Charting Public Health: A Taxonomic Study of Visualization Practices in the Public Health Field

Public health organizations use data visualizations to communicate critical information to the public and policymakers. Despite their importance, the actual design practices used in these high-stakes settings have remained largely unexamined. This study addresses this gap by constructing and analyzing a large-scale corpus of over 4,000 real-world data visualizations collected from more than two dozen U.S. and international public health organizations.

The researchers developed a semi-automated pipeline to scrape, filter, and categorize visualizations from public health websites. Using a combination of optical character recognition (OCR) and large language models (LLMs), they identified key design characteristics, including chart types, structural conventions, and accessibility features. The study also categorized organizations into distinct communication styles based on their recurring design choices, such as their use of embellishments and annotation practices.

The analysis reveals that public health communication is dominated by a narrow set of chart types, with line charts being the most prevalent. While organizations generally follow standard structural conventions, there are persistent design flaws. Specifically, nearly 23% of the visualizations contain color combinations that pose accessibility risks for individuals with color vision deficiencies. Furthermore, the researchers identified five distinct communication styles—ranging from evidence-forward reporting to imaging-heavy storytelling—suggesting that institutional norms and specific communication goals significantly shape how health data is presented.

This work provides a foundational taxonomy for understanding how public health organizations operationalize data communication under real-world constraints. By identifying common vulnerabilities and design profiles, the study offers actionable insights for public health officials to improve the accessibility and clarity of their communications. The released dataset and corpus serve as a valuable resource for future research into the effectiveness and equity of data visualization in public-facing roles.