ARD + Tables

Daniel D. Sjoberg and Becca Krouse

Workshop outline

Introduction to the Analysis Results Standard and ARDs with {cards}
Introduction to the {cardx} Package and ARD Extras
ARD to Tables with {gtsummary} and {tfrmt}

Tables in the {pharmaverse}

Data frame-first packages

Tables in the {pharmaverse}

ARD-first packages

Introduction to {gtsummary}

How it started

Began to address reproducible issues while working in academia
Goal was to build a package to summarize study results with code that was both simple and customizable
First release in May 2019

How it’s going

The stats
- 1,300,000+ installations from CRAN
- 1100+ GitHub stars
- 300+ contributors
- 50+ code contributors
Won the 2021 American Statistical Association (ASA) Innovation in Programming Award
Won the 2024 Posit Pharma Table Contest

{gtsummary} overview

Create tabular summaries with sensible defaults but highly customizable
Types of summaries:
- Demographic- or “Table 1”-types
- Cross-tabulation
- Regression models
- Survival data
- Nested summaries (AE tables)
- Custom tables

Stack and/or merge any table type
Use themes to standardize across tables
Choose from different print engines

{gtsummary} runs on ARDs!

Simple Summary Table

library(gtsummary)
ADSL <- cards::ADSL |> dplyr::filter(startsWith(ARM, "Xanomeline"))
ADAE <- cards::ADAE |> dplyr::filter(startsWith(TRTA, "Xanomeline"))

tbl <- ADSL |> 
  tbl_summary(
    by = ARM, 
    include = c(AGE, RACE),
    type = AGE ~ "continuous2",
    statistic = all_continuous() ~ c("{mean} ({sd})", "{min}, {max}")
  ) |> 
  add_overall()

Characteristic	Overall N = 168¹	Xanomeline High Dose N = 84¹	Xanomeline Low Dose N = 84¹
Age
Mean (SD)	75 (8)	74 (8)	76 (8)
Min, Max	51, 88	56, 88	51, 88
Race
AMERICAN INDIAN OR ALASKA NATIVE	1 (0.6%)	1 (1.2%)	0 (0.0%)
BLACK OR AFRICAN AMERICAN	15 (8.9%)	9 (10.7%)	6 (7.1%)
WHITE	152 (90.5%)	74 (88.1%)	78 (92.9%)
¹ n (%)

Simple Summary Table

But where are the ARDs?!

gather_ard(tbl) |> 
  purrr::pluck("tbl_summary") |> 
  cards::apply_fmt_fn()

   group1 group1_level variable variable_level stat_name stat_label  stat stat_fmt
1     ARM    Xanomeli…     RACE      AMERICAN…         n          n     1        1
2     ARM    Xanomeli…     RACE      AMERICAN…         N          N    84       84
3     ARM    Xanomeli…     RACE      AMERICAN…         p          % 0.012      1.2
4     ARM    Xanomeli…     RACE      BLACK OR…         n          n     9        9
5     ARM    Xanomeli…     RACE      BLACK OR…         N          N    84       84
6     ARM    Xanomeli…     RACE      BLACK OR…         p          % 0.107     10.7
7     ARM    Xanomeli…     RACE          WHITE         n          n    74       74
8     ARM    Xanomeli…     RACE          WHITE         N          N    84       84
9     ARM    Xanomeli…     RACE          WHITE         p          % 0.881     88.1
10    ARM    Xanomeli…     RACE      AMERICAN…         n          n     0        0

Because the ARD is created as a byproduct of every gtsummary function, QC is a breeze.
Simply compare the ARD above to the ARD created with {cards}, which allows you to compare both the underlying numeric values and the rounded/formatted value that appears in the table.

