R ggplot2中geom_point与geom_errorbar对齐问题及双轴绘图需求

针对你遇到的点与误差棒不对齐、部分数据不显示问题，结合需求（OR对数刻度、cont/RD第二Y轴、形状/颜色映射），给出以下修复步骤：

1. 统一点与误差棒的位置对齐 #

点和误差棒错位核心原因是两者的位置调整参数不一致，需统一position_dodge的宽度：

# 定义固定的dodge宽度，确保点和误差棒错开距离一致
dodge_width <- 0.8

# 绘图时绑定x轴为结局，用dodge区分不同Method
ggplot(data, aes(x = outcome, y = est_transform, color = effect_type, shape = Method)) +
  # 误差棒先画，避免被点遮挡
  geom_errorbar(aes(ymin = LCI_transform, ymax = UCI_transform),
                position = position_dodge(width = dodge_width),
                width = 0.2, na.rm = TRUE) +
  geom_point(position = position_dodge(width = dodge_width),
             size = 3, na.rm = TRUE)

2. 处理双Y轴与OR对数刻度 #

OR需用对数刻度，cont/RD用线性刻度，必须先对OR类数据做对数转换，再通过sec.axis反转转换实现双刻度：

# 预处理数据：对OR的效应值、置信区间做对数转换
library(dplyr)
data <- data %>%
  mutate(
    est_transform = ifelse(effect_type == "OR", log(est), est),
    LCI_transform = ifelse(effect_type == "OR", log(LCI), LCI),
    UCI_transform = ifelse(effect_type == "OR", log(UCI), UCI)
  )

# 配置双Y轴
ggplot(data, aes(x = outcome, y = est_transform, color = effect_type, shape = Method)) +
  geom_errorbar(aes(ymin = LCI_transform, ymax = UCI_transform),
                position = position_dodge(width = dodge_width),
                width = 0.2, na.rm = TRUE) +
  geom_point(position = position_dodge(width = dodge_width),
             size = 3, na.rm = TRUE) +
  scale_y_continuous(
    name = "Continuous/RD 效应估计",
    # 第二Y轴：将对数转换后的值反转回原始OR
    sec.axis = sec_axis(trans = ~exp(.), name = "Odds Ratio (对数刻度)")
  )

3. 修复缺失数据（NA）不显示问题 #

部分数据不显示是因为原始数据缺少某些结局+Method的组合，需先构建完整的分组网格再合并数据：

# 生成所有结局、Method、effect_type的完整组合
full_grid <- expand.grid(
  outcome = unique(data$outcome),
  Method = unique(data$Method),
  effect_type = unique(data$effect_type)
)

# 合并原始数据，保留所有组合（缺失值为NA）
data_full <- full_join(full_grid, data, by = c("outcome", "Method", "effect_type"))

# 使用data_full绘图，NA对应的位置会自动留空

4. 完善映射与主题设置 #

添加颜色、形状的手动映射，优化图表可读性：

ggplot(data_full, aes(x = outcome, y = est_transform, color = effect_type, shape = Method)) +
  geom_errorbar(aes(ymin = LCI_transform, ymax = UCI_transform),
                position = position_dodge(width = dodge_width),
                width = 0.2, na.rm = TRUE) +
  geom_point(position = position_dodge(width = dodge_width),
             size = 3, na.rm = TRUE) +
  scale_y_continuous(
    name = "Continuous/RD 效应估计",
    sec.axis = sec_axis(trans = ~exp(.), name = "Odds Ratio (对数刻度)")
  ) +
  # 手动设置颜色与形状
  scale_color_manual(values = c("OR" = "#E63946", "cont" = "#457B9D", "RD" = "#1D3557")) +
  scale_shape_manual(values = c("Method1" = 16, "Method2" = 17, "Method3" = 15, "Method4" = 18, "Method5" = 8)) +
  theme_bw() +
  labs(x = "结局", color = "效应类型", shape = "方法") +
  # 若结局较多，用分面避免x轴拥挤
  facet_wrap(~outcome, scales = "free_x")

内容的提问来源于stack exchange，提问作者user5481267