Background/Purpose: The preclinical period of rheumatoid arthritis (RA) is characterized by elevated inflammatory biomarkers but it is unclear whether inflammatory diet may contribute to RA risk. We aimed to examine whether a dietary pattern associated with inflammatory biomarkers predicts subsequent risk of RA using the Nurses’ Health Study (NHS) and NHS II.

Methods: We prospectively followed 79,988 women in NHS aged 38-63 years at 1984 and 93,585 women in NHS II aged 27-44 years at 1991, who were free from RA or other connective tissue diseases. Lifestyle, environmental, and anthropometric information were collected at baseline and updated biennially. Dietary data were obtained from validated food frequency questionnaires at baseline and approximately every 4 years during follow-up. RA cases were self-reported and confirmed by a connective tissue disease screening questionnaire and medical record review according to the 1987 ACR criteria. Seropositive RA was defined as positive rheumatoid factor or anti-citrullinated peptide antibody and was determined by medical record review. The inflammatory dietary pattern was measured by the Empirical Dietary Inflammatory Index (EDII), a weighted score including 9 pro- and anti-inflammatory food groups, which was predictive of 3 plasma inflammatory biomarkers (CRP, IL-6 and sTNFR2). Higher (more positive) scores indicate more pro-inflammatory diets and lower (more negative) scores indicate anti-inflammatory diets. We pooled the data from both cohorts to examine the association between cumulative averaged EDII and risk of RA among younger (≤55 years) and older age groups (>55 years) since diet has been shown to have different effects on RA risk according to age of onset in previous studies. The EDII was categorized according to baseline quartile cutoffs. Time-varying Cox regression models were used to calculate the hazard ratios and 95% confidence intervals (CI) after adjusting for potential confounding factors.  

Results: During 34 years of follow-up, we identified 1,188 incident RA cases. Among women ≤55 years old, a higher EDII was associated with a higher RA risk: HRs (95% CI) across increasing quartiles of EDII score were 1.00 (reference), 1.16 (0.87 to 1.53), 1.38 (1.05 to 1.81) and 1.43 (1.08 to 1.89) (p-trend 0.007). After additional adjustment for BMI, the observed results were attenuated. When further stratifying by serostatus, the significant association was observed only for seropositive RA (p-trend: 0.028) (Table). There was no significant association for RA among women aged >55 years (p for EDII-age interaction <0.01).

Conclusion: In these two large prospective cohort studies, inflammatory dietary pattern was associated with increased risk of developing RA among young and middle-aged women. The observed association may be partially mediated through BMI.