Dynamite plots in surgical research over 10 years: a meta-study using machine-learning analysis.

Area of Science:

• Medical Research
• Surgical Fields
• Data Visualization

Background:

• Dynamite plots, a type of bar chart, are often misleading due to within-the-bar bias and lack of distributional information.
• The machine learning tool Barzooka aids in screening graph types in scientific literature.

Purpose of the Study:

• To determine the prevalence of dynamite plot usage in original research articles across nine surgical fields.
• To analyze trends in the use of dynamite plots over time.

Main Methods:

• Over 40,000 original research articles from nine surgical fields were sampled and analyzed using Barzooka.
• Trends in dynamite plot usage were assessed using Kendall τ coefficient and the Mann-Kendall test.

Main Results:

• A statistically significant decrease in dynamite plot use was observed in eight of nine surgical fields over a decade.
• In 2022, dynamite plot usage ranged from approximately 30% to 70% depending on the surgical field.

Conclusions:

• While the use of dynamite plots in surgical research has declined, it remains prevalent.
• Further research and education are needed to improve data visualization practices among surgical researchers.