Abstract
Objectives
To evaluate the association between long-term dietary quality, measured by the 2010 Alternative Healthy Eating Index (AHEI-2010), and risk of rheumatoid arthritis in women.
Methods
We prospectively followed 76,597 women in the Nurses’ Health Study aged 30–55 years and 93,392 women in the Nurses’ Health Study II aged 25–42 years at baseline and free from rheumatoid arthritis or other connective tissue diseases. The lifestyle, environmental exposure and anthropometric information were collected at baseline and updated biennially. Cumulative follow-up rates were more than 90% for both cohorts. The primary outcome was rheumatoid arthritis alone with two subtypes of the disease: seropositive and seronegative rheumatoid arthritis.
RESULTS
During 3,678,104 person-years, 1,007 RA cases were confirmed. In the multivariable-adjusted model, long-term adherence to healthy eating patterns was marginally associated with reduced RA risk. To assess potential effect modification by age at diagnosis, we stratified by age. Among women ≤ 55 years old, better quality diet was associated with lower RA risk (HRQ4 vs Q1: 0.67; 95% CI: 0.51 to 0.88; p trend: 0.002), but no significant association was found for women >55 years (p interaction: 0.005). When stratifying by serostatus, the inverse association among those aged ≤55 years was strongest for seropositive RA (HRQ4 vs Q1: 0.60; 95% CI: 0.42 to 0.86; p-trend: 0.003).
Conclusion
A healthier diet was associated with a reduced risk of RA occurring at 55 years of age or younger, particularly seropositive RA.
INTRODUCTION
Rheumatoid arthritis (RA) is a common autoimmune disease, affecting approximately 1.3 million Americans including 2.5 times as many women as men.1 Although the etiology of RA is not completely understood, it is widely recognized that both genetic and environmental exposures including lifestyle and dietary risk factors contribute to the development of RA. Our group and others have reported that high intake of alcohol, fish, olive oil and cooked vegetables confer a protective effect against the development of RA,2–6 whereas red meat and sugar-sweetened soda intake may increase the risk of RA.7, 8 However, not all studies confirmed such associations.9–11 One explanation for the inconsistent findings may be that individual food/nutrients only confer modest beneficial effects, so that any protective effects would not be observed unless groups of dietary factors are consumed together. Dietary pattern analysis in relation to disease risk addresses these limitations as it integrates the complexity of overall diet.12
The 2010 Alternative Healthy Eating Index (AHEI-2010) is a dietary quality score based on the recent Dietary Guidelines for Americans composed of 11 foods and nutrients that have been consistently associated with risk of chronic diseases.13 Among 11 foods/nutrients included in the dietary index, fruits, vegetables, whole grains, nuts, long-chain (n-3) fat (% of total energy), polyunsaturated fat (% of total energy) and moderate alcohol consumption were considered as healthy items while sugar-sweetened beverages (including fruit juice), red/processed meat, trans fat (% of total energy) and sodium intake were deemed as unhealthy components. Higher scores of AHEI-2010 have been strongly associated with reduced risk for major chronic diseases including cardiovascular diseases (CVD), type 2 diabetes (T2D) and certain types of cancer.14–19 No study has investigated the associations between overall dietary quality and risk of autoimmune diseases, such as RA. Our primary aim was to investigate associations between AHEI-2010 score and risk of RA in two large prospective cohort studies of women: Nurses’ Health Study and Nurses’ Health Study II.
Previous studies on CVD have showed that the influence of established risk factors such as smoking, BMI, hypertension, diabetes, hyperlipidemia diminished with advancing age,20–22 which is likely due to blunted response to autonomic nervous system stimuli 23, 24 that is triggered by these modifiable factors. Similar phenomenon was observed in our previous analysis that demonstrated a significant trend toward increased risk of RA diagnosed at age ≤55 years among overweight and obese women, but not in women >55 years.25 Moreover, literature suggest that RA occurred at younger and older age are etiologically different in many respects. 26, 27 Thus, our secondary aim was also to examine the diet quality and RA associations in both younger (age≤ 55) and older age (>55 years).
METHODS
Study population
In 1976, NHS was initiated and included 121,700 female registered nurses aged 30 to 55 years. In 1989, 116,671 female registered nurses aged 25–42 years enrolled in NHSII. Participants from both prospective cohorts provided information regarding lifestyle behaviors, medical history, and diseases through self-administered questionnaires at entry and updated every 2 years. The response rate was maintained over 90% for every 2 year period in both cohorts.29, 30 The baselines for NHS and NHS II in the current study were 1984 and 1991 when a comprehensive Food Frequency Questionnaire (FFQ) was introduced in each cohort. A total of 76,597 and 93,392 participants from each cohort were included in the final analysis after excluding those who did not answer the FFQ and those who had RA or other connective tissue diseases at baseline. We censored all women who reported psoriasis, psoriatic arthritis or other connective tissue diseases at the self-reported date, unless the diagnosis was subsequently confirmed as RA. Women lost to follow-up were censored at their last response to questionnaires. The institutional review board at Partners Healthcare approved the study protocol.
