如何在R中合并多变量FND症状并统计单症状分层频率

我正在开展一项研究，旨在对比接受特定脑外科手术患者的术后并发症。其中一项分析需使用CreateTableOne函数对比患者术后的FND（局灶性神经功能缺损）症状。目前症状被收集为四个分类变量FNDtype、FNDtype2、FNDtype3、FNDtype4，每个患者最多有4种症状，部分患者更少。我尝试用paste函数将这些变量合并为一个，但使用CreateTableOne生成描述性分析时，结果显示的是症状组合的统计数据，而非单个症状在各分层组中的频率。

相关R代码 #

# Postcraniectomia Infection
data$postcraniotomy_infection <- factor(data$`Postcraniotomy infection (y=1, n=0)`, levels = c(0,1), labels = c("no", "yes"))

# FND Type 
data$FNDtype <- factor(data$`Type of FND (1)`, levels = c(0,1,2,3,4,5,6,7,8,9), labels = c("none", "hemiparesis", "hypoesthesia", "aphasia", "visual field defects", "coordiantion disturbance", "cranial nerve dysfunction", "neglect","dysarthria","other"))

# FND Type 2 
data$FNDtype2 <- factor(data$`Type of FND (2)`, levels = c(0,1,2,3,4,5,6,7,8,9), labels = c("none", "hemiparesis", "hypoesthesia", "aphasia", "visual field defects", "coordiantion disturbance", "cranial nerve dysfunction", "neglect","dysarthria","other"))
                        
# FND Type 3 
data$FNDtype3 <- factor(data$`Type of FND (3)`, levels = c(0,1,2,3,4,5,6,7,8,9), labels = c("none", "hemiparesis", "hypoesthesia", "aphasia", "visual field defects", "coordiantion disturbance", "cranial nerve dysfunction", "neglect","dysarthria","other"))
                       
# FND Type 4 
data$FNDtype4 <- factor(data$`Type of FND (4)`, levels = c(0,1,2,3,4,5,6,7,8,9), labels = c("none", "hemiparesis", "hypoesthesia", "aphasia", "visual field defects", "coordiantion disturbance", "cranial nerve dysfunction", "neglect","dysarthria","other"))

# Trying to merge FND Types into one variable
data$fndtype <- paste(data$FNDtype, data$FNDtype2, data$FNDtype3, data$FNDtype4, sep = ",") 

vars <- c("fndtype")

table <- CreateTableOne(vars = vars, data = data, strata = c("primary_infection"), testNonNormal = kruskal.test, addOverall = TRUE)

kableone(table, nonnormal = vars)

