Development of empirical anti-inflammatory diet index: a cross-sectional study.

Background: There is evidence that some foods and dietary patterns may influence low-grade inflammation status. We aimed to develop a user-friendly empirical Anti-inflammatory Diet Index (eADI) that predicts low-grade chronic inflammation.

Methods: In this cross-sectional study of 4,432 men (aged 74 ± 6 years) from the Cohort of Swedish Men-Clinical, inflammatory status was assessed by high-sensitivity C-reactive protein (hsCRP), interleukin 6 (IL-6), tumor necrosis factor receptor 1 (TNF-R1), and tumor necrosis factor receptor 2 (TNF-R2). Dietary intake was assessed using a food frequency questionnaire. The eADI was developed in a randomly chosen Discovery group (n = 2,216) using a 10-fold feature selection with filtering (based on Lasso regression) to select food groups most correlated with inflammatory biomarkers. From the selected foods, the eADI was then constructed based on summed scores of the consumption tertiles (corresponding to 0, 0.5, and 1 point). Next, in the Replication group (n = 2,216), the association of eADI with inflammatory biomarkers was examined using multivariable-adjusted linear regression models.

Results: eADI-17 included 17 food groups (11 with anti-inflammatory, 6 with pro-inflammatory potential). In the Replication group, the median of eADI-17 was 9 (range: 2-16) scores and the Spearman correlation coefficients for eADI-17 vs. hsCRP, IL-6, TNF-R1, and TNF-R2 were -0.17, -0.23, -0.28, and -0.26, respectively. Each increment by 4.5-point eADI-17 (2 SD) was associated with concentrations that were 12% lower for hsCRP, 6% lower for IL-6, 8% lower for TNF-R1, and 9% lower for TNF-R2. These results obtained for the Replication group were robust as they were essentially the same as those of the Discovery group.

Conclusions: The eADI-17 is a validated, robust and user-friendly anti-inflammatory diet index developed to predict low-grade chronic inflammation. This index has the potential to further refine future dietary guidelines and to be used in personalized nutrition. However, its predictive validity should be further evaluated in diverse populations.