AE Summary Table

tbl <- ADAE |> 
  dplyr::filter(AESOC %in% c("EYE DISORDERS", "VASCULAR DISORDERS")) |> 
  tbl_hierarchical(
    by = TRTA,
    variables = c(AESOC, AETERM),
    id = USUBJID,
    denominator = ADSL |> dplyr::rename(TRTA = ARM)
  )

Primary System Organ Class Reported Term for the Adverse Event	Xanomeline High Dose N = 84¹	Xanomeline Low Dose N = 84¹
EYE DISORDERS	1 (1.2%)	2 (2.4%)
CONJUNCTIVAL HAEMORRHAGE	0 (0.0%)	1 (1.2%)
VISION BLURRED	1 (1.2%)	1 (1.2%)
VASCULAR DISORDERS	2 (2.4%)	3 (3.6%)
HOT FLUSH	0 (0.0%)	1 (1.2%)
HYPERTENSION	1 (1.2%)	1 (1.2%)
HYPOTENSION	0 (0.0%)	1 (1.2%)
WOUND HAEMORRHAGE	1 (1.2%)	0 (0.0%)
¹ n (%)

tbl_merge()/tbl_stack()

tbl_merge() for side-by-side tables

library(survival)

tbl_uvreg <- cards::ADTTE |> 
  tbl_uvregression(
    y = Surv(AVAL, 1 - CNSR),
    include = c(TRTA, AGE),
    method = coxph,
    exponentiate = TRUE, 
    hide_n = TRUE)
tbl_uvreg

Characteristic	HR	95% CI	p-value
Actual Treatment
Placebo	—	—
Xanomeline High Dose	5.03	3.18, 7.94	<0.0001
Xanomeline Low Dose	4.15	2.65, 6.50	<0.0001
Age	0.99	0.97, 1.01	0.2494
Abbreviations: CI = Confidence Interval, HR = Hazard Ratio

tbl_reg <- 
  coxph(
    Surv(AVAL, 1-CNSR)~TRTA+AGE, 
    data = cards::ADTTE) |> 
  tbl_regression(exponentiate=TRUE)




tbl_reg

Characteristic	HR	95% CI	p-value
Actual Treatment
Placebo	—	—
Xanomeline High Dose	5.08	3.21, 8.03	<0.0001
Xanomeline Low Dose	4.22	2.69, 6.63	<0.0001
Age	0.99	0.97, 1.00	0.1326
Abbreviations: CI = Confidence Interval, HR = Hazard Ratio

tbl_merge() for side-by-side tables

# combine the tables side by side
tbl <- list(tbl_uvreg, tbl_reg) |> 
  tbl_merge(tab_spanner = c("**Univariable**", "**Multivariable**"))

Characteristic	Univariable			Multivariable
Characteristic	HR	95% CI	p-value	HR	95% CI	p-value
Actual Treatment
Placebo	—	—		—	—
Xanomeline High Dose	5.03	3.18, 7.94	<0.0001	5.08	3.21, 8.03	<0.0001
Xanomeline Low Dose	4.15	2.65, 6.50	<0.0001	4.22	2.69, 6.63	<0.0001
Age	0.99	0.97, 1.01	0.2494	0.99	0.97, 1.00	0.1326
Abbreviations: CI = Confidence Interval, HR = Hazard Ratio

tbl_stack() to combine vertically

# stack the two tables 
tbl <- list(tbl_uvreg, tbl_reg) |> 
  tbl_stack(group_header = c("Univariable", "Multivariable"))

	HR	95% CI	p-value
Univariable
Actual Treatment
Placebo	—	—
Xanomeline High Dose	5.03	3.18, 7.94	<0.0001
Xanomeline Low Dose	4.15	2.65, 6.50	<0.0001
Age	0.99	0.97, 1.01	0.2494
Multivariable
Actual Treatment
Placebo	—	—
Xanomeline High Dose	5.08	3.21, 8.03	<0.0001
Xanomeline Low Dose	4.22	2.69, 6.63	<0.0001
Age	0.99	0.97, 1.00	0.1326
Abbreviations: CI = Confidence Interval, HR = Hazard Ratio

ARD-first tables

ARD-first Tables

Similar to functions that accept a data frame, the package exports functions with nearly identical APIs that accept an ARD.

tbl_summary()

tbl_continuous()

tbl_wide_summary()

tbl_ard_summary()

tbl_ard_continuous()

tbl_ard_wide_summary()

