Estimating and displaying population attributable fractions using the R package: graphPAF.

Area of Science:

• Epidemiology
• Biostatistics
• Public Health

Background:

• Population attributable fractions (PAF) are crucial for understanding the public health impact of risk factors.
• Existing methods for PAF estimation and visualization can be limited, especially in complex multi-risk factor scenarios.

Purpose of the Study:

• To introduce graphPAF, a novel R package for comprehensive estimation, inference, and display of PAF and impact fractions.
• To provide tools for advanced PAF analyses, including continuous exposures, pathway-specific fractions, and multi-risk factor scenarios.

Main Methods:

• Development of the graphPAF R package.
• Implementation of methods for standard and advanced PAF calculations.
• Integration of visualization tools such as fan-plots and nomograms.
• Application of Bayesian network approaches for complex multi-risk factor scenarios.

Main Results:

• graphPAF enables robust estimation and inference for various types of attributable fractions.
• The package facilitates clear visualization of attributable fractions, including over multiple risk factors.
• It supports calculations for continuous exposures and pathway-specific attributable fractions.
• Bayesian network methods are incorporated for joint, sequential, and average PAF in complex settings.

Conclusions:

• graphPAF is a comprehensive R package that enhances the estimation, inference, and display of population attributable fractions.
• The package provides valuable tools for epidemiologists and biostatisticians to assess the public health impact of risk factors.
• graphPAF serves as both a theoretical guide and a practical tutorial for attributable fraction analysis.