Statistical Methods for the Analysis of Food Composition Databases: A Review.

Area of Science:

• Nutrition Science
• Data Science
• Food Science

Background:

• Dietary guidelines and policy rely on understanding food-nutrient relationships.
• Analyzing food composition databases (FCDBs) requires appropriate statistical methods.
• Existing literature on statistical analysis of FCDBs needs synthesis.

Purpose of the Study:

• To review and collate statistical methods used to analyze FCDBs.
• To identify key trends in the application of statistical methods to FCDBs.
• To highlight research gaps in the statistical analysis of food composition data.

Main Methods:

• Systematic literature search yielding 24 relevant studies from 4238 references.
• Extraction of data on objectives, statistical methods, and results.
• Categorization of statistical methods and their applications.

Main Results:

• Statistical methods were primarily used for grouping similar food items (37.5%).
• Other applications included determining nutrient associations (25.0%), co-occurrence (20.8%), and temporal changes (16.7%).
• Standard statistical tests (33.3%) and clustering (29.1%) were the most common methods.

Conclusions:

• Statistical analysis of FCDBs is essential for evidence-based dietary recommendations.
• Current methods often focus on food item categorization and nutrient relationships.
• Future research should address the unique compositional nature of nutrient data and develop tailored statistical approaches.