在R中执行加权分层Logistic回归：分层结果展示问题求助

问题根源 #

你当前用svyglm()拟合的是整合所有分层的全局模型，而非按gd分层分别拟合独立模型，因此summary()和tbl_regression()只会返回整体回归结果，无法展示各分层的单独结果。tbl_uvregression()能输出分层单变量结果是因为它内置了by参数支持自动分层拟合，而tbl_regression()需要手动实现分层拟合逻辑。

解决方案：按分层批量拟合模型并合并结果 #

方法1：手动拆分+合并表格 #

先按分层变量gd拆分数据集，分别拟合模型后，用tbl_regression()生成各分层结果表格，再通过tbl_merge()合并展示：

library(tidyverse)
library(gtsummary)
library(survey)

# 原数据集构建（同你提供的代码）
gender <- factor(c(1, 2, 1, 1, 1, 1, 2, 2, 1, 2, 1, 2, 1, 2, 2, 1, 1, 1, 1, 2))
age <- factor(c(3, 4, 2, 1, 3, 1, 1, 2, 3, 3, 4, 1, 3, 1, 3, 4, 2, 1, 2, 1))
prema.5cl <- factor(c(5, 4, 5, 2, 1, 1, 3, 1, 5, 1, 2, 4, 4, 3, 1, 2, 1, 4, 5, 2))
meduc.4cl <- factor(c(1, 2, 3, 3, 1, 4, 2, 1, 4, 2, 1, 3, 3, 4, 4, 2, 4, 3, 1, 2))
mhealth <- factor(c(1, 1, 1, 0, 0, 1, 0, 0, 1, 1, 1, 0, 1, 1, 0, 1, 0, 1, 0, 0))
parttime <- factor(c(1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 0, 0, 1, 0, 1, 0, 0))
ref1 <- factor(c(0, 0, 0, 0, 0, 1, 1, 1, 0, 1, 0, 0, 1, 0, 1, 0, 1, 1, 0, 0))
ref2 <- factor(c(0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 0, 1, 1, 0, 0, 1, 1, 0))
ref3 <- factor(c(0, 1, 0, 1, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 0, 1, 1, 1))
poids <- c(0.52, 1.43, 0.74, 1.60, 1.71, 0.08, 0.96, 1.62, 1.00, 0.83, 1.74, 0.82, 1.23, 1.04, 0.19, 1.63, 0.45, 0.08, 0.60, 1.73)
gd <- factor(c("gp_2", "gp_1", "gp_2", "gp_1", "gp_2", "gp_1", "gp_1", "gp_1", "gp_2", "gp_2", "gp_1", "gp_1", "gp_2", "gp_1", "gp_1", "gp_2", "gp_2", "gp_1", "gp_2", "gp_2"))
bth_2y <- factor(c(1, 1, 0, 0, 1, 0, 0, 0, 1, 1, 0, 1, 0, 0, 0, 1, 0, 1, 1, 0))
db <- data.frame(gender, age, prema.5cl, meduc.4cl, mhealth, parttime, ref1, ref2, ref3, poids, gd, bth_2y)

# 按gd分层拆分数据集
db_split <- split(db, db$gd)

# 分别拟合两个分层的模型
reg_gp1 <- svyglm(
  bth_2y ~ gender + age + prema.5cl + meduc.4cl + mhealth + ref1 + ref2 + ref3,
  family = quasibinomial,
  design = svydesign(id=~1, data = db_split$gp_1, weights = ~poids))

reg_gp2 <- svyglm(
  bth_2y ~ gender + age + prema.5cl + meduc.4cl + mhealth + ref1 + ref2 + ref3,
  family = quasibinomial,
  design = svydesign(id=~1, data = db_split$gp_2, weights = ~poids))

# 生成各分层的结果表格并设置变量标签
tbl_gp1 <- tbl_regression(reg_gp1, label = list(
  gender ~ "性别", age ~ "年龄", prema.5cl ~ "孕前抑郁程度",
  meduc.4cl ~ "母亲教育水平", mhealth ~ "母亲健康状况",
  ref1 ~ "参考变量1", ref2 ~ "参考变量2", ref3 ~ "参考变量3"
)) %>% modify_caption("**分层gp_1的回归结果**")

tbl_gp2 <- tbl_regression(reg_gp2, label = list(
  gender ~ "性别", age ~ "年龄", prema.5cl ~ "孕前抑郁程度",
  meduc.4cl ~ "母亲教育水平", mhealth ~ "母亲健康状况",
  ref1 ~ "参考变量1", ref2 ~ "参考变量2", ref3 ~ "参考变量3"
)) %>% modify_caption("**分层gp_2的回归结果**")

# 合并两个分层表格
tbl_merge(list(tbl_gp1, tbl_gp2), tab_spanner = c("**gp_1**", "**gp_2**"))

方法2：用purrr批量处理（适配多分层场景） #

如果分层数量较多，用map()函数批量完成模型拟合和表格生成：

# 批量拟合所有分层的模型
reg_list <- map(db_split, ~svyglm(
  bth_2y ~ gender + age + prema.5cl + meduc.4cl + mhealth + ref1 + ref2 + ref3,
  family = quasibinomial,
  design = svydesign(id=~1, data = ., weights = ~poids)
))

# 批量生成各分层的结果表格
tbl_list <- map2(reg_list, names(reg_list), ~tbl_regression(.x, label = list(
  gender ~ "性别", age ~ "年龄", prema.5cl ~ "孕前抑郁程度",
  meduc.4cl ~ "母亲教育水平", mhealth ~ "母亲健康状况",
  ref1 ~ "参考变量1", ref2 ~ "参考变量2", ref3 ~ "参考变量3"
)) %>% modify_caption(sprintf("**分层%s的回归结果**", .y)))

# 合并所有分层表格
tbl_merge(tbl_list, tab_spanner = paste0("**", names(reg_list), "**"))

补充说明 #

• 若需检验分层间的系数差异，可在全局模型中加入自变量与分层变量的交互项（如bth_2y ~ gd*(gender + age + prema.5cl + ...)），通过summary()查看交互项的显著性。
• gtsummary的tbl_merge()支持自定义表格标题、列标签等格式，可根据需求进一步调整。

内容的提问来源于stack exchange，提问作者Curbice