如何在Disposition表中设置特定列留空，仅保留总计列数值？

需求背景 #

制作入组情况计算用的Disposition表，要求保留total number列（示例中为"Overall"列）的数值，指定的treatment列需保留列结构但内容为空，最终需通过rbind()函数与其他表部分合并。

待解决问题 #

• 如何使指定列的内容为空？
• 能否指定两列内容完全为空，仅保留total number列的数值？

现有代码 #

lyt <- basic_table(show_colcounts = FALSE ) %>%
  split_cols_by(
    "TRT01P",
    split_fun = remove_split_levels(c("Not Assigned", "Screen Failure"))
  ) %>%
  add_overall_col(label="Overall")%>%
  count_patients_with_flags(
    var = "USUBJID",
    flag_variables = "ENRLFL",
    .stats = "count",
  ) %>%
  count_patients_with_flags(
    var = "USUBJID",
    flag_variables = "NERFL",
    .stats = "count",
    .indent_mods = 1
  ) %>%
  analyze_vars(
    "ENTERCOM_f",
    .stats = "count",
    denom = "N_col",
    show_labels = "hidden",
    .indent_mods = c(count = 2L)
  )

lyt2 <- basic_table(show_colcounts = FALSE ) %>%
  split_cols_by(
    "TRT01P",
    split_fun = remove_split_levels(c("Not Assigned", "Screen Failure"))
  ) %>%
  add_overall_col(label="Overall")%>%
  count_patients_with_flags(
    var = "USUBJID",
    flag_variables = "RANDFL_",
    .stats = "count",
  )%>%
  # count_patients_with_flags(
  #   var = "USUBJID",
  #   flag_variables = "NOTRTFL",
  #   # .stats = "count",
  #   .indent_mods = c(count = 2L),
  # )%>%
  count_values("NOTRTFL",
               values = "Y",
               .labels = c(count_fraction = "Not treated"),
               .indent_mods = c(count_fraction = 2L)
  )%>%
  count_values("TRTFL",
               values = "Y",
               .labels = c(count_fraction = "Treated, i.e. Randomised Treatment"),
               .indent_mods = c(count_fraction = 2L)
  )%>%
  count_patients_with_flags(
    var = "USUBJID",
    flag_variables = "OVRTFL",
    .stats = "count",
    .indent_mods = 3
  )%>%
  count_patients_with_flags(
    var = "USUBJID",
    flag_variables = "OVRCOM",
    .stats = "count",
    .indent_mods = 4
  )%>%
  count_patients_with_flags(
    var = "USUBJID",
    flag_variables = "OVRONGO",
    .stats = "count",
    .indent_mods = 4
  )%>%
  count_patients_with_flags(
    var = "USUBJID",
    flag_variables = "OVRDIS",
    .stats = "count",
    .indent_mods = 4
  )%>%
  analyze_vars(
    "OVRDIS_f",
    .stats = "count_fraction",
    denom = "N_col",
    show_labels = "hidden",
    .indent_mods = c(count_fraction = 5L)
  )%>%
  count_patients_with_flags(
    var = "USUBJID",
    flag_variables = "RANDTFL",
    .stats = "count",
    .indent_mods = 3
  )%>%
  count_patients_with_flags(
    var = "USUBJID",
    flag_variables = "RANDCOM",
    .stats = "count",
    .indent_mods = 4
  )%>%
  count_patients_with_flags(
    var = "USUBJID",
    flag_variables = "RANDONGO",
    .stats = "count",
    .indent_mods = 4
  )%>%
  count_patients_with_flags(
    var = "USUBJID",
    flag_variables = "RANDDIS",
    .stats = "count",
    .indent_mods = 4
  )%>%
  analyze_vars(
    "RANDDIS_f",
    .stats = "count_fraction",
    denom = "N_col",
    show_labels = "hidden",
    .indent_mods = c(count_fraction = 5L)
  )

result1 <- build_table(lyt = lyt, df = adsl)
result1
result2 <- build_table(lyt = lyt2, df = adsl2)
result2

col_info(result2) <- col_info(result1)

result <- rbind(result1, result2)

解决方法 #

1. 清空指定列内容 #

构建完表格后，直接通过列索引或列名定位目标列，将内容替换为空字符串（或NA）即可：

# 方式1：通过列索引定位（假设treatment列是第2、3列，根据实际结构调整）
result1[, 2:3] <- ""

# 方式2：通过列名精准匹配（假设"Overall"是total number列，其余为treatment列）
trt_cols <- setdiff(colnames(result1), "Overall")
result1[, trt_cols] <- ""

2. 仅保留total number列数值，清空其他指定列 #

逻辑和上述一致，先确定total number对应的列（示例中为"Overall"），然后将所有需要清空的列内容置空：

# 保留Overall列，清空所有treatment列
trt_cols <- setdiff(colnames(result2), "Overall")
result2[, trt_cols] <- ""

处理后的表格列结构保持不变，仅指定列内容为空，可直接通过rbind()与其他表格合并。

内容的提问来源于stack exchange，提问作者xiecheng gu