Assessment of dietary factors and AHEI-2010
We used a 133-item semi-quantitative FFQ to collect dietary data approximately every 4 years from 1984–2010 in the NHS and 1991–2011 in NHSII. 28 About 84% of NHS participants and 90% of NHS II participants answered FFQ at respective baseline. Participants who returned FFQ with more than 70 missing items, total energy intake less than 500 kcal/day or greater than 3500 kcal/d were excluded from the analysis.8 Validation studies from two parallel cohorts showed high correlation coefficients between FFQ and multiple dietary records for the dietary factors.28, 31 We calculated nutrient intakes by multiplying the frequency of intake for each food by its nutrient content and summing nutrient contributions across all food items, using previously validated methods.32
As a scoring system, the AHEI-2010 integrates the dietary data to evaluate the overall dietary quality. The rationale of food items selection and score criteria of AHEI-2010 has been discussed elsewhere.13 Women with moderate alcohol intake defined as 0.5–1.5 drinks/d received a maximum score of 10 but non-drinkers or heavy drinkers were assigned lower scores, reflecting the U-shaped association of alcohol consumption with chronic diseases.13 All AHEI-2010 components are scored from 0 (worst) to 10 (best), and the total AHEI-2010 score can range from 0 to 110 (non to perfect adherence). A higher AHEI-2010 score indicates better diet quality. In order to investigate the potential cumulative effect of a long-term diet, we calculated cumulative average AHEI-2010 from all of the preceding measures over the study period.8, 33 The missing cycles of FFQ were excluded from the estimates of cumulative average dietary measures.
Identification of RA
The identification of RA cases consisted of two steps. In the first step, participants who self-reported a physician diagnosis of RA received the connective tissue disease (CTD) screening questionnaire (CSQ) through mail. For those with positive screening for RA or other CTDs based on CSQ, medical records were requested.34 In the second step, two rheumatologists independently reviewed medical records according to the 1987 American College of Rheumatology classification.35 The serologic phenotype of RA was determined from medical records by positive rheumatoid factors (RF) (available since baseline) or cyclic citrullinated peptide (CCP) antibodies (available since the early 2000s).34, 36
Assessment of time-varying covariates
Information regarding participants’ demographic, reproductive, clinical and behavioral factors has been collected with the biennial follow-up questionnaires. Body mass index (BMI) was calculated as weight in kilograms divided by height in meters and was categorized into 5 levels: <18.49, 18.5–22.9, 23–24.9, 25–30, and ≥30 kg/m2. Four categories were created for smoking status: never, past, current 1–14 cigarettes/day, current ≥15 cigarettes/day. Recreational physical activity levels were categorized as: 0–2.9, 3–8.9, 9–17.9, 18–26.9 and ≥27 metabolic equivalents (METs) /week. 37 Parity and breastfeeding status were jointly defined as: nulliparous, parous/no breastfeeding, parous/1–12 months breastfeeding, parous/ >12 months breastfeeding. Age at menarche was classified into 3 categories: <12, 12, >12 years and we defined the hormone use status as: pre-menopausal, post-menopausal with never use, current use and past use. Census tract median family income and total energy intake were ranked using quartiles and quintiles respectively.
Statistical analysis
Baseline characteristics were expressed as mean (SD) or median for continuous variables and percentage was used for categorical variables according to quartiles of AHEI-2010 score. In order to assess age-specific effects and potential effect modifiers, we pooled the data from both cohorts to include a wide-range of age groups. To control for possible cohort effects given secular trends in food supply according to calendar year, an indicator variable for cohorts was used in model stratification. For the primary analysis, time-varying Cox proportional hazards models were employed to assess the association between AHEI-2010 and risk of RA. All RA, seropositive and seronegative RA were separate outcomes in the analysis. Cumulative averaged AHEI-2010 score was included in the model as a time-varying exposure for which the values were updated during the follow-up. The proportional hazard assumption was evaluated by including an interaction term between continuous AHEI-2010 and logarithm of person-time and was met in all analyses. The main model adjusted for potential risk factors of RA that are also associated with AHEI including age, smoking, age at menarche, parity/breastfeeding, hormone use and total energy intake. We additionally adjusted for BMI to assess its potential mediation effects in a separate model. The median value of each quartile was used as a continuous variable to calculate the p-value for trend.
