Comparison of the Predictive Ability of an Empirical and a Literature‐derived Index of Dietary Inflammation Potential

Background Inflammation plays an important role in the development of many chronic diseases and dietary factors can modulate inflammation. An empirical dietary inflammatory pattern (EDIP) comprised of food groups as reported on a food frequency questionnaire (FFQ) and a literature‐derived dietary inflammatory index (DII) comprised mainly of nutrients, exist to describe dietary inflammation potential. Objective To compare the ability of the EDIP and DII in predicting levels of inflammation markers in two cohorts. Design Both EDIP and DII scores were calculated from FFQ data reported by 5826 women in the Nurses' Health Study‐II and 5227 men in the Health Professionals Follow‐up Study. We used multivariable‐adjusted linear regression analyses to calculate relative concentrations of four plasma inflammation markers: C‐reactive protein (CRP), interleukin‐6 (IL6), tumor necrosis factor alpha receptor 2 (TNFαR2) and adiponectin across quintiles of the dietary indices. Results Correlations between the EDIP and DII scores were modest (Spearman r =0.28, women; 0.23, men; all P <0.0001). In both women and men, higher index scores were associated with higher (lower for adiponectin) concentrations of inflammation markers, except for no association between DII and adiponectin in men. EDIP scores explained a higher proportion of variance in biomarker concentrations and with associations of higher magnitudes compared to DII scores except for IL6 in men; e.g., in EDIP quintile 5, the percent change in relative concentrations of biomarkers was: CRP, +23.2%; IL6, +2.4%; TNFαR2, +0.1%; and adiponectin, +13.9% higher compared to DII quintile 5 in women; and CRP, +5.8%; IL6, −9.7%; TNFαR2, +1.9%; and adiponectin, +9.1% higher in men. Conclusion Despite design differences, the EDIP and DII both assess dietary inflammation potential, with the EDIP showing a higher ability to predict concentrations of plasma inflammation markers in these cohorts. Both indices may be useful in predicting inflammation‐mediated diet‐disease associations. Support or Funding Information Drs. Fred K. Tabung, Jorge E. Chavarro and Frank B. Hu were supported by National Institutes of Health (NIH) grants K99 CA207736, P30 DK046200 and U54 CA155426 respectively. The HPFS and NHS‐II cohorts are supported by NIH grants: UM1 CA167552 and UM1 CA176726, respectively