Subgroup Analyses

Function adapted from gtforester::tbl_subgroups().

tbl_roche_subgroups(data, rsp, by, subgroups, .tbl_fun, time_to_event = NULL)

Arguments

data

(data.frame, survey.design)
a data frame or survey object

rsp

(tidy-select)
Variable to use in responder rate calculations.

by

(tidy-select)
Variable to make comparison between groups.

subgroups

(tidy-select)
Variables to perform stratified analyses for.

.tbl_fun

(function) A function or formula. If a function, it is used as is. If a formula, e.g. ~ .x %>% tbl_summary() %>% add_p(), it is converted to a function. The stratified data frame is passed to this function.

time_to_event

(tidy-select)
Variable to use in time-to-event analyses. If specified, the mid table will show the median time-to-event instead of responder rates.

Value

a 'gtsummary' table

Examples

set.seed(1)

# prepare sample data
df_adtte <- data.frame(
  time = rexp(100, rate = 0.1),
  status = sample(c(0, 1), 100, replace = TRUE),
  arm = sample(c("Arm A", "Arm B"), 100, replace = TRUE),
  grade = sample(c("I", "II"), 100, replace = TRUE),
  strata = sample(c("1", "2"), 100, replace = TRUE)
) |>
  mutate(arm = relevel(factor(arm), ref = "Arm A")) # Set Reference

# logistic regression -------------------------------------------------------
df_adtte |>
  tbl_roche_subgroups(
    rsp = "status",
    by = "arm",
    subgroups = c("grade"),
    .tbl_fun =
      ~ glm(status ~ arm, data = .x) |>
        tbl_regression(
          show_single_row = arm,
          exponentiate = TRUE # , tidy_fun = broom.helpers::tidy_parameters
        )
  ) |>
  modify_header(starts_with("estimate") ~ "**Odds Ratio**")


  
Baseline Risk Factors
      
Total n
      

        
Arm A
      
      

        
Arm B
      
      
Odds Ratio
      
95% CI
      
p-value
    
n Response (%)
      
n Response (%)
    
All Participants
100
50 (64.0 %)
50 (40.0 %)
0.79
0.65, 0.95
0.0161
grade











    I
57
33 (66.7 %)
24 (37.5 %)
0.75
0.58, 0.96
0.0292
    II
43
17 (58.8 %)
26 (42.3 %)
0.85
0.62, 1.15
0.3007

# coxph regression ----------------------------------------------------------
# please use browser() inside .tbl_fun to check if the coxph model throws an error
# and use tryCatch to modify the input/output accordingly
df_adtte |>
  tbl_roche_subgroups(
    rsp = status,
    by = arm,
    time_to_event = time, # Specify time variable for time-to-event analyses (different mid table)
    subgroups = c(grade, strata),
    ~ survival::coxph( # Please use coxph for time-to-event analyses
      survival::Surv(time, status) ~ arm,
      data = .x,
      ties = "exact"
    ) |> # Exact Ties
      tbl_regression(
        show_single_row = arm,
        exponentiate = TRUE # Get Hazard Ratios
      )
  ) |>
  modify_header(starts_with("estimate") ~ "**Hazard Ratio**")


  
Baseline Risk Factors
      
Total n
      

        
Arm A
      
      

        
Arm B
      
      
Hazard Ratio
      
95% CI
      
p-value
    
n
      
Median (Months)
      
n
      
Median (Months)
    
All Participants
100
50
7.8
50
8.1
0.70
0.40, 1.23
0.2126
grade















    I
57
33
6.8
24
10.2
0.48
0.22, 1.07
0.0718
    II
43
17
8.4
26
7.6
1.26
0.48, 3.34
0.6396
strata















    1
52
27
8.0
25
7.7
0.65
0.29, 1.45
0.2901
    2
48
23
5.5
25
9.6
0.73
0.33, 1.63
0.4442