For the secondary aim, we stratified the participants by age rounded to integer (age ≤55 and age >55 years) based on our pre-specified hypotheses.8, 25 Women were followed up from baseline to age 55 years, and from age 56 years to the end of cohorts respectively. The interaction was tested by comparing two β coefficients for the continuous AHEI-2010 score from respective multivariable models in a χ2 test. We also conducted an exploratory analysis to evaluate the effect of each individual component score of AHEI-2010 in relation to RA risk.
In addition, we performed sensitivity analyses to examine the robustness of our findings. First, to maximally control for residual confounding effects, we further included variables that varied by AHEI but are less associated with RA in the NHS including census tract median family income, physical activity, multivitamin use and self-reported T2D. Secondly, after we identified that red meat and alcohol were two individual components of AHEI-2010 that showed particular beneficial effects against RA risk for women at age 55 years old or younger, we repeated the analysis using AHEI-2010 without either of these two components, to assess whether the observed associations were driven by single food item. Missing covariate data was carried forward one questionnaire cycle. All statistical tests were 2-sided with significant level of 0.05 and performed using SAS 9.3 (SAS Institute, Cary, NC).
RESULTS
A total of 1,007 RA cases (624 seropositive, 383 seronegative) were identified during 3,678,104 person-years of follow-up. The mean follow-up time was 21.6 years. The ranges of AHEI-2010 score were 15.6–100.1 and 13.7–95.8 at baseline for NHS and NHS II respectively. In both cohorts, higher baseline AHEI-2010 score was positively associated with desirable lifestyle and socioeconomic status including family median income, physical activity, alcohol consumption and multivitamin use (Table 1). Participants with higher AHEI-2010 score also tended to have lower BMI and total energy intake, and were less likely to be parous and current smokers. BMI and AHEI were significantly negatively correlated but the Spearman correlation coefficient was small (−0.08, p<0.001).
Table 1.
Age-standardized baseline characteristics of participants from Nurses’ Health Study (N=76,597 in 1984) and study participants from Nurses’ Health Study II (N=93,392 in 1991) according to AHEI-2010 quartile*
| AHEI-2010 quartile, NHS | AHEI-2010 quartile, NHSII | |||||||
|---|---|---|---|---|---|---|---|---|
| Q1 (n=19,149) |
Q2 (n=19,149) |
Q3 (n=19,150) |
Q4 (n=19,149) |
Q1 (n=23,348) |
Q2 (n=23,348) |
Q3 (n=23,348) |
Q4 (n=23,348) |
|
| AHEI-2010, median (range)† | 34.6 | 43.6 | 50.7 | 60.0 | 36.0 | 44.5 | 51.7 | 61.6 |
| Age (years)† | 49.3(7.1) | 50.3(7.1) | 51.2(7.1) | 52.4(6.9) | 35.9(4.7) | 36.4(4.6) | 36.8(4.6) | 37.3(4.5) |
| Body mass index (kg/m²) | 25.4(5.2) | 25.2(4.9) | 25.0(4.6) | 24.4(4.1) | 25.3(6.0) | 24.8(5.4) | 24.5(5.1) | 23.9(4.6) |
| Census tract median family income $ (*1000), mean (SD) | 60.4(22.7) | 63.1(24.5) | 66.0(26.4) | 69.6(28.8) | 56.8(18.9) | 59.9(21.2) | 62.6(23.1) | 66.1(25.4) |
| Physical activity(MET-h/wk)‡ | 4.6(1.9, 12.7) | 6.5(2.4, 16.0) | 8.3(2.9, 20.1) | 11.1(4.0, 24.9) | 8.3(3.2, 18.7) | 10.9(4.4, 23.3) | 13.8(5.8, 27.7) | 19.4(8.6, 37.1) |
| Alcohol consumption (g/day) ‡ | 1.1(0, 5.3) | 1.9(0, 7.8) | 2.8(0, 9.9) | 3.9(0.9, 11.0) | 0.5(0, 1.8) | 0.9(0, 2.8) | 1.4(0, 4.2) | 2.0(0.5, 6.3) |
| Current smoker, % | 30.3 | 25.8 | 23.1 | 18.1 | 14.7 | 13.1 | 11.8 | 9.8 |
| Multivitamin use, % | 31.1 | 34.7 | 38.3 | 43.3 | 40.0 | 42.5 | 45.1 | 47.7 |
| Menarche before age 12, % | 21.3 | 22.5 | 23.9 | 25.3 | 22.3 | 24.0 | 25.4 | 26.0 |
| Oral contraceptive use, % | 47.9 | 48.5 | 48.8 | 50.9 | 83.6 | 83.6 | 83.5 | 83.2 |
| Parous, % | 93.4 | 93.0 | 92.7 | 91.0 | 75.8 | 73.1 | 68.4 | 60.6 |
| Breastfeeding ≥ 12 months, % | 16.0 | 16.5 | 17.6 | 18.0 | 24.6 | 27.1 | 26.8 | 25.8 |
| Postmenopausal, % | 48.8 | 48.5 | 49.0 | 49.0 | 3.3 | 3.3 | 3.2 | 3.2 |
| Current postmenopausal hormone use, % | 18.7 | 19.5 | 21.0 | 23.9 | 2.7 | 2.7 | 2.7 | 2.6 |
| Total energy intake (Kcal/d) | 1909(501) | 1769(532) | 1688(528) | 1599(509) | 1937(531) | 1804(552) | 1727(545) | 1690(529) |
Values are means (SD) or percentages and are standardized to the age distribution of the study population.
