Combining two different groups into a single bar on a bar plot

[sjelacic74]

I have 3 different groups of patients in my data frame. I would like to create a bar plot with 4 bars (3 bars representing each group and 4th bar representing groups 2 and 3 combined).

The code I am using for my bar plot is the following which allows me to plot the first 3 bars representing each group.

ggbarplot(
  MVdata, x = "fgroupplot", y = "MVtimeft", fill = "fgroupplot",
  add = c("mean_se"), width = 0.5,
  xlab = FALSE,
  ylab = "Proportion of Patients\n Extubated within 6 Hours",
  yticks.by = 0.25
) +
  geom_bracket(
    xmin = c("Pre-intervention", "FT Protocol Non-compliant", "Pre-intervention"), xmax = c("FT Protocol Compliant", "FT Protocol Compliant", "FT Protocol Non-compliant"),
    y.position = c(0.8, 0.9, 1), label = c("p<0.001", "p<0.001", "ns"),
    tip.length = 0.02
) +
  scale_y_continuous(labels = scales::percent) +
  scale_x_discrete(labels= c("Pre-intervention"="Pre\n Fast-Track Eligible","FT Protocol Non-compliant"="Post TOF Response\n Not Measured", "FT Protocol Compliant"="Post TOF\n Response Measured")) +
  scale_fill_grey() + 
  theme(legend.position='none') +
  theme(axis.text.x = element_text(size = 8))

This is my data frame:

Thanks.
Srdjan

[scottyd22]

Welcome to the community @sjelacic74! Below is one way to combine two groups into a single bar. Since only a portion of the data was provided, I first duplicated the pre-intervention data twice to create (what I'm assuming are) groups 2 and 3. This resulted in a data frame with 6 observations for each of the 3 groups. Then, I created a new category (2 & 3 combined) by binding these two groups to MVdata and renaming their fgroupplot name. This added 12 observations for 2 & 3 combined (as expected). Finally, the plot was generated, adding a new x label for the combined category (other plot features may need to be updated).

library(tidyverse)
library(ggpubr)

# sample data provided
MVdata = data.frame(
  Mvtime = c(308, 191, 537, 173, 16, 188),
  Reintubation = c(0,0,0,0,1,0),
  NIPPV = c(0,0,1,0,0,0),
  twitch = c(0,1,0,0,0,0),
  group = 0,
  MVtimeft = c(1,1,0,1,1,1),
  fgroup = 'Pre-intervention group',
  fgroupplot = 'Pre-intervention'
) %>%
  # add the exact same rows with new fgroupplot names (more sample data)
  bind_rows(
    (.) %>% mutate(fgroupplot = 'FT Protocol Non-compliant'),
    (.) %>% mutate(fgroupplot = 'FT Protocol Compliant')
  )

count(MVdata, fgroupplot)
#>                  fgroupplot n
#> 1     FT Protocol Compliant 6
#> 2 FT Protocol Non-compliant 6
#> 3          Pre-intervention 6

# add a "new" category (2 & 3 combined)
# by filtering to groups 2 & 3 and then 
# changing the fgroupplot name
MVdata = MVdata %>%
  bind_rows(
    (.) %>%
      filter(fgroupplot %in% c('FT Protocol Compliant', 'FT Protocol Non-compliant')) %>%
      mutate(fgroupplot = '2 & 3 combined')
    )

count(MVdata, fgroupplot)
#>                  fgroupplot  n
#> 1            2 & 3 combined 12
#> 2     FT Protocol Compliant  6
#> 3 FT Protocol Non-compliant  6
#> 4          Pre-intervention  6

# generate plot
ggbarplot(
  MVdata, x = "fgroupplot", y = "MVtimeft", fill = "fgroupplot",
  add = c("mean_se"), width = 0.5,
  xlab = FALSE,
  ylab = "Proportion of Patients\n Extubated within 6 Hours",
  yticks.by = 0.25
) +
  geom_bracket(
    xmin = c("Pre-intervention", "FT Protocol Non-compliant", "Pre-intervention"), 
    xmax = c("FT Protocol Compliant", "FT Protocol Compliant", "FT Protocol Non-compliant"),
    y.position = c(0.8, 0.9, 1), label = c("p<0.001", "p<0.001", "ns"),
    tip.length = 0.02
  ) +
  scale_y_continuous(labels = scales::percent) +
  scale_x_discrete(labels= c("Pre-intervention"="Pre\n Fast-Track Eligible",
                             "FT Protocol Non-compliant"="Post TOF Response\n Not Measured", 
                             "FT Protocol Compliant"="Post TOF\n Response Measured",
                             '2 & 3 combined' = "This is my new label"
                             )
                   ) +
  scale_fill_grey() + 
  theme(legend.position='none') +
  theme(axis.text.x = element_text(size = 8))

The plot looks a little "funny" because all of the data is duplicated. Once the actual data is used, I think this will get to your intended result.

Created on 2022-12-24 with reprex v2.0.2.9000