Shiny中用gtsummary生成分组生存表时遇'symbol不可子集化'错误

构建Shiny应用生成指定时间点的生存百分比表，支持分组对比（如性别、药物组）。无分组的生存表可正常运行，但添加分组变量后，运行报错："object of type 'symbol' is not subsettable"。调试确认survfit模型拟合正常，问题出在gtsummary的tbl_ard_summary构建环节。

问题核心代码片段：

tbl_survival <- 
  fit |> 
  cardx::ard_survival_survfit(times = input$timepoint) |> 
  cards::update_ard_fmt_fn(
    stat_names = c("estimate", "conf.low", "conf.high"),
    fmt_fn = label_style_sigfig(digits = 2, scale = 100)
  ) |> 
  dplyr::mutate(context = "categorical") |> 
  tbl_ard_summary(
    by = if (input$group_var == "None") NULL else input$group_var, # 此处直接传字符串触发错误
    label = list(time = paste0(input$timepoint, " months Survival Probability")),
    statistic = time ~ "{estimate}% (95% CI {conf.low}%, {conf.high}%)"
  )

问题根源 #

tbl_ard_summary的by参数要求传入符号（symbol）（如gender）而非字符串（如"gender"），但Shiny中input$group_var返回的是字符串类型，直接传递会触发类型不匹配的错误。

修改步骤 #

1. 将字符串类型的分组变量转换为符号：用rlang::sym()函数把input$group_var的字符串值转为gtsummary可识别的符号。
2. 移除dplyr::mutate(context = "categorical")：该手动修改上下文的操作会干扰gtsummary对生存分析结果的自动解析，无需保留。

修改后的完整Server代码 #

server <- function(input, output, session) {
  
  output$surv_table <- render_gt({
    # Validate inputs
    req(input$time_var, input$status_var, input$timepoint)
    
    # Create the survival object
    surv_obj <- Surv(time = data[[input$time_var]], event = data[[input$status_var]])
    
    # Determine grouping formula
    if (input$group_var == "None") {
      formula <- surv_obj ~ 1
    } else {
      formula <- as.formula(paste("surv_obj ~", input$group_var))
    }
    
    # Fit the survival model
    fit <- survfit(formula, data = data)
    
    # Build tbl_survival using the input value
    tbl_survival <- 
      fit |> 
      cardx::ard_survival_survfit(times = input$timepoint) |> 
      cards::update_ard_fmt_fn(
        stat_names = c("estimate", "conf.low", "conf.high"),
        fmt_fn = label_style_sigfig(digits = 2, scale = 100)
      ) |> 
      tbl_ard_summary(
        # 关键修改：将字符串转换为符号
        by = if (input$group_var == "None") NULL else rlang::sym(input$group_var),
        label = list(time = paste0(input$timepoint, " months Survival Probability")),
        statistic = time ~ "{estimate}% (95% CI {conf.low}%, {conf.high}%)"
      )
    
    # Convert to gt table before rendering
    gt_table <- as_gt(tbl_survival)
    
    # Render the table
    gt_table
  })
}

替代方案（不依赖rlang包） #

使用.data代词直接引用列名：

by = if (input$group_var == "None") NULL else .data[[input$group_var]]

内容的提问来源于stack exchange，提问作者Ivan Dlouhy