Value is not age adjusted
Values are median (interquartile range)
In multivariable-adjusted model, women in the highest quartile of AHEI-2010 had 15% reduced risk (HR Q4 vs Q1= 0.85, 95% CI: 0.70 to 1.02; p for trend 0.08) compared to the women in the lowest quartile. The association was slightly attenuated after additional adjustment for BMI (HRQ4 vs Q1= 0.87, 95% CI: 0.73 to 1.05; p trend 0.16). Results for both serologic subtypes of RA suggested an inverse association, but neither was statistically significant. In analyses restricted to women at age 55 years or younger (444 RA cases), a significant inverse dose-response relationship was found between AHEI-2010 and RA incidence (Table 3). Women in the top quartile of AHEI-2010 had 33% (HRQ4 vs Q1= 0.67, 95% CI: 0.51 to 0.88; p trend 0.002) lower risk of RA compared with those in the lowest quartile and the observed association remained statistically significant after additional BMI adjustment (HRQ4 vs Q1= 0.71, 95% CI: 0.54 to 0.94; p for trend 0.009). Further stratification analysis by two serologic subtypes of RA showed significant inverse associations for seropositive RA. No significant associations were detected for women at age over 55 years (p for age by AHEI-2010 interaction 0.005).
Table 3.
Pooled hazard ratios (95% CI) for incident RA according to cumulative AHEI-2010 quartile in Nurses’ Health Study (NHS, 1984–2012) and Nurses’ Health Study II (NHS II, 1991–2013) stratified by age before or after 55 years
| AHEI Score, quartile
|
|||||
|---|---|---|---|---|---|
| Q1 | Q2 | Q3 | Q4 | p for trend ‡ | |
| Age ≤55 years | |||||
|
| |||||
| All RA | |||||
| Case/person-years | 136/558,968 | 125/535,171 | 99/513,734 | 84/485,162 | |
| Age adjusted model 1 | 1.00 | 0.90(0.70 to 1.15) | 0.76(0.58 to 0.98) | 0.68(0.52 to 0.89) | 0.003 |
| Multivariable model 2 (main)* | 1.00 | 0.89(0.69 to 1.14) | 0.75(0.57 to 0.97) | 0.67(0.51 to 0.88) | 0.002 |
| Multivariable model 3† | 1.00 | 0.90(0.70 to 1.15) | 0.76(0.59 to 1.00) | 0.71(0.54 to 0.94) | 0.009 |
|
| |||||
| Seropositive RA | |||||
| Case/person-years | 87/548,349 | 83/525,054 | 70/503,397 | 47/476,102 | |
| Age adjusted model 1 | 1.00 | 0.95(0.70 to 1.29) | 0.81(0.58 to 1.11) | 0.63(0.44 to 0.89) | 0.005 |
| Multivariable model 2 (main)* | 1.00 | 0.93(0.68 to 1.27) | 0.78(0.57 to 1.09) | 0.60(0.42 to 0.86) | 0.003 |
| Multivariable model 3† | 1.00 | 0.95(0.69 to 1.28) | 0.81(0.58 to 1.12) | 0.65(0.45 to 0.92) | 0.011 |
|
| |||||
| Seronegative RA | |||||
| Case/person-years | 49/556,795 | 42/533,240 | 29/511,879 | 37/483,609 | |
| Age adjusted model 1 | 1.00 | 0.80(0.52 to 1.23) | 0.67(0.43 to 1.05) | 0.77(0.50 to 1.18) | 0.18 |
| Multivariable model 2 (main)* | 1.00 | 0.81(0.52 to 1.24) | 0.68(0.43 to 1.07) | 0.79(0.51 to 1.24) | 0.26 |
| Multivariable model 3† | 1.00 | 0.81(0.52 to 1.24) | 0.69(0.44 to 1.09) | 0.83(0.53 to 1.29) | 0.34 |
|
| |||||
| Age>55 years | |||||
|
| |||||
| All RA | |||||
| Case/person-years | 116/365,249 | 136/390,708 | 156/412,404 | 154/440,769 | |
| Age adjusted model 1 | 1.00 | 1.08(0.85 to 1.37) | 1.10(0.87 to 1.40) | 1.11(0.87 to 1.41) | 0.39 |
| Multivariable model 2 (main)* | 1.00 | 1.08(0.85 to 1.37) | 1.10(0.86 to 1.40) | 1.10(0.86 to 1.41) | 0.45 |
