如何去除R语言gtsummary包生成表格（含回归表格）中的前导零以符合APA规范？

Great question—gtsummary is such a powerful tool for flexible table building, and tweaking it to fit APA standards is totally doable with a few targeted functions. Let’s break this down step by step:

Core Tools to Use #

The star player here is modify_fmt_fun()—it lets you apply custom formatting rules to specific columns in your table. We’ll pair this with built-in gtsummary formatters and a tiny custom helper function to strip leading zeros.

Step 1: Define a Leading-Zero Remover Helper #

First, make a quick function to turn values like 0.85 into .85 by stripping that leading 0:

remove_leading_zero <- function(x) {
  stringr::str_replace(x, "^0\\.", "\\.")
}

This uses stringr (which integrates seamlessly with gtsummary) to target strings starting with 0. and swap them for just ..

Step 2: Format Regression Tables for APA Compliance #

Let’s use a linear regression example to put this into action. We’ll handle three key APA-aligned adjustments: coefficients, p-values, and confidence intervals:

library(gtsummary)
library(stringr)

# Fit a sample regression model
my_model <- lm(mpg ~ wt + hp + am, data = mtcars)

# Build and format the table to meet APA specs
tbl_regression(my_model) %>%
  # Format coefficients: remove leading zero, keep 2 decimal places
  modify_fmt_fun(
    columns = estimate,
    fun = function(x) remove_leading_zero(fmt_number(x, digits = 2))
  ) %>%
  # Format p-values: APA style (no leading zero, <.001 for tiny values)
  modify_fmt_fun(
    columns = p.value,
    fun = fmt_pvalue,
    leading_zero = FALSE,  # This removes the leading zero for p-values
    digits = 3,
    cutoff = 0.001  # Shows "<.001" instead of exact values below this threshold
  ) %>%
  # Format confidence intervals: remove leading zeros from both bounds
  modify_fmt_fun(
    columns = ci,
    fun = function(x) {
      # Split CI into lower/upper bounds, clean, remove zeros, then rejoin
      ci_parts <- str_split(x, ", ", simplify = TRUE)
      ci_parts[,1] <- remove_leading_zero(str_remove(ci_parts[,1], "\\("))
      ci_parts[,2] <- remove_leading_zero(str_remove(ci_parts[,2], "\\)"))
      paste0("(", ci_parts[,1], ", ", ci_parts[,2], ")")
    }
  )

What This Achieves: #

• Coefficients: Transforms 0.56 to .56
• P-values: Turns 0.023 into .023, and 0.0008 into <.001
• Confidence Intervals: Converts (0.12, 0.98) to (.12, .98)

Bonus: Apply to Descriptive Tables #

If you’re working with tbl_summary() for descriptive stats, you can reuse the same remove_leading_zero function to format continuous variables. For example:

tbl_summary(mtcars, include = c(mpg, wt)) %>%
  modify_fmt_fun(
    columns = stat_0,
    fun = function(x) remove_leading_zero(x)
  )

Quick Pro Tips #

• Ensure you have the latest version of gtsummary installed—some formatting functions get polished in updates.
• To save time across multiple tables, wrap all these modify_fmt_fun() calls into a custom function (e.g., add_apa_format()) that you can reuse with a single line.

内容的提问来源于stack exchange，提问作者veronicalm