Associations and Multidimensional Characterization of the Nutritional-Inflammatory Profile in Children with Previous SARS-CoV-2 Infection: A Retrospective Cohort Study

Background/Objectives: Nutritional, inflammatory, and lifestyle-related characteristics may contribute to substantial biological heterogeneity among children with documented previous SARS-CoV-2 infection. This study aimed to characterize an integrated nutritional-inflammatory profile and explore its relationships with biological and clinical features in a hospitalized pediatric cohort. Methods: This retrospective cohort study included 260 children with documented previous SARS-CoV-2 infection. Clinical, anthropometric, and laboratory data were complemented by questionnaire-derived dietary and lifestyle information (n = 213). Analyses combined conventional statistics with correlation and multivariate techniques, including PCA and K-means clustering (n = 231) and exploratory Ridge regression (n = 188). Results: Suboptimal serum 25(OH)D concentrations (<30 ng/mL) were present in 55.8% of participants (95% CI [49.7%, 61.7%]), while 43.5% were overweight or obese (95% CI [37.6%, 49.5%]). BMI correlated with hemoglobin (ρ = 0.23) and CRP (ρ = 0.12), while 25(OH)D correlated with potassium (ρ = 0.22). Increasing daily device use was associated with progressively lower 25(OH)D concentrations (H = 24, p < 0.001, ε2 = 0.11), with median values decreasing from 33 ng/mL at <1 h/day to 24 ng/mL at >4 h/day. PCA identified three complementary biological dimensions explaining 55.0% of the total variance, while K-means clustering identified three reproducible exploratory nutritional–inflammatory phenotypes differentiated by age, BMI, hematological and inflammatory characteristics, and selected lifestyle factors. Conclusions: The findings show multidimensional nutritional–inflammatory heterogeneity in children with documented previous SARS-CoV-2 infection. The identified patterns and exploratory phenotypes provide an integrated framework for biological characterization and warrant validation in prospective multicenter cohorts.