A heuristic approach to the formulas for population attributable fraction

Area of Science:

• Epidemiology
• Biostatistics

Background:

• Standard epidemiology texts often focus on Levin's formula for dichotomous exposures, neglecting Miettinen's formula and multi-level exposure scenarios.
• This leads to limited understanding and confidence among health researchers and practitioners regarding population attributable fraction (PAF) calculations.
• Many are unaware of or struggle with PAF formulas for complex exposure distributions and confounding.

Purpose of the Study:

• To provide a clear, heuristic, and visual method for understanding and interconnecting the Levin and Miettinen formulas for population attributable fraction estimation.
• To address the common limitations in epidemiological resources concerning multi-level exposures and confounding factors in PAF calculations.
• To enhance the practical application of PAF estimation in public health research.

Main Methods:

• A heuristic approach combined with pictorial representations to explain the underlying structures of the Levin and Miettinen formulas.
• Illustrations demonstrating correct methods for handling multiple exposure levels.
• Presentation of correct and incorrect approaches for aggregating estimates over strata of confounding factors.

Main Results:

• The study offers a visual and intuitive method to understand the relationship between Levin's and Miettinen's formulas.
• It clarifies how to correctly manage situations with multiple exposure levels, highlighting flaws in commonly used methods.
• Guidance is provided on appropriate strategies for dealing with confounding factors in PAF estimation.

Conclusions:

• The developed approach enhances understanding and correct application of population attributable fraction formulas, especially in complex scenarios.
• Health researchers and practitioners can gain confidence in using PAF estimation with multi-level exposures and confounders.
• This work aims to bridge the gap between theoretical PAF formulas and their practical implementation in public health.