| Multivariable model 3† | 1.00 | 1.08(0.85 to 1.37) | 1.10(0.87 to 1.41) | 1.11(0.87 to 1.42) | 0.40 |
|
| |||||
| Seropositive RA | |||||
| Case/person-years | 68/363,288 | 78/388,184 | 90/409,690 | 101/438,302 | |
| Age adjusted model 1 | 1.00 | 1.04(0.76 to 1.42) | 1.13(0.83 to 1.54) | 1.19(0.88 to 1.62) | 0.21 |
| Multivariable model 2 (main)* | 1.00 | 1.06(0.77 to 1.45) | 1.17(0.85 to 1.59) | 1.22(0.89 to 1.68) | 0.17 |
| Multivariable model 3† | 1.00 | 1.06(0.77 to 1.45) | 1.17(0.85 to 1.60) | 1.23(0.90 to 1.69) | 0.16 |
|
| |||||
| Seronegative RA | |||||
| Case/person-years | 48/364,396 | 58/389,885 | 66/411,265 | 53/439,561 | |
| Age adjusted model | 1.00 | 1.13(0.78 to 1.64) | 1.07(0.73 to 1.56) | 0.99(0.67 to 1.45) | 0.87 |
| Multivariable model 1 (main)* | 1.00 | 1.09(0.75 to 1.58) | 1.01(0.69 to 1.48) | 0.91(0.62 to 1.36) | 0.57 |
| Multivariable model 2† | 1.00 | 1.09(0.75 to 1.58) | 1.02(0.69 to 1.49) | 0.93(0.63 to 1.39) | 0.66 |
Hazard ratios were calculated by using time-varying Cox proportional hazards models
Adjustment for age, cohort, smoking (never, past, current 1–14 cigarettes/d, current ≥15 cigarettes/d), age at menarche (<12, 12, >12 years), parity and breast feeding (nulliparous, parous/no breastfeeding, parous/1–12 months breastfeeding, parous/ >12 months breastfeeding), hormone use (pre-menopausal, post-menopausal with never use, current use and past use) and total energy (quintiles)
Additional adjustment for BMI (<20, 20–22.9, 23–24.9, 25–29.9, ≥30kg/m2)
p for trend was derived from tests of linear trend across categories of AHEI scores by treating the median value of each category as a continuous variable
As an exploratory analysis, we further examined the association between each AHEI component and RA risk among women at 55 years or younger age (Table 4). A higher score of alcohol (moderate intake) or red meat consumption (lower intake) was significantly associated with a reduced RA risk independent of current BMI and other covariates, while no significant associations were observed for other AHEI components.
Table 4.
Pooled hazard ratios (95% CI) for incident RA according to cumulative AHEI-2010 individual component score quartile for women aged 55 years or younger in Nurses’ Health Study (NHS, 1984–2012) and Nurses’ Health Study II (NHS II, 1991–2013)*
| AHEI-2010
|
|||||
|---|---|---|---|---|---|
| Q1 | Q2 | Q3 | Q4 | p for trend § | |
| Unhealthy food/nutrient | |||||
|
| |||||
| Red/processed meat | |||||
| Median, servings/d | 1.50 | 1.04 | 0.72 | 0.37 | |
| Cases/ Person-years | 131/544,851 | 138/530,911 | 93/522,149 | 82/495,159 | |
| Age adjusted model 1 | 1.00 | 0.93(0.73 to 1.18) | 0.71(0.54 to 0.92) | 0.61(0.46 to 0.81) | <0.001 |
| Multivariable-adjusted model 2 (main)† | 1.00 | 0.89(0.69 to 1.14) | 0.67(0.50 to 0.88) | 0.58(0.43 to 0.79) | <0.001 |
| Multivariable-adjusted model 3‡ | 1.00 | 0.91(0.71 to 1.17) | 0.70(0.53 to 0.93) | 0.64(0.47 to 0.87) | 0.002 |
|
| |||||
| Sugar-sweetened beverages | |||||
| Median, servings/d | 1.00 | 0.83 | 0.56 | 0.18 | |
| Cases/ Person-years | 127/647,550 | 85/402,430 | 124/517,586 | 108/525,503 | |
