Dietary Nutrient Intake and Self-Reported Dermatitis/Eczema in U.S. Adults: A Survey-Weighted Cross-Sectional Analysis of National Health and Nutrition Examination Survey (NHANES) 1999-2004

This is the authors' abstract. We don't add key points for this paper.

Introduction: Dermatitis and eczema are major chronic inflammatory skin conditions that may substantially affect quality of life.Data regarding the association between dietary intake and dermatitis/eczema remain limited and inconsistent.Thus, our aim is to characterize dietary differences between U.S. adults with and without self-reported dermatitis/eczema.Methods: This is a survey-weighted cross-sectional analysis of pooled National Health and Nutrition Examination Survey (NHANES) 1999-2004 data among adults aged 20-59 years with valid Day 1 24-hour dietary recall data.Nutrient intake was compared by dermatitis/eczema status, and survey-weighted univariable and multivariable logistic regression models assessed associations between individual dietary nutrients and dermatitis/eczema.We used a random forest classifier with temporal validation (development on 1999-2002 data, testing on 2003-2004 data) and SHapley Additive exPlanations (SHAP)-based interpretation to evaluate the discriminative capacity of dietary variables.Results: Among 8,702 participants, 1,076 (12.36%) reported dermatitis/eczema within the previous 12 months.Those affected tended to be older, more commonly non-Hispanic White, more educated, and more likely to be obese or former smokers.Overall dietary intake was similar across groups, with one exception: riboflavin was the only nutrient significantly linked to dermatitis/eczema, both in descriptive comparisons (2.29 ± 0.05 mg vs. 2.19 ± 0.03 mg, p=0.049) and in adjusted regression analysis (adjusted odds ratio (aOR) 1.075, 95% CI 1.001-1.154,p=0.046).A random forest model based solely on dietary variables showed only modest predictive ability (area under the receiver operating characteristic curve (AUROC) 0.531), performing worse than a model using clinical and demographic variables alone (AUROC 0.585).Combining both dietary and clinical/ demographic data did not improve predictive performance.Conclusion: Except for a slight, borderline correlation with riboflavin, dietary nutrient consumption in U.S. adults exhibited little independent connection with self-reported dermatitis/eczema.For differentiating illness status, sociodemographic and clinical factors performed better than nutritional variables, indicating that diet has a minor impact compared with other factors.