10.4 Example: Visualising Lung Cancer Deaths by Prevalence and Age in Germany

## # A tibble: 6 × 8
##   age   sex    prevalence prev_upper prev_lower    dx dx_uppe dx_lower
##   <chr> <chr>       <dbl>      <dbl>      <dbl> <dbl>   <dbl>    <dbl>
## 1 10-14 male         0.08       0.13       0.05 0.322   0.461    0.217
## 2 10-14 female       0.18       0.32       0.09 0.457   0.761    0.248
## 3 10-14 both         0.13       0.22       0.07 0.779   1.21     0.468
## 4 15-19 male         0.48       0.77       0.29 1.27    1.75     0.916
## 5 15-19 female       0.9        1.52       0.5  1.56    2.46     0.941
## 6 15-19 both         0.68       1.02       0.44 2.83    3.88     2.07

Scatter plot

library(ggplot2)

lung_scatter <- hmsidwR::germany_lungc |>
  
  # Initialise ggplot with arguments to set the x and y 
  # aesthetics
  ggplot(aes(x = prevalence, y = dx)) +
  
  # Create a scatter plot, using sex to modify the shape 
  # aesthetic
  geom_point(aes(shape = sex)) + 
  
  # Set the title and names for x and y axes
  labs(title = "Lung Cancer Deaths by Prevalence in Germany",
       x = "Prevalence",
       y = "Deaths")

lung_scatter

A scatterplot of lung cancer deaths in Germany, showing estimated prevalence rate on the x-axis and estimated death rate on the y-axis. There are three categorical variables of sex: both, female and male, each with different shapes. The plot shows some points from the 'both' category with a higher estimated prevalence and death rate. Males have a higher prevalence and death rate than females.

Things to do:

  1. Change the size of the points with the size aesthetic within geom_point.

  2. Change the colour of the points with the colour argument within geom_point.

  3. Remove the grey background with theme:

    • theme_classic() or theme_bw() with extra parameters do the same job.

  4. Check the readability of colours for people with colour vision deficiency.

hmsidwR::germany_lungc |>
  ggplot(
    aes(
      x = prevalence,
      y = dx,
      shape = sex,
      colour = sex)) +
  geom_point(size = 4) + 
  labs(title = "Lung Cancer Deaths by Prevalence in Germany",
       x = "Prevalence",
       y = "Deaths") +
  theme_classic() +
  theme(text = element_text(family = "Lato",
                            size = 14),
        axis.text = element_text(colour = "black"))

A scatterplot of lung cancer deaths in Germany, showing estimated prevalence rate on the x-axis and estimated death rate on the y-axis. There are three categorical variables of sex: both, female and male. The points are larger than in the previous plot and are different colours.

hmsidwR::germany_lungc |>
  ggplot(
    aes(
      x = prevalence,
      y = dx,
      shape = sex,
      colour = sex)) +
  geom_point(size = 4) + 
  labs(title = "Lung Cancer Deaths by Prevalence in Germany",
       x = "Prevalence",
       y = "Deaths") +
  theme_bw() +
  theme(axis.line = element_line(colour = "black"),
        plot.background = element_blank(),
        panel.grid.major = element_blank(),
        panel.grid.minor = element_blank(),
        panel.border = element_blank(),
        text = element_text(family = "Lato",
                            size = 14),
        axis.text = element_text(colour = "black"))

A scatterplot of lung cancer deaths in Germany, showing estimated prevalence rate on the x-axis and estimated death rate on the y-axis. There are three categorical variables of sex: both, female and male. The plot uses a lightly different theme with additional parameters to create the same features as the previous plot.

It’s ok, but maybe not for people who are colourblind.

How to check?

Four scatterplots of lung cancer deaths in Germany, showing estimated prevalence rate on the x-axis and estimated death rate on the y-axis. The four plots simulate colour vision deficiency to demonstrate how the colours appear to people with red, green or blue colour blindness, and monochromatic vision.

Possible options:

  1. Don’t use colour.

  2. Put black outlines around the shapes.

  3. Choose a better palette.

  • The Okabe-Ito palette (below) was designed for qualitative data.

  • The viridis palette may be suitable for continuous data.

  • {colorbrewer} package has colourblind-friendly palettes.

Okabe-Ito colour chart showing the colours and their hex codes. The codes are: Black #000000; Orange #E69F00; Sky blue #56B4E9; Bluish green #009E73; Yellow #F0E442; Blue #0072B2; Vermillion #D55E00; Reddish purple #CC79A7; Grey #999999.

  • Control colour and shape using scale_colour_manual and scale_shape_manual (necessary if modifying the legend title).
scatter +
  labs(title = "Lung Cancer Deaths by Prevalence in Germany") +
  scale_color_manual(name = "Sex",
                     values = c("#000000", "#009E73", "#D55E00"),
                     labels = c("Both", "Female", "Male")) +
  scale_shape_manual(name = "Sex",
                     values = c(16, 17, 15),
                     labels = c("Both", "Female", "Male"))

A scatterplot of lung cancer deaths in Germany, showing estimated prevalence rate on the x-axis and estimated death rate on the y-axis. The colours of the points have been changed to demonstrate how this is done with the scale_colour_manual function.

  • Can also use scale_color_OkabeIto and scale_fill_OkabeIto.