| Age adjusted model 1 | 1.00 | 0.95(0.71 to 1.27) | 1.10(0.86 to 1.40) | 1.02(0.79 to 1.31) | 0.71 |
| Multivariable-adjusted model 2 (main)† | 1.00 | 0.93(0.69 to 1.25) | 1.07(0.83 to 1.38) | 1.01(0.77 to 1.32) | 0.76 |
| Multivariable-adjusted model 3‡ | 1.00 | 0.90(0.67 to 1.21) | 1.02(0.80 to 1.32) | 0.96(0.74 to 1.26) | 0.97 |
|
| |||||
| Trans fat | |||||
| Median, % of total energy | 2.0 | 1.6 | 1.4 | 1.0 | |
| Cases/ Person-years | 127/543,068 | 108/534,883 | 110/521,468 | 99/493,651 | |
| Age adjusted model 1 | 1.00 | 0.95(0.74 to 1.22) | 0.81(0.62 to 1.06) | 0.86(0.66 to 1.12) | 0.15 |
| Multivariable-adjusted model 2 (main)† | 1.00 | 0.94(0.73 to 1.21) | 0.81(0.62 to 1.05) | 0.86(0.65 to 1.12) | 0.15 |
| Multivariable-adjusted model 3‡ | 1.00 | 0.96(0.74 to 1.23) | 0.84(0.64 to 1.10) | 0.93(0.71 to 1.22) | 0.39 |
|
| |||||
| Sodium¶ | |||||
| Cases/ Person-years | 125/546,210 | 113/468,167 | 112/565,577 | 94/513,114 | |
| Age adjusted model 1 | 1.00 | 1.04(0.81 to 1.35) | 0.87(0.67 to 1.12) | 0.79(0.60 to 1.03) | 0.04 |
| Multivariable-adjusted model 2 (main)† | 1.00 | 0.92(0.69 to 1.23) | 0.73(0.52 to 1.02) | 0.65(0.44 to 0.98) | 0.02 |
| Multivariable-adjusted model 3‡ | 1.00 | 0.95(0.71 to 1.27) | 0.77(0.55 to 1.08) | 0.71(0.48 to 1.07) | 0.07 |
|
| |||||
| Healthy food/nutrient | |||||
|
| |||||
| Nuts | |||||
| Median, servings/d | 0.07 | 0.18 | 0.29 | 0.57 | |
| Cases/ Person-years | 125/552,650 | 101/527,416 | 109/508,006 | 109/504,997 | |
| Age adjusted model 1 | 1.00 | 0.82(0.63 to 1.07) | 0.84(0.65 to 1.09) | 0.95(0.73 to 1.22) | >0.99 |
| Multivariable-adjusted model 2 (main)† | 1.00 | 0.80(0.61 to 1.04) | 0.81(0.62 to 1.06) | 0.92(0.70 to 1.20) | 0.89 |
| Multivariable-adjusted model 3‡ | 1.00 | 0.80(0.61 to 1.05) | 0.82(0.62 to 1.07) | 0.95(0.73 to 1.25) | 0.88 |
|
| |||||
| Vegetables | |||||
| Median, servings/d | 1.4 | 2.3 | 3.2 | 4.4 | |
| Cases/ Person-years | 123/542,983 | 101/530,967 | 118/517,891 | 102/501,229 | |
| Age adjusted model 1 | 1.00 | 0.83(0.63 to 1.08) | 1.01(0.78 to 1.30) | 0.90(0.69 to 1.17) | 0.73 |
| Multivariable-adjusted model 2 (main)† | 1.00 | 0.80(0.61 to 1.05) | 0.96(0.73 to 1.25) | 0.84(0.63 to 1.12) | 0.45 |
| Multivariable-adjusted model 3‡ | 1.00 | 0.80(0.61 to 1.05) | 0.95(0.73 to 1.24) | 0.84(0.63 to 1.11) | 0.43 |
|
| |||||
| Fruits | |||||
| Median, servings/d | 0.42 | 0.85 | 1.34 | 2.15 | |
| Cases/ Person-years | 130/553,732 | 114/539,303 | 97/510,931 | 103/489,103 | |
| Age adjusted model 1 | 1.00 | 1.21(0.93 to 1.57) | 0.85(0.64 to 1.13) | 1.05(0.81 to 1.37) | 0.73 |
| Multivariable-adjusted model 2 (main)† | 1.00 | 1.20(0.92 to 1.56) | 0.83(0.62 to 1.11) | 1.03(0.77 to 1.36) | 0.62 |
| Multivariable-adjusted model 3‡ | 1.00 | 1.20(0.92 to 1.56) | 0.84(0.63 to 1.12) | 1.05(0.79 to 1.39) | 0.73 |
|
| |||||
| Poly-unsaturated fat | |||||
| Median, % of total energy | 4.3 | 5.2 | 5.9 | 7.0 | |
| Cases/ Person-years | 107/517,432 | 99/527,495 | 111/526,417 | 127/521,723 | |
| Age adjusted model 1 | 1.00 | 0.92(0.69 to 1.23) | 1.13(0.86 to 1.49) | 1.15(0.88 to 1.51) | 0.16 |
| Multivariable-adjusted model 2 (main)† | 1.00 | 0.92(0.69 to 1.22) | 1.12(0.85 to 1.47) | 1.13(0.86 to 1.49) | 0.20 |
