Nutritional Variation Among Foods Included in an Ecuadorian Food Composition Database Using Multivariate Analysis.

Background/

Objectives

This study characterized compositional variation among 996 foods and beverages retained from the Universidad San Francisco de Quito Food Chemical Composition Table after record-level quality control.

Methods

Nineteen nutritional variables were evaluated using descriptive statistics, heteroscedastic one-way analysis of variance, rank correlations, principal component analysis, and hierarchical clustering. Missingness was substantial and non-random across food groups; 427 records contained all 19 variables and were used for multivariate analyses.

Results

Welch tests with Holm correction identified food-group differences for all variables, although the magnitude of group separation ranged from small for vitamin A (η2 = 0.029, 95% CI 0.016-0.061) to large for carbohydrates (η2 = 0.590, 95% CI 0.555-0.632) and energy (η2 = 0.588, 95% CI 0.530-0.654). Spearman analysis showed strong expected compositional associations between total fat and monounsaturated fatty acids (ρ = 0.956) and between total fat and saturated fatty acids (ρ = 0.937), together with non-arithmetic associations involving protein, phosphorus, zinc, cholesterol, and vitamin B12. Parallel analysis retained four principal components, which explained 61.4% of total variance; the first two accounted for 40.2% and primarily represented energy-lipid-mineral covariation and lipid composition. Ward clustering in the full 19-dimensional standardized space supported a three-cluster solution (silhouette = 0.482; median subsampling adjusted Rand index = 0.845), comprising a broad mixed profile, a fiber- and mineral-dense profile, and an oil-dominant profile.

Conclusions

These results describe compositional patterns within the USFQ database and provide a reproducible basis for database harmonization, nutrient profiling, and hypothesis generation. They do not estimate dietary intake, food availability, nutritional adequacy, or health outcomes in the Ecuadorian population.