ARD-first Tables

We can use the skills we learned earlier today to create ARDs for simple and complex gtsummary tables.

tbl <- ADSL |> 
  ard_stack( 
    .by = ARM,
    ard_continuous(variables = AGE),
    ard_categorical(variables = RACE),
    # add this for best-looking tables
    .attributes = TRUE
  ) |> 
  tbl_ard_summary(
    by = ARM,
    type = all_continuous() ~ "continuous2",
    statistic = all_continuous() ~ c("{mean} ({sd})", "{min} - {max}")
  ) |> 
  modify_caption("**Table 1. Subject Demographics**")

ARD-first Tables

**Table 1. Subject Demographics**
Characteristic	Xanomeline High Dose¹	Xanomeline Low Dose¹
Age
Mean (SD)	74.4 (7.9)	75.7 (8.3)
Min - Max	56.0 - 88.0	51.0 - 88.0
Race
AMERICAN INDIAN OR ALASKA NATIVE	1 (1.2%)	0 (0.0%)
BLACK OR AFRICAN AMERICAN	9 (10.7%)	6 (7.1%)
WHITE	74 (88.1%)	78 (92.9%)
¹ n (%)

What About Other Tables?

While our examples have focused on simple demographics tables, the ARD structure is general and any statistic can be presented.

ADSL |> 
  cardx::ard_stats_t_test_onesample(by = c(ARM, SEX), variables = AGE) |> 
  cards::update_ard_fmt_fn(stat_names = "p.value", 
                           fmt_fn = label_style_pvalue(prepend_p = TRUE)) |> 
  tbl_ard_continuous(
    by = ARM, 
    include = SEX,
    variable = AGE,
    statistic = ~"{estimate} ({conf.low}, {conf.high}; {p.value})"
  ) |> 
  modify_footnote(all_stat_cols() ~ "Mean (95% CI; p-value)")

Characteristic	Xanomeline High Dose¹	Xanomeline Low Dose¹
SEX
F	74.7 (72.2, 77.1; p<0.001)	75.7 (73.4, 78.0; p<0.001)
M	74.1 (71.6, 76.6; p<0.001)	75.6 (72.6, 78.7; p<0.001)
¹ Mean (95% CI; p-value)

ARD-first Table Shells

Characteristic	Drug A N = xx¹	Drug B N = xx¹
Patient Age, yrs
Mean (SD)	xx.x (xx.x)	xx.x (xx.x)
Min - Max	xx.x - xx.x	xx.x - xx.x
Grade
I	xx (xx.x%)	xx (xx.x%)
II	xx (xx.x%)	xx (xx.x%)
III	xx (xx.x%)	xx (xx.x%)
¹ n (%)

`as_gtsummary()`

For tables that just don’t fit any mold, use as_gtsummary().

Create a data frame of the data summary convert to gtsummary

ADAE[1:5,] |> 
  dplyr::select(USUBJID, AESOC, AETERM) |> 
  dplyr::mutate(
    .by = USUBJID, 
    USUBJID = ifelse(dplyr::row_number() == 1, USUBJID, NA)
  ) |> 
  as_gtsummary() |> 
  modify_header(USUBJID = "Patient ID") |> 
  modify_column_alignment(everything(), "left")

`as_gtsummary()`

Patient ID	Primary System Organ Class	Reported Term for the Adverse Event
01-701-1028	GENERAL DISORDERS AND ADMINISTRATION SITE CONDITIONS	APPLICATION SITE ERYTHEMA
	GENERAL DISORDERS AND ADMINISTRATION SITE CONDITIONS	APPLICATION SITE PRURITUS
01-701-1034	GENERAL DISORDERS AND ADMINISTRATION SITE CONDITIONS	APPLICATION SITE PRURITUS
	GENERAL DISORDERS AND ADMINISTRATION SITE CONDITIONS	FATIGUE
01-701-1097	SKIN AND SUBCUTANEOUS TISSUE DISORDERS	ERYTHEMA