Relationship between Non-Energy-Adjusted and Energy-Adjusted Dietary Inflammatory Index and the Healthy Eating Index-2015: an analysis of the National Health and Nutrition Examination Survey (NHANES) 2015–2018

OBJECTIVES: Acknowledging the association between diet and systemic inflammation, the Dietary Inflammatory Index (DII(®)) and the Energy-Adjusted DII (E-DII(TM)) were developed to categorize diet from anti- to pro-inflammatory. The purpose of this study was to evaluate differences in the relationship between DII and E-DII against the Healthy Eating Index (HEI) to assess the use of energy-adjustment when analyzing the inflammatory potential of the diet. METHODS: This cross-sectional secondary data analysis included 5289 adults participating in the National Health and Nutrition Examination Survey (NHANES) between 2015 and 2018. DII and E-DII scores were calculated and grouped into quartiles. Multivariable linear regression was used to evaluate the association between HEI with DII and E-DII separately, adjusting for age, gender, race/ethnicity, education, family-to-poverty ratio and body mass index. Bootstrap methods were used to estimate the difference between coefficients for E-DII and DII from their respective models. RESULTS: Results suggest that HEI scores were significantly lower between quartile 2 and quartile 1 of DII scores (Q2 vs. Q1: β(SE) = −7.45(0.54), p < 0.05) and quartile 3 and quartile 4 against quartile 1 of E-DII scores (Q3 vs. Q1: β(SE) = −16.56(0.65), p < 0.05 and Q4 vs. Q1: β(SE) = −24.93(0.87), p < 0.05) in unadjusted models. Similar results were observed in adjusted models (HEI and DII Q3 vs. Q1: β(SE) = −10.68(0.82), p = 0.049; HEI and E-DII Q2 vs. Q1: β(SE) = −9.14(0.64) and Q3 vs. Q1: β(SE) = −15.76(0.44) and Q4 vs. Q1: β(SE) = −23.77(0.58), p < 0.05). Further, 52% of the variance in HEI scores was explained by the E-DII in both adjusted and unadjusted models (R(2) = 0.52). In contrast, 17% of the variance in HEI score is explained by the DII in the unadjusted model (R(2) = 0.17), yet this increased to 26% of the variance in the adjusted model (R(2) = 0.26). The difference between HEI scores for the first versus the fourth quartile of DII scores (−15.64) was significantly larger than the difference between HEI scores for the first versus the fourth quartile of E-DII scores (−25.90; bootstrap estimated 95% CI: 53.41–62.41). CONCLUSIONS: Use of an inflammatory index along with the HEI may provide further understanding into relationships between dietary quality by nutrient and food group consumption on the inflammatory potential of the diet.