Comparing the Value of Data Visualization Methods for Communicating Harms in Clinical Trials

Area of Science:

• Clinical Trials
• Data Visualization
• Pharmacovigilance

Background:

• Adverse events in clinical trials are typically reported by frequency counts, omitting crucial data dimensions like severity, seriousness, and timing.
• Current reporting methods fail to capture the full spectrum of harm information, hindering comprehensive stakeholder understanding.
• Existing reporting practices often exclude many observed harms due to selection criteria.

Purpose of the Study:

• To evaluate and compare six distinct data visualization techniques for reporting clinical trial harms.
• To assess the utility of visualizations in communicating multidimensional adverse event data.
• To identify optimal visualization methods favored by content experts for improved harm communication.

Main Methods:

• Replication and comparison of six harm visualization approaches: dot plot, stacked bar chart, volcano plot, heat map, treemap, and tendril plot.
• Utilized individual participant data from a gabapentin neuropathic pain randomized trial for binary event analysis.
• Assessed visualization value through a heuristic approach and expert content review, with figures generated using R.

Main Results:

• Visualizations can present multiple dimensions of harms, unlike frequency counts alone.
• Dot plots and volcano plots emerged as favored methods for summarizing overall harms data.
• Most visualization methods, except the tendril plot, do not necessitate individual participant data.

Conclusions:

• Data visualization significantly enhances the communication of multidimensional clinical trial harms.
• Dot plots and volcano plots are recommended for their effectiveness in summarizing and conveying adverse event information.
• Open-source R code is provided to facilitate the implementation of these visualization techniques by trialists.