如何覆盖 TERN 函数中的自动格式（例如 count_患者_with_flags）？

Question

TERN (Rtables) 包格式化函数 这里有人熟悉 TERN 函数中的覆盖自动格式（例如 count_患者_with_flags）吗？

lyt_adae <- basic_table(show_colcounts = TRUE) %>%
   split_cols_by("ACTARM") %>%
   analyze_num_patients(
     vars = "USUBJID",
     .stats = c("unique", "nonunique"),
     .labels = c(
       unique = "Total number of patients with at least one AE",
       nonunique = "Total number of AEs"
     ),
     .formats = list(unique = format_count_fraction_fixed_dp, nonunique = "xx"),
     show_labels = "hidden"
   ) %>%
   count_patients_with_flags(
     "USUBJID",
     flag_variables = aesi_vars,
     denom = "N_col",
     var_labels = "Total number of patients with at least one",
     show_labels = "visible"
   )

result_adae <- build_table(lyt_adae, df = adae)

我正在尝试实现一件非常简单的事情：将变量计数和百分比更改为分数和百分比。所以从 10 (50%) 到 10/20 (50%)。

与上面的analyze_num_患者函数一样，我认为我可以使用添加了.formats的代码覆盖count_患者_with_flags函数中的默认函数：

 count_patients_with_flags(
     "USUBJID",
     flag_variables = aesi_vars,
     denom = "N_col",
     var_labels = "Total number of patients with at least one",
     show_labels = "visible",
    .formats = c(count_fraction = format_count_fraction))
   )                     

但是，format_count_fraction是一种仅适用于向量的格式，并且被扔进函数的对象是S4。另外，format_count_fraction

错误：

任何帮助将不胜感激。对于初学者来说，还有比 Hadley's Advanced 更好的学习 R 的 S3、S4 等 OOP 的书籍吗？他的基础数据科学书很棒，但是高级书很难理解。

谢谢！

我尝试使用 as.vector 函数手动将表格的每个单元格转换为向量，然后应用粗略的自定义格式函数来实现分数 + 百分比，但这太乏味了，我一直坚持将向量重新转换回 S4 兼容要输出到最终表格单元格的对象。

format_rcell（result_adae[2,1]，custom_format）

Answer 1

您可以使用自定义格式函数覆盖格式。请注意，您需要再次从分数中检索总计数。

library(tern)
library(dplyr)

# Add labelled flag variables to analysis dataset.
adae <- tern_ex_adae %>%
  mutate(
    fl1 = TRUE %>% with_label("Total AEs"),
    fl2 = (TRTEMFL == "Y") %>%
      with_label("Total number of patients with at least one adverse event"),
    fl3 = (TRTEMFL == "Y" & AEOUT == "FATAL") %>%
      with_label("Total number of patients with fatal AEs"),
    fl4 = (TRTEMFL == "Y" & AEOUT == "FATAL" & AEREL == "Y") %>%
      with_label("Total number of patients with related fatal AEs")
  )
my_format_func <- function(x, ...) {
  attr(x, "label") <- NULL
  
  if (any(is.na(x))) {
    return("NA")
  }
  # browser()
  checkmate::assert_vector(x)
  checkmate::assert_int(x[1])
  checkmate::assert_number(x[2], lower = 0, upper = 1) # the stat returns a fraction
  
  result <- if (x[1] == 0) {
    paste0(x[1])
  } else {
    dn <- round(x[1] / x[2]) # Retrieving the denominator
    paste0(
      x[1], "/", dn,
      " (", round(x[2] * 100, 1), "%)"
    )
  }
  
  return(result)
}
lyt_adae <- basic_table(show_colcounts = TRUE) %>%
  split_cols_by("ACTARM") %>%
  analyze_num_patients(
    vars = "USUBJID",
    .stats = c("unique", "nonunique"),
    .labels = c(
      unique = "Total number of patients with at least one AE",
      nonunique = "Total number of AEs"
    ),
    .formats = list(unique = format_count_fraction_fixed_dp, nonunique = "xx"),
    show_labels = "hidden"
  ) %>%
  count_patients_with_flags(
    "USUBJID",
    flag_variables = c("fl1", "fl2", "fl3", "fl4"),
    denom = "N_col",
    var_labels = "Total number of patients with at least one",
    show_labels = "visible",
    .formats = c(count_fraction = my_format_func)
  )

result_adae <- build_table(lyt_adae, df = adae)
result_adae
#>                                                                A: Drug X        B: Placebo     C: Combination
#>                                                                 (N=202)          (N=177)          (N=162)    
#> —————————————————————————————————————————————————————————————————————————————————————————————————————————————
#> Total number of patients with at least one AE                  59 (29.2%)       57 (32.2%)       48 (29.6%)  
#> Total number of AEs                                               202              177              162      
#> Total number of patients with at least one                                                                   
#>   Total AEs                                                  59/202 (29.2%)   57/177 (32.2%)   48/162 (29.6%)
#>   Total number of patients with at least one adverse event   59/202 (29.2%)   57/177 (32.2%)   48/162 (29.6%)
#>   Total number of patients with fatal AEs                    28/202 (13.9%)   31/177 (17.5%)   20/162 (12.3%)
#>   Total number of patients with related fatal AEs            28/202 (13.9%)   31/177 (17.5%)   20/162 (12.3%)

^{创建于 2024-07-03，使用 reprex v2.1.0}