向包含多个模型的 gtsummary 回归表添加多级标题

我正在尝试为一个生成多个级别的标头gtsummary回归表包含应在表中按年份分组的回归模型。这是一个玩具示例：

library(dplyr)
library(gtsummary)
library(purrr)

set.seed(92922)
df <- tibble(y_1980 = rbinom(n = 10, size = 1, prob = .4),
             y_1990 = rbinom(n = 10, size = 1, prob = .7),
             x1 = rnorm(10, sd = 1),
             x2 = rnorm(10, sd = 2))

tbls <- c("y_1980 ~ x1", "y_1980 ~ x1 + x2", "y_1990 ~ x1", "y_1990 ~ x1 + x2") %>% 
  map(as.formula) %>% 
  map(glm,
      data = df, 
      family = binomial(link = "logit")) %>% 
  map(tbl_regression, exponentiate = TRUE) %>% 
  map(add_significance_stars, hide_ci = TRUE, hide_p = TRUE, hide_se = FALSE) %>% 
  map(add_glance_table, include = nobs) 

我可以得到这个：

tbls %>% 
  tbl_merge(tab_spanner = c("1980 (1)", "1980 (2)", "1990 (1)", "1990 (2)")) %>% 
  modify_table_body(~.x %>% dplyr::arrange(row_type == "glance_statistic"))

但我想要两个级别——年份在上，型号在下：

 1980      1990
_______   ______
(1) (2)   (1) (2)

我怎样才能做到这一点？

您需要将 gtsummary 表转换为 gt 表。然后你可以使用gt::tab_spanner()放置更高级别的跨越标题。

library(dplyr)
library(gtsummary)
library(purrr)

set.seed(92922)
df <- tibble(y_1980 = rbinom(n = 10, size = 1, prob = .4),
             y_1990 = rbinom(n = 10, size = 1, prob = .7),
             x1 = rnorm(10, sd = 1),
             x2 = rnorm(10, sd = 2))

tbls <- c("y_1980 ~ x1", "y_1980 ~ x1 + x2", "y_1990 ~ x1", "y_1990 ~ x1 + x2") %>% 
  map(as.formula) %>% 
  map(glm,
      data = df, 
      family = binomial(link = "logit")) %>% 
  map(tbl_regression, exponentiate = TRUE) %>% 
  map(add_significance_stars, hide_ci = TRUE, hide_p = TRUE, hide_se = FALSE) %>% 
  map(add_glance_table, include = nobs) 


tbl <-
  tbls %>% 
  tbl_merge(tab_spanner = FALSE) %>% 
  modify_table_body(~.x %>% dplyr::arrange(row_type == "glance_statistic"))

show_header_names(tbl)

gt_tbl <- 
  as_gt(tbl) %>%
  gt::tab_spanner(
    columns = c(estimate_1, std.error_1, estimate_3, std.error_3),
    label = "(1)",
    gather = FALSE
  ) %>%
  gt::tab_spanner(
    columns = c(estimate_2, std.error_2, estimate_4, std.error_4),
    label = "(2)",
    gather = FALSE
  ) %>%
  gt::tab_spanner(
    columns = c(estimate_1, std.error_1, estimate_2, std.error_2),
    label = "1980",
    level = 2,
    gather = FALSE
  ) %>%
  gt::tab_spanner(
    columns = c(estimate_3, std.error_3, estimate_4, std.error_4),
    label = "1990",
    level = 2,
    gather = FALSE
  )