| Multivariable-adjusted model 3‡ | 1.00 | 0.89(0.67 to 1.19) | 1.08(0.82 to 1.42) | 1.09(0.83 to 1.43) | 0.32 |
|
| |||||
| Alcohol¶ | |||||
| Range, drinks/d | 0 | 2.0–2.5 | 1.5–2.0 | 0.5–1.5 | |
| Cases/ Person-years | 153/685,795 | 87/307,692 | 120/637,042 | 84/462,539 | |
| Age adjusted model 1 | 1.00 | 1.08(0.82 to 1.42) | 0.86(0.68 to 1.10) | 0.72(0.55 to 0.94) | 0.006 |
| Multivariable-adjusted model 2 (main)† | 1.00 | 1.03(0.78 to 1.36) | 0.84(0.66 to 1.06) | 0.67(0.51 to 0.89) | 0.002 |
| Multivariable-adjusted model 3‡ | 1.00 | 1.03(0.78 to 1.36) | 0.87(0.68 to 1.10) | 0.73(0.55 to 0.96) | 0.013 |
|
| |||||
| Whole grain | |||||
| Median, g/d | 6.5 | 13.9 | 22.2 | 36.1 | |
| Cases/ Person-years | 129/543,264 | 106/536,438 | 108/520,691 | 101/492,676 | |
| Age adjusted model 1 | 1.00 | 0.85(0.66 to 1.09) | 0.81(0.62 to 1.05) | 0.84(0.64 to 1.10) | 0.24 |
| Multivariable-adjusted model 2 (main)† | 1.00 | 0.84(0.65 to 1.09) | 0.80(0.61 to 1.04) | 0.84(0.64 to 1.11) | 0.25 |
| Multivariable-adjusted model 3‡ | 1.00 | 0.84(0.65 to 1.08) | 0.81(0.62 to 1.06) | 0.88(0.67 to 1.16) | 0.44 |
|
| |||||
| Long-chain (n-3) fats (EPA+DHA) | |||||
| Median, mg/d | 63 | 120 | 178 | 250 | |
| Cases/ Person-years | 116/529,182 | 118/544,526 | 100/493,951 | 110/525,410 | |
| Age adjusted model 1 | 1.00 | 0.93(0.72 to 1.21) | 0.85(0.65 to 1.12) | 0.90(0.69 to 1.17) | 0.37 |
| Multivariable-adjusted model 2 (main)† | 1.00 | 0.91(0.70 to 1.18) | 0.82(0.62 to 1.08) | 0.86(0.66 to 1.13) | 0.23 |
| Multivariable-adjusted model 3‡ | 1.00 | 0.90(0.69 to 1.18) | 0.81(0.62 to 1.07) | 0.86(0.66 to 1.13) | 0.24 |
Hazard ratios were calculated by using time-varying Cox proportional hazards models
The AHEI-2010 score is reversely coded for red meat, sugar-sweetened beverages, Trans fat and sodium: higher score means lower intake. The AHEI-2010 score is directly coded for nuts, vegetables, fruits, poly-unsaturated fat, alcohol, whole grain and long-chain (n-3) fats: higher score means higher intake
Adjustment for age, cohort, smoking (never, past, current 1–14 cigarettes/d, current ≥15 cigarettes/d), age at menarche (<12, 12, >12 years), parity and breast feeding (nulliparous, parous/no breastfeeding, parous/1–12 months breastfeeding, parous/ >12 months breastfeeding), hormone use (pre-menopausal, post-menopausal with never use, current use and past use) and total energy (quintiles)
Additional adjustment for BMI (<20, 20–22.9, 23–24.9, 25–29.9, ≥30kg/m2)
p for trend was derived from tests of linear trend across categories of AHEI scores by treating the median value of each category as a continuous variable
Sodium score was derived according to an empirical quintile cutoff in the repeated dietary assessment, so the unit was not available
Alcohol score was derived using pre-specified cutoff, so only range was available
In sensitivity analyses that additionally adjusted for census tract median family income, physical activity, multi-vitamin use and diabetes history, the results were unchanged. Both alcohol-excluded and red meat-excluded AHEI-2010 yielded comparable estimates to the primary stratified analysis, so it is unlikely that these items alone were responsible for the findings.
