Log-ratio transformations for dietary compositions: numerical and conceptual questions

Area of Science:

• Nutritional Epidemiology
• Biostatistics
• Dietary Data Analysis

Background:

• Traditional substitution models struggle to differentiate the impact of individual macronutrients from total energy intake.
• Dietary data are compositional, inherently expressing relative proportions that are best analyzed using ratios.

Purpose of the Study:

• To explore the application and interpretative implications of log-ratio transformations for analyzing dietary macronutrient composition.
• To assess the ability of log-ratio models to disentangle the effects of individual macronutrients and total energy intake on health outcomes.

Main Methods:

• Application of distinct log-ratio transformations to dietary macronutrient data.
• Use of regression models incorporating log-ratio transformed variables and total energy intake.
• Analysis of serum glucose levels as a health outcome using data from an Italian population-based study.

Main Results:

• Log-ratio transformations provide numerical and conceptual advantages over traditional substitution models.
• This approach allows for the separate evaluation of individual macronutrient effects and total energy intake.
• Demonstrated utility in analyzing the relationship between dietary composition and serum glucose levels.

Conclusions:

• Log-ratio transformations represent a robust statistical method for analyzing compositional dietary data.
• This methodology overcomes limitations of conventional models, offering clearer insights into nutrient-health relationships.
• The approach is valuable for nutritional epidemiology research, particularly for understanding metabolic health markers.