Barplot

library(ggpattern)

# Create a barplot
hmsidwR::germany_lungc |>
    
  # This time, the aesthetics are age and estimated death rate 
  # with patterns for sex
  ggplot(aes(x = age, y = dx, group = sex))  +
  
  # The ggpattern package creates stacked charts with 
  # geom_col_pattern different patterns based on the sex 
  # variable
  ggpattern::geom_col_pattern(aes(pattern=sex),
                              position="stack",
                              fill= 'white',
                              colour= 'black') +
  
  # Again, set the title and names for x and y axes
  labs(title = "Lung Cancer Deaths by Age in Germany",
       x = "Age",
       y = "Deaths")

A stacked barplot of lung cancer deaths in Germany, showing age categories from 10 years to 85+ in 5-year increments on the x-axis, and estimated death rate on the y-axis. There are three categorical variables of sex: both, female and male. The bars have patterns based on the sex category of 'both', 'female', or 'male'. Deaths from lung cancer increase with age from 30-34 to 75-79, then start to decline.

Things to do:

  1. Create a bar chart with geom_bar.

  2. Use faceting to separate the categories with facet_grid.

  3. Angle the x-axis category labels with hjust.

  4. Set the y-axis origin to 0.

# Create a barplot
barplot <- hmsidwR::germany_lungc |>
    
  # This time, the aesthetics are age and number of deaths with 
  # patterns for sex
  ggplot(aes(x = age, y = dx))  +
  
  # The geom_bar argument is used for bar charts
  geom_bar(fill = "white",
           
           # stat = "identity" tells R to calculate the sum of 
           # the y variable, grouped by the x variable
           stat = "identity",
           
           # Add a black outline
           colour = "black") +
  
  # Set the origin for the y-axis at 0
  scale_y_continuous(expand = c(0, 0)) +
  
  # Again, set the title and names for x and y axes
  labs(title = "Lung Cancer Deaths by Age in Germany",
       x = "Age",
       y = "Deaths") +
  
  # Separate panels for each sex category, and modify the 
  # fonts and some styling
  facet_grid(sex ~ .) +
  theme_classic() +
  theme(text = element_text(family = "Lato",
                            size = 12),
        axis.text = element_text(colour = "black"),
        axis.text.x = element_text(angle = 45,
                                   hjust = 1))

barplot

A facted grid of barplots of lung cancer deaths in Germany, showing age categories from 10 years to 85+ in 5-year increments on the x-axis, and estimated death rate on the y-axis. The barplots are arranged in three separate panels stacked on top of one another.There are three categorical variables of sex: both, female and male. Deaths from lung cancer increase with age from 30-34 to 75-79, then start to decline. Female deaths are lower than 'male' and 'both'.

To format the strip text in the facets, use the labeller argument inside facet_grid:

# Put the categories in a list
sex_categories <- list(
  "both" = "Both",
  "female" = "Female",
  "male" = "Male"
)

# Create a function to store the new names
sex_labeller <- function(variable,value){
  return(sex_categories[value])
}

# Add the names to the strip panels
barplot +
  facet_grid(~ sex, labeller = sex_labeller) +
  theme_classic() +
  theme(text = element_text(size = 10),
        axis.text = element_text(colour = "black"),
        axis.text.x = element_text(angle = 45,
                                   hjust = 1))

A facted grid of barplots of lung cancer deaths in Germany, showing age categories from 10 years to 85+ in 5-year increments on the x-axis, and estimated death rate on the y-axis. The barplots are arranged in three separate panels in a row. This demonstrates a different way to use the facet_grid function and changes the sex category labels to capitalise the first letter of each label.

Line plot

lineplot <- hmsidwR::germany_lungc |>
  
  # Variables are age, deaths, grouped by sex
  ggplot(aes(x = age, y = dx, group = sex)) +
  
  # geom_line creates the lines, varying the type of line by sex
  geom_line(aes(linetype = sex)) +
  
  # Add points with geom_point
  geom_point() +
  labs(title = "Lung Cancer Deaths by Age in Germany",
       x = "Age",
       y = "Deaths")

lineplot

A lineplot showing lung cancer deaths in Germany, showing age categories from 10 years to 85+ in 5-year increments on the x-axis, and estimated death rate on the y-axis. The sex category variable of 'both', 'female' and 'male' is shown as three separate lines of with different line styles.

Lineplot

Things to do:

  1. Remove the legend and add labels to the ends of the lines with package {ggtext}.

  2. Plain background and similar formatting as before.

library(ggtext)

# Subset the data to create labels based on the oldest age 
# category to put the labels at the end of the line
dat_label <- subset(hmsidwR::germany_lungc, age == max(age))

hmsidwR::germany_lungc |>
  
  # Same variables as before
  ggplot(aes(x = age, y = dx, group = sex)) +
  
  # geom_line creates the lines, varying the type of line by 
  # sex; remove the legend
  geom_line(aes(linetype = sex),
            show.legend = FALSE) +
  
  # Use the subsetted data created to add the labels
  ggtext::geom_richtext(data = dat_label,
                        label.color = NA,
                        aes(label = sex)) +
  labs(title = "Lung Cancer Deaths by Age in Germany",
       x = "Age",
       y = "Deaths") +
  theme_classic() +
  theme(text = element_text(family = "Lato",
                            size = 14),
        axis.text = element_text(colour = "black"),
        axis.text.x = element_text(angle = 45,
                                   hjust = 1))

A lineplot showing lung cancer deaths in Germany, showing age categories from 10 years to 85+ in 5-year increments on the x-axis, and estimated death rate on the y-axis. The sex category variable of 'both', 'female' and 'male' is shown as three separate lines of with different line styles. The legend has been removed and the sex category labels are at the end of each line. This is to demonstrate the ggtext package.