Chapter 9 Creating tables

In our final document we will want tables to describe our data. Below we create a table which contains demographic information on our sample.

For most tables you can use the examples shown below with Rmisc and rstatix. However for more complex tables we need to use a combination of gtsummary and flextable.

To this point I have emphasized that using spaces in column names is bad practice, however when we get the final form of our table, we may want to insert spaces so that they are properly displayed in our manuscript. I use the janitor package to make this switch as easy as possible.

tbl.demo %>% 
  janitor::clean_names(case = "title")

9.1 Simple Tables

# Method 1 using rstatix
tbl.demo  <- df.nirs %>%
  distinct(Subject, .keep_all = TRUE) %>%
  group_by(group, gender) %>%
  rstatix::get_summary_stats(age, type = "mean_sd") %>%
  select(-c("variable")) %>%
  dplyr::rename(
    "Group" = "group",
    "Sex" = "gender",
    "Mean Age" = "mean",
    "SD" = "sd"
  )

# Method 2 using RMisc

journey_time <- Rmisc::summarySE(data = df,
                                      measurevar = "journey_time_avg",
                                      groupvars = c("service","year"),
                                      conf.interval = 0.95,
                                      na.rm = TRUE,
                                      .drop = TRUE)

# Taking the mean and confidence interval =============
    # __Method #1 - Easy #########
      journey_time <- Rmisc::summarySE(data = df,
                                       measurevar = "journey_time_avg",
                                       groupvars = c("service","year"),
                                       conf.interval = 0.95,
                                       na.rm = TRUE,
                                       .drop = TRUE)
                                       
    # To get a descriptive table

  tbl.descFullACAP <- Rmisc::summarySE(data = df.mri %>%
                           filter(metric == "FA", mriloc == "CC_FMajor"),
                         measurevar = "age",
                         groupvars = c("group","gender"),
                         conf.interval = 0.95,
                         na.rm = TRUE,
                         .drop = TRUE) %>%
          janitor::clean_names("upper_camel") %>%
          rename(
            "CI" = "Ci",
            "SE" = "Se",
            "SD" = "Sd"
          ) %>%
            mutate_at(vars(Age,SD,SE,CI), funs(round(., 3)))
    
              cgwtools::resave(tbl.descFullACAP, file = "data/tables.RData") #resave a list of tables that I'll use in the .Rmd file.

    # __Method 2 - More options
    df %>%
        group_by(Channel) %>%
        summarise_each(funs(mean, sd))
        
        
    dt <- data.table(df)
    group_by(dt, Channel_46)
    # __Method 2b -  Not so Easy ##################

        # descriptives <- demo%>%dplyr::group_by(AthleteType)%>%
        #   dplyr::summarise(
        #     "Mean Height (cm)" = round(mean(Height_cm),2)
        #     , "Mean Weight (kg)" = round(mean(Weight_kg),2)
        #     , "Mean Age (Years)" = round(mean(AgeInYears),2)
        #     , "Number of Concussions" = round(mean(NumPriorConc),2)
        #     , "Number of Diagnosed Concussions" = round(mean(NumDiagConc),2)
        #     , "Number of Undiagnosed Concussions" = round(mean(NumUndiagConc),2)
        #   )

If you have only categorical variables than you can use the tabyl function in the janitor package.

# Example 1 using ToothGrowth dataset
tbl <- ToothGrowth %>% 
  janitor::tabyl(supp)
 
# Example 2 using mtcars dataset
tbl <- mtcars %>%
    janitor::tabyl(cyl,am) %>%
      janitor::adorn_percentages("col") %>%
  janitor::adorn_pct_formatting(digits = 1)

9.2 Rounding numbers

In many cases, you might want to remove leading and/or trailing zeros. There are a few packages which claim to do this. Normally I will use the following piece of code

df %>%
        mutate(across(where(is.numeric), round, 4))

However, alternatives are the numform and snip functions.

9.3 Saving Tables to xlsx

You may also choose to write these to an xlsx file, you can use the xlsx package (which does have the benefit of letting you append) however I prefer using rio for all my import/export tasks. It functions using list which is a bit different.

# using xlsx package
df <- iris
df1 <- mtcars
rio::export(df, "raw/test.xlsx") # uses the rio package

rio::export_list(list(df, df1), "raw/multidf.xlsx") # uses the rio package

# using xlsx package
df <- iris
xlsx::write.xlsx2(df, "test.xlsx", row.names = FALSE, sheetName = "SheetNumber1", append = TRUE) # uses the xlsx package

9.4 Complex Tables

For more complex tables you may want to check into gtsummary which was a lifesaver for me on a recent manuscript. It has a ton of custom options you can mess with to get exactly what you are looking for. Objects from gtsummary are saved as flextables

pacman::p_load_gh("ddsjoberg/gtsummary")
pacman::p_load(flextable)

# Andrew's example from a recent manuscript (microstates paper)
tbl.demo2 <- tbl.demo %>% 
  janitor::clean_names(case = "title") %>% 
  dplyr::rename(
    "Weight (kg)" = "Weight Kg",
    "Height (m)" = "Ht m",
    "BMI" = "Bmi",
    "Age (years)" = "Age",
    "Was nitrous used?" = "Nitrous"
  ) %>% 
  gtsummary::tbl_summary(by = `Anesthetic Maintenance`) %>% 
  add_n() %>% 
  add_overall() %>%
  modify_spanning_header(starts_with("stat") ~ "**Anesthetic Received**") %>% 
  bold_labels()

Below is a reprex example with gtsummary from their documentation

pacman::p_load(gtsummary)

trial2 <- trial %>% select(trt, age, grade)
trial2 %>% tbl_summary()

trial2 %>% tbl_summary(by = trt) %>% add_p()

trial2 %>%
  tbl_summary(by = trt) %>%
  add_p(pvalue_fun = ~style_pvalue(.x, digits = 2)) %>%
  add_overall() %>%
  add_n() %>%
  modify_header(label ~ "**Variable**") %>%
  modify_spanning_header(c("stat_1", "stat_2") ~ "**Treatment Received**") %>%
  modify_footnote(
    all_stat_cols() ~ "Median (IQR) or Frequency (%)"
  ) %>%
  modify_caption("**Table 1. Patient Characteristics**") %>%
  bold_labels()

tbl <- trial2 %>%
  tbl_summary(by = trt) %>%
  add_p(pvalue_fun = ~style_pvalue(.x, digits = 2)) %>%
  add_overall() %>%
  add_n() %>%
  modify_header(label ~ "**Variable**") %>%
  modify_spanning_header(c("stat_1", "stat_2") ~ "**Treatment Received**") %>%
  modify_footnote(
    all_stat_cols() ~ "Median (IQR) or Frequency (%)"
  ) %>%
  modify_caption("**Table 1. Patient Characteristics**") %>%
  bold_labels()

9.5 Supplementary Material

Item Analysis