Odds Ratio, CI = Confidence Interval

A multivariable table:

tbl_mvreg <- 
  glm(
    Raised_BP ~ Age_Group + Smoking + BMI + LDL,
    data = df,
    family = binomial(link = "logit")
  ) |> 
  tbl_regression(
    exponentiate = TRUE
  ) |> 
  add_global_p() 
tbl_mvreg
Characteristic OR1 95% CI1 p-value
Age_Group <0.001
    18-30 — —
    30-44 2.47 1.79, 3.44
    44-59 5.82 4.23, 8.12
    59-69 14.6 10.3, 21.0
Smoking 0.8
    No — —
    Yes 1.03 0.77, 1.38
BMI 1.11 1.09, 1.13 <0.001
LDL 1.00 1.00, 1.00 0.7
1 OR = Odds Ratio, CI = Confidence Interval

tbl_merge() for side-by-side tables

tbl_merge(
  list(tbl_uvreg, tbl_mvreg),
  tab_spanner = c("**Univariable**", "**Multivariable**")
)
Characteristic Univariable Multivariable
N OR1 95% CI1 p-value OR1 95% CI1 p-value
Gender 3,000
    Men — —
    Women 0.93 0.79, 1.11 0.4
Age 3,000 1.07 1.06, 1.08 <0.001
Age_Group 3,000 <0.001
    18-30 — — — —
    30-44 3.05 2.25, 4.20 <0.001 2.47 1.79, 3.44
    44-59 7.10 5.25, 9.77 <0.001 5.82 4.23, 8.12
    59-69 15.6 11.2, 22.0 <0.001 14.6 10.3, 21.0
Education 3,000
    Higher Secondary + — —
    Illiterate 1.29 1.06, 1.58 0.010
    Primary 0.89 0.69, 1.15 0.4
    Secondary 0.75 0.58, 0.97 0.031
BMI 2,922 1.12 1.09, 1.14 <0.001 1.11 1.09, 1.13 <0.001
LDL 3,000 1.01 1.00, 1.01 <0.001 1.00 1.00, 1.00 0.7
Smoking 3,000 0.8
    No — — — —
    Yes 1.27 0.97, 1.65 0.080 1.03 0.77, 1.38
Physical_Inactivity 3,000
    No — —
    Yes 1.05 0.86, 1.28 0.6
1 OR = Odds Ratio, CI = Confidence Interval

tbl_strata() for stratified tables

df |>
  mutate(Raised_BP = if_else(Raised_BP == 1, "Raised", "Normal")) |>
  tbl_strata(
    strata = District,
    ~tbl_summary(.x, by = Raised_BP,
                 percent = "row",
                 missing = "no",
                 include = c(Age_Group, Gender),
                 digits = list(all_categorical() ~ 1),
                 statistic = list(all_categorical() ~ "{p}")) |>
      modify_header(all_stat_cols() ~ "**{level}**")
  )
Characteristic Balaghat Bhopal Damoh Dhar Dindori Hoshangabad Ratlam Satna Sheopur Shivpuri
Normal1 Raised1 Normal1 Raised1 Normal1 Raised1 Normal1 Raised1 Normal1 Raised1 Normal1 Raised1 Normal1 Raised1 Normal1 Raised1 Normal1 Raised1 Normal1 Raised1
Age_Group
    18-30 89.5 10.5 91.8 8.2 93.3 6.7 87.5 12.5 97.0 3.0 93.8 6.2 93.8 6.2 90.7 9.3 93.9 6.1 96.3 3.7
    30-44 74.2 25.8 77.5 22.5 89.8 10.2 68.0 32.0 78.7 21.3 90.2 9.8 76.8 23.2 84.8 15.2 81.8 18.2 76.5 23.5
    44-59 60.9 39.1 58.0 42.0 60.0 40.0 59.5 40.5 63.4 36.6 64.3 35.7 69.2 30.8 68.9 31.1 35.0 65.0 61.0 39.0
    59-69 37.7 62.3 26.8 73.2 44.0 56.0 34.2 65.8 40.0 60.0 46.2 53.8 63.4 36.6 60.3 39.7 44.4 55.6 33.3 66.7
Gender
    Men 67.2 32.8 69.9 30.1 87.1 12.9 63.2 36.8 66.7 33.3 78.7 21.3 73.0 27.0 79.0 21.0 71.1 28.9 62.5 37.5
    Women 65.5 34.5 69.0 31.0 69.6 30.4 71.3 28.7 76.3 23.7 76.6 23.4 81.3 18.7 78.6 21.4 71.4 28.6 74.8 25.2
1 %