数据集 #

list(postcraniotomy_infection = structure(c(2L, 1L, 2L, 2L, 2L, 
1L, 2L, 1L, 2L, 1L, 2L, 2L, 1L, 1L, 2L, 1L, 1L, 1L, 2L, 2L, 1L, 
2L, 2L, 2L, 2L, 2L, 1L, 1L, 2L, 2L, 1L, 2L, 2L, 2L, 1L, 2L, 1L, 
2L, 2L, 1L, 2L, 1L, 2L, 2L, 1L, 1L, 2L, 2L, 2L, 1L, 1L, 1L, 2L, 
2L, 1L, 2L, 2L, 1L, 2L, 2L, 1L, 1L, 2L, 1L, 1L, 2L, 1L, 2L, 2L, 
2L, 1L, 2L, 2L, 1L, 1L, 2L, 2L, 1L, 1L, 1L, 1L, 2L, 2L, 1L), levels = c("no", 
"yes"), class = "factor"), FNDtype = structure(c(1L, 2L, 2L, 
1L, 2L, 1L, 8L, 2L, 4L, 4L, 1L, 3L, 4L, 2L, 2L, 4L, 2L, 3L, 1L, 
1L, 1L, 3L, 1L, 9L, 7L, 1L, 2L, 2L, 1L, 1L, 5L, 1L, 3L, 1L, 2L, 
3L, 3L, 2L, 4L, 2L, 1L, 10L, 2L, 1L, 3L, 6L, 1L, 1L, 1L, 1L, 
2L, 2L, 1L, 1L, 6L, 1L, 5L, 10L, 1L, 1L, 2L, 10L, 1L, 1L, 1L, 
2L, 1L, 1L, 1L, 5L, 5L, 1L, 1L, 2L, 2L, 1L, 2L, 2L, 2L, 5L, 3L, 
1L, 3L, 1L), levels = c("none", "hemiparesis", "hypoesthesia", 
"aphasia", "visual field defects", "coordiantion disturbance", 
"cranial nerve dysfunction", "neglect", "dysarthria", "other"
), class = "factor"), FNDtype2 = structure(c(1L, 1L, 1L, 1L, 
3L, 1L, 1L, 5L, 1L, 8L, 1L, 7L, 1L, 4L, 1L, 7L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 8L, 7L, 1L, 1L, 1L, 1L, 1L, 1L, 3L, 7L, 
1L, 6L, 6L, 3L, 1L, 1L, 6L, 1L, 9L, 10L, 1L, 1L, 1L, 1L, 1L, 
3L, 1L, 1L, 8L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 3L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 3L, 4L, 1L, 4L, 9L, 6L, 1L, 4L, 1L, 6L, 
1L), levels = c("none", "hemiparesis", "hypoesthesia", "aphasia", 
"visual field defects", "coordiantion disturbance", "cranial nerve dysfunction", 
"neglect", "dysarthria", "other"), class = "factor"), FNDtype3 = structure(c(1L, 
1L, 1L, 1L, 6L, 1L, 1L, 8L, 1L, 1L, 1L, 8L, 1L, 7L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 9L, 8L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 6L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 4L, 1L, 1L, 10L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 4L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 8L, 1L, 6L, 1L, 9L, 1L, 
1L, 1L, 1L, 1L), levels = c("none", "hemiparesis", "hypoesthesia", 
"aphasia", "visual field defects", "coordiantion disturbance", 
"cranial nerve dysfunction", "neglect", "dysarthria", "other"
), class = "factor"), FNDtype4 = structure(c(1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 9L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 8L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 10L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L
), levels = c("none", "hemiparesis", "hypoesthesia", "aphasia", 
"visual field defects", "coordiantion disturbance", "cranial nerve dysfunction", 
"neglect", "dysarthria", "other"), class = "factor"))

要统计单个症状在各分层组中的频率，不能直接用paste合并变量（这会生成症状组合），正确的做法是将多个FND变量转换为长格式数据，标记每个症状的存在情况后再转为宽格式的二元变量，最后用CreateTableOne分析这些二元变量。

步骤1：加载必要的包 #

library(tableone)
library(dplyr)
library(tidyr)

步骤2：转换数据格式并生成症状二元变量 #

# 提取FND相关列，转换为长格式
fnd_long <- data %>%
  select(postcraniotomy_infection, starts_with("FNDtype")) %>%
  pivot_longer(cols = starts_with("FNDtype"), names_to = "fnd_col", values_to = "symptom") %>%
  # 排除无病症的记录
  filter(symptom != "none") %>%
  # 标记症状存在
  mutate(present = 1) %>%
  # 去重，避免同一患者同一症状重复计数
  distinct(postcraniotomy_infection, symptom, .keep_all = TRUE)

# 转换为宽格式，每个症状对应一列，未出现的症状填0
fnd_wide <- fnd_long %>%
  pivot_wider(names_from = symptom, values_from = present, values_fill = 0) %>%
  # 合并回原数据的分层变量（若原分层变量是primary_infection，替换此处的postcraniotomy_infection）
  right_join(data %>% select(postcraniotomy_infection), by = "postcraniotomy_infection") %>%
  # 将缺失值替换为0，表示未出现该症状
  mutate(across(where(is.numeric), ~replace_na(.x, 0)))

步骤3：用CreateTableOne分析单个症状 #

# 获取所有症状变量名
symptom_vars <- setdiff(colnames(fnd_wide), "postcraniotomy_infection")

# 创建描述性表格，按分层变量分组（若原分层变量是primary_infection，替换此处的postcraniotomy_infection）
table <- CreateTableOne(vars = symptom_vars, data = fnd_wide, strata = "postcraniotomy_infection", addOverall = TRUE)

# 输出结果
kableone(table)

关键说明 #

• 该方法会生成每个症状在各分层组中的频率（百分比），而非症状组合的统计结果。
• 若原代码中的分层变量是primary_infection，请将上述代码中的postcraniotomy_infection统一替换为该变量名。
• 转换过程自动过滤了"none"的记录，仅统计实际出现的症状。

内容的提问来源于stack exchange，提问作者Massimo Barbagallo