Association Between the Dietary Inflammatory Index and Cardiorenal Anemia Syndrome: Evidence From NHANES and Related-Phenotype Exploration in KNHANES.

Cardiorenal anemia syndrome (CRAS) links heart failure, chronic kidney disease, and anemia, but the role of diet-related inflammation remains unclear. In the National Health and Nutrition Examination Survey (NHANES), we examined associations of the Dietary Inflammatory Index (DII) with CRAS, dose-response patterns, inflammatory mediation, and machine-learning prediction, alongside analysis of a myocardial infarction-CKD-anemia phenotype in the Korea National Health and Nutrition Examination Survey (KNHANES). This study included 25,388 NHANES and 32,935 KNHANES adults. Survey-weighted regression and restricted cubic splines assessed associations with CRAS in NHANES and the related phenotype in KNHANES. Exploratory mediation evaluated six inflammatory indices. In NHANES participants with heart failure, least absolute shrinkage and selection operator regression selected predictors for 12 internally validated models, with cross-phenotype application in KNHANES and interpretation using Shapley additive explanations (SHAP). Higher DII was associated with greater odds of CRAS in NHANES (odds ratio [OR] = 1.22, 95% CI: 1.07-1.40) and of the related phenotype in KNHANES (OR = 1.19, 95% CI: 1.10-1.29). The association was largely linear in NHANES but nonlinear in KNHANES. In NHANES, all six indices showed positive indirect associations, with estimated proportions ranging from 5.22% to 16.26%. CatBoost showed the highest discrimination in NHANES internal validation (ROC-AUC = 0.741; PR-AUC = 0.556) and in the KNHANES cross-phenotype application (ROC-AUC = 0.753; PR-AUC = 0.698). Higher DII was associated with greater odds of CRAS in NHANES, with directionally consistent findings for the related phenotype in KNHANES. The model may support model-based CRAS identification and showed preliminary applicability to the related phenotype.