Table 13 Summary of graphic format to display complications collected in trials
VisualizationData format for creationCharacteristics presentedProsCons
Bar Chart
Unique harms are arranged along
x-axis in an order of the authors’ choice (e.g., descending or ascending effect size, alphabetical).
Aggregate data
  • Occurrence of each harm by trial arm (y-axis)

  • Distribution of severity for each harm (colour distribution in each bar)

  • Easy to understand

  • Moderate value for communicating harms as rated by experts

  • Figure becomes wide with many events (each unique harm gets a column)

  • Limited information presented

Dot Plot
Unique harms are arranged along the y-axis in an order of the authors’ choice.
Aggregate data
  • Comparative measure of effect (x-axis on the right panel of the figure)

  • Uncertainty (95% CI around points on the right panel of the figure)

  • Incidence/occurrence of each harm by trial arm (x-axis on the left panel of the figure)

Note: Additional “panels” of information can be added to present data on other characteristics (e.g., the distribution of severity ratings for each unique harm)
  • Easy to understand

  • Expandable format using a panel approach

  • High value for communicating harms as rated by experts

  • Figure becomes long with many events (each unique harm gets a row)

Heatmap
Unique harms are arranged along the y-axis in an order of the authors’ choice.
Aggregate data
  • Standardized comparative measure of effect (colour)

  • Subgroups of participants experiencing each harm (x-axis)

Note: Choose whichever subgroups are desired for exploration (e.g., male/female, young/
old, low/high dose, severe/not-severe, etc.) and restrict to events in that subgroup before calculating the standardized measure of effect.
  • Allows exploration of harms within and across subgroups

  • Moderate value for communicating harms as rated by experts

  • Difficult to understand

  • Figure becomes long with many events (each unique harm gets a row)

  • Overwhelming with too many events and subgroups

Volcano Plot
Unique harms are represented by bubbles placed on an X-Y grid.
Aggregate data
  • Comparative measure of effect (x-axis)

  • Overall occurrence of each harm (size of bubble)

  • P value (y-axis and opacity of bubble)

  • Trial arm positively associated with each harm (colour)

Note: ‘Colour’ and ‘opacity’ of bubbles could be used to represent other dimensions
(e.g., seriousness or whether the harms are recurrent) as their characteristics are already represented by the harms’ positions on the x and y-axis, respectively.
  • Can present up to five dimensions of data in a relatively condensed space

  • High value for communicating harms as rated by experts

  • Events with the same data (i.e., counts in each arm) will occupy the same space

Tendril Plot
Unique harms are represented by distinct tendrils which are created by adding vectors created by
the connection of two instances/events for a unique harm in a coordinate space.
Individual participant data
  • Overall occurrence of each harm (size of each harm’s points)

  • P value (colour of each harm’s points)

  • Duration between each event (length of each vector)

  • Trial arm that had the event (direction of the vector)

  • Detailed visual for the timing of every event allows users to see which harms occur more often soon after exposure and which occur later or equally over time.

  • More useful for a few unique harms of interest

  • Very difficult to understand

  • Low value for communicating many harms as rated by experts