In Closing

{gtsummary} website

http://www.danieldsjoberg.com/gtsummary/

Package Authors/Contributors

Daniel D. Sjoberg

Michael Curry

Joseph Larmarange

Jessica Lavery

Karissa Whiting

Emily C. Zabor

Xing Bai

Esther Drill

Jessica Flynn

Margie Hannum

Stephanie Lobaugh

Shannon Pileggi

Amy Tin

Gustavo Zapata Wainberg

Other Contributors

@ablack3, @ABorakati, @aghaynes, @ahinton-mmc, @aito123, @akarsteve, @akefley, @albertostefanelli, @alexis-catherine, @amygimma, @anaavu, @andrader, @angelgar, @arbet003, @arnmayer, @aspina7, @asshah4, @awcm0n, @barthelmes, @bcjaeger, @BeauMeche, @benediktclaus, @berg-michael, @bhattmaulik, @BioYork, @brachem-christian, @bwiernik, @bx259, @calebasaraba, @CarolineXGao, @ChongTienGoh, @Chris-M-P, @chrisleitzinger, @cjprobst, @clmawhorter, @CodieMonster, @coeus-analytics, @coreysparks, @ctlamb, @davidgohel, @davidkane9, @dax44, @dchiu911, @ddsjoberg, @DeFilippis, @denis-or, @dereksonderegger, @dieuv0, @discoleo, @djbirke, @dmenne, @ElfatihHasabo, @emilyvertosick, @ercbk, @erikvona, @eweisbrod, @feizhadj, @fh-jsnider, @ge-generation, @ghost, @gjones1219, @gorkang, @GuiMarthe, @hass91, @HichemLa, @hughjonesd, @iaingallagher, @ilyamusabirov, @IndrajeetPatil, @IsadoraBM, @j-tamad, @jalavery, @jeanmanguy, @jemus42, @jenifav, @jennybc, @JeremyPasco, @JesseRop, @jflynn264, @jjallaire, @jmbarajas, @jmbarbone, @JoanneF1229, @joelgautschi, @jojosgithub, @JonGretar, @jordan49er, @jthomasmock, @juseer, @jwilliman, @karissawhiting, @kendonB, @kmdono02, @kwakuduahc1, @lamhine, @larmarange, @leejasme, @loukesio, @lspeetluk, @ltin1214, @lucavd, @LuiNov, @maia-sh, @Marsus1972, @matthieu-faron, @mbac, @mdidish, @MelissaAssel, @michaelcurry1123, @mljaniczek, @moleps, @motocci, @msberends, @mvuorre, @myensr, @MyKo101, @oranwutang, @palantre, @Pascal-Schmidt, @pedersebastian, @perlatex, @philsf, @polc1410, @postgres-newbie, @proshano, @raphidoc, @RaviBot, @rich-iannone, @RiversPharmD, @rmgpanw, @roman2023, @ryzhu75, @sachijay, @saifelayan, @sammo3182, @sandhyapc, @sbalci, @sda030, @shannonpileggi, @shengchaohou, @ShixiangWang, @simonpcouch, @slb2240, @slobaugh, @spiralparagon, @StaffanBetner, @Stephonomon, @storopoli, @szimmer, @tamytsujimoto, @TarJae, @themichjam, @THIB20, @tibirkrajc, @tjmeyers, @tldrcharlene, @tormodb, @toshifumikuroda, @UAB-BST-680, @uakimix, @uriahf, @Valja64, @vvm02, @xkcococo, @yonicd, @yoursdearboy, @zabore, @zachariae, @zaddyzad, @zeyunlu, @zhengnow, @zlkrvsm, @zongell-star, and @Zoulf001.

Thank you

Ask on stackoverflow.com

Use the gtsummary tag

Hundreds of Qs already answered!

danieldsjoberg.com

@statistishdan

linkedin.com/in/ddsjoberg/

github.com/ddsjoberg

[Public Health Data Tabulation with {gtsummary})