将因子水平从冒号更改为行

Question

我使用 gtsummary 的 tbl_summary 制作了调查问题的汇总表，其中答案有 5 或 7 个级别同意、不同意等，但表格很长。有什么方法可以制作表格，其中答案的级别位于列中？

这是我的代码：

table4<-physician[ ,32:43] %>% tbl_summary() %>%  
  modify_footnote(all_stat_cols()~"Frequency (%)") %>% 
  modify_caption("** Table 7. The frequency with which prescribers provided antibiotics or resources related to prudent use of antibiotics,
behavior on drivers for initiating prescriptions and antibiotic prescribing behaviors **") %>% 
  modify_header(label~"Item") %>% bold_labels() 

我希望我的桌子看起来像这样：

Item        SA   A   D   SD  N/A  U  IDU
quesion     56   47  35  89  15   28  10

问题在行中，每个级别的级别和计数在冒号中

Answer 1

它有助于提供示例数据。如果没有它，我在下面创建了一个简短的示例：

set.seed(44)

df <- data.frame(
  question1 = sample(c("Strongly Agree", "Agree", "Disagree", "Strongly Disagree", NA), 15, replace = TRUE),
  question2 = sample(c("Strongly Agree", "Agree", "Disagree", "Strongly Disagree", NA), 15, replace = TRUE),
  question3 = sample(c("Strongly Agree", "Agree", "Disagree", "Strongly Disagree", NA), 15, replace = TRUE)
)

这里每个问题都有一栏，每一栏都包含一个 Likert 类型的答案。

           question1         question2         question3
1     Strongly Agree              <NA>    Strongly Agree
2           Disagree Strongly Disagree              <NA>
3           Disagree              <NA>              <NA>
4     Strongly Agree              <NA>             Agree
5  Strongly Disagree    Strongly Agree             Agree
6     Strongly Agree    Strongly Agree          Disagree
7               <NA>             Agree Strongly Disagree
8     Strongly Agree              <NA>              <NA>
9     Strongly Agree              <NA>          Disagree
10             Agree              <NA>             Agree
11             Agree    Strongly Agree    Strongly Agree
12              <NA>    Strongly Agree    Strongly Agree
13    Strongly Agree Strongly Disagree          Disagree
14    Strongly Agree             Agree             Agree
15    Strongly Agree          Disagree Strongly Disagree

这可能是

bstfun

包的理想选择，它可以创建

gtsummary

类似李克特型量表的表格。目前CRAN上还没有，但是你可以下载开发版本（我注释掉了）。

# devtools::install_github("MSKCC-Epi-Bio/bstfun")

library(bstfun)

tbl_likert(df)

输出

或者，您可以单独使用

gtsummary

。您可以创建单独的表格行，然后将它们堆叠起来以创建完整的表格。您也可以轻松指定缺失的列。

library(tidyverse)
library(gtsummary)

map(
  names(df),
  ~ df |>
    select(all_of(.x)) |>
    mutate(across(all_of(.x), ~fct_na_value_to_level(., "Missing")),
           label_char = TRUE) |> 
    tbl_summary(
      by = all_of(.x),
      statistic = ~"{n}",
      missing = "no",
      label = list(label_char = .x)
    ) |>
    modify_header(
      all_stat_cols() ~ "**{level}**",
      label = "**Question**"
    )
) |>
  tbl_stack()

输出