Skip to contents

Adverse Events Tornado Plot

Source: vignettes/tornado_plot.Rmd
tornado_plot.Rmd

Introduction

This article describes creating the Adverse Events Tornado Plot for Oncology Studies. It can be used for the reports adae_tornado.tot; as well as for CDISC-recommended Tornado Plot. Examples are created using in-built ADAE and ADSL dataset.

Programming Flow

Read in Data

To start, load required libraries and the in-built example adae and adsl dataset.

Process Data with Subsets and Variables

Before creating Adverse Events tornado plot, it’s essential to use the process_tornado_data(). This function helps prepare the data for visualization, making it a crucial step. With this function, users can customize many parts of the plot to match their specific data and analysis needs.
While it is recommended that this be used for Oncology data, it can also work with non-Therapeutic Area specific ADaM datasets, which is shown in this example.
This function block consists mainly adsl_merge(), ae_pre_processor() and mentry() functions to combine ADSL variables, perform adverse-event specific pre-processing and filter records/set grouping variables. It also filters out the maximum severity based on the given inputs. Below, we’ll explain the parameters used in process_tornado_data:-

  • dataset_adsl: This parameter is used to pass the variable containing ADSL dataset.

  • dataset_analysis: This parameter is used to pass the variable containing analysis dataset. In this example, ADAE dataset.

  • adsl_subset: This parameter specifies subset condition(s) for the ADSL dataset. It allows the user to filter data before plotting.

  • analysis_subset: Similar to adsl_subset, this parameter specifies subset condition(s) for the analysis dataset.

  • split_by: This parameter is used for splitting the data by a specific variable. The default value is NA_character_ which suggests that by default, there’s no splitting. If a variable is provided, it would split the data accordingly. Example: Region - separate plots would be generated for each region.

  • trtvar: This parameter represents the treatment variable. It indicates the variable that represents different treatments or groups in the data.

  • trt_left: This parameter represents the treatment value. It indicates the value that displays on the left hand side of the plot.

  • trt_right: This parameter represents the treatment value. It indicates the value that displays on the right hand side of the plot.

  • trtsort: This parameter is used for sorting the treatment variable.

  • obs_residual: This parameter is used to pass a period (numeric) to extend the observation period. If passed as NA, overall study duration is considered for analysis. eg. if 5, only events occurring upto 5 days past the TRTEDT are considered..

  • fmq_data: This parameter is required if FMQ terms are to be used as high or low terms, the FMQ_Consolidated_List should be passed. Default value is NULL.

  • ae_catvar: This parameter is used for series categorical variable for severity analysis and filter out the maximum severity.

  • denom_subset: This parameter is a condition which is used to subset data set for calculating denominator. Default value is NA.

  • pctdisp: This parameter is a method that is used to calculate denominator (for %). Valid values: "TRT", "VAR", "COL", "SUBGRP", "CAT", "NONE", "NO", "DPTVAR", "BYVARxyN".

  • legendbign: This parameter determines whether N should be displayed in the legend. The default value is Y which suggests that by default, N is displayed in legend. Default value is N.

  • yvar: This parameter represents the variable to be plotted on the y-axis. The default value is AESOC.

NOTE: Default/Example values are provided to give users a starting point. However, these values can be modified as necessary to better suit the specific data and visualization objectives.

The first step for creating tornado_plot for Adverse Events is to read the required datasets into the environment. In this example, ADSL and ADAE are used.

Below is an example of calling process_tornado_data function for preprocessing the data for tornado plot.

df <- process_tornado_data(
  dataset_adsl = adsl,
  dataset_analysis = adae,
  adsl_subset = "SAFFL == 'Y'",
  analysis_subset = "TRTEMFL == 'Y'",
  obs_residual = "30",
  fmq_data = NA,
  ae_catvar = "AESEV/AESEVN",
  trtvar = "ARM",
  trt_left = "Xanomeline High Dose",
  trt_right = "Xanomeline Low Dose",
  pctdisp = "TRT",
  denom_subset = NA_character_,
  legendbign = "N",
  yvar = "AESOC"
)

The process_tornado_data returns data.frame which gets stored in df above. A pre-processed dataset which is the input for Oncology tornado plot.

df

Tornado Plot

The tornado_plot function is designed to create flipped bar plots from input data, with customizable options. It offers flexibility in plotting data and allows users to tailor the plot aesthetics to their preferences. The parameters used in tornado_plot are explained below:

  • datain: The input data for plotting. The output data from process_tornado_data function.

  • trt_left_label: This parameter is used to label the trt_left in the plot.

  • trt_right_label: This parameter is used to label the trt_right in the plot..

  • bar_width: Bar width option. This parameter allows customization of the bar width displayed on the plot.

  • axis_opts: List of options for axis appearance. This parameter allows customization of the appearance of the x-axis and y-axis, including labels, and other options retrieved from plot_axis_opts().

  • legend_opts: List of options for legend appearance. This parameter allows customization of the legend displayed on the plot, including its label, position, and direction.

  • series_opts: Vector of color options with named ae_catvar variable for the severity analysis.

  • griddisplay: Grid display option. This parameter controls whether grid lines are displayed on the plot. Set it to “N” to disable grid lines, and “Y” to enable them.

Below is an example of how oncology tornado plot is called using the defined parameters:

series_opts <- g_seriescol(df, "red~yellow~blue", "BYVAR1")

plot <- tornado_plot(
  df,
  trt_left_label = "DRUG B",
  trt_right_label = "DRUG C",
  bar_width = 0.5,
  axis_opts = plot_axis_opts(
    xaxis_label = "Primary System Organ Class",
    yaxis_label = "% of Subjects",
    ylinearopts = list(
      breaks = seq(-100, 100, 10),
      labels = c(seq(100, 0, -10), seq(10, 100, 10))
    )
  ),
  legend_opts = list(
    label = "Severity",
    pos = c(0.2, 0.15),
    dir = "vertical"
  ),
  series_opts = series_opts,
  griddisplay = "N"
)
print(plot)