DISUCSSION
Data from these two large cohort studies of women suggest that long-term adherence to higher dietary quality guidelines was associated with reduced RA risk, in particular for seropositive RA with onset at age 55 years or earlier. Among the individual components of AHEI-2010 score, moderate alcohol consumption and lower red meat intake were found to be most associated with decreased early-onset RA risk, while the associations of other components with RA risk were not strong enough to reach statistical significance. These results indicate that an overall healthy diet quality may be more beneficial for RA risk reduction than individual foods and nutrients, particularly for early-onset seropositive RA.
The observed inverse associations between AHEI-2010 score and RA risk are generally expected as most of the score’s components are biologically relevant in reducing RA risk. Owing to the rich contents of dietary antioxidants, fruits and vegetables have long been suggested as healthy foods that may be effective for RA prevention. In our previous study evaluating association of Alternate Mediterranean Diet Score (AMED) with RA risk, we did not observe a significant association of AMED score as well as the fruit and vegetable consumption components with overall RA risk, but did not investigate whether this was associated with earlier onset RA.33 AMED is the simple sum of 9 dichotomized components at median, not including detailed amount of actual dietary intakes. We previously found another AHEI-2010 component, sugar-sweetened beverages, to be linked to increased risk of later onset RA (diagnosed after age 55 years) using different comparison groups from the current study.8 While the EPIC-Norfolk study found an elevated risk of RA among people with higher red meat consumption,7 our group assessed the red meat-RA association in NHS 9 years ago and did not confirm such associations.38 However, our current findings do not necessarily contradict our previous report. In the current analysis, we combined data from two cohorts with more incident RA cases and longer follow-up of both younger and older women than the previous analyses, allowing us to detect potential differing effects of diet quality on RA occurred in younger and older age.
The polyunsaturated, particularly long-chain n-3 fatty acids, have been found to be effective in both reducing RA risk and improving symptoms in RA patients through the mechanism of promoting the anti-inflammatory eicosanoid synthesis.39, 40 Additionally, we have also demonstrated a U-shaped association between alcohol consumption and RA risk, in which women with moderate alcohol consumption had the lowest risk of RA.2 A recently published nested case-control studies reported that high sodium consumption was associated with increased risk of RA particularly among smokers,41, 42 and experimental studies have suggested that sodium may be able to promote the development of autoimmune diseases through the induction of pathogenic T helper 17 (Th17) in a process mediated by serum glucocorticoid kinase 1 (SGK1).43, 44 Finally, legumes, nuts and whole grains, which are abundant in dietary fiber and antioxidants may have potential to mitigate the RA risk.33
In the current study, we have found that the protective effects of higher diet quality on RA risk are strongest among women < age 55 years. This result is similar to our previous finding that the positive association between obesity and RA is also restricted among women with RA diagnosed at younger age. This may imply different risk factors for earlier and later onset RA. It is proposed that later onset RA may be different from earlier onset RA in terms of genetic predisposition and immune dysfunction, and hormonal changes with menopause can be a source of RA activation in older female patients.45
The strengths of our study include the use of two large cohort studies with long-term follow-up, dietary pattern analysis, repeated assessments of diet and potential confounders, and high rates of follow-up. There are several limitations in our study. First, residual confounding by other unmeasured or unknown factors might still exist, and better diet quality could be a marker for healthy lifestyle. However, because we have carefully adjusted for major lifestyle factors, it’s unlikely that the residual confounding completely explain the observed inverse associations. Second, the AHEI-2010 is a hypothesis-driven dietary index based on observed associations between single dietary components and the risk of major diseases, 13 that was not specifically derived for RA. The AHEI-2010 did not incorporate all dietary items that may have beneficial effects for lowering RA risk. Thus it is not clear that AHEI-2010 is the optimal dietary index for RA prevention. Third, the pooled analysis may generate spurious results when the distribution of exposure variable are different in each study (Simpson’s paradox). 46, 47 However, NHS and NHS II have the similar research designs, so it is unlikely that our findings were severely biased by Simpson’s paradox. Fourth, since the dietary information was collected every 4 years, we were unable to capture the variation of diet intake within this interval. However, the use of cumulative average dietary intake largely reduced the impact of day-to-day variation and could better reflect the long-term intake. Fifth, the majority of RA cases in our cohorts were diagnosed from 45 to 65 years old, we were unable to have enough RA cases to repeat the analyses among more extreme age groups, such as age<45 years or >65 years. Lastly, the dietary quality within these cohorts of mostly white, well-educated health professionals may not be representative of the dietary quality in the United States and other countries. However, it is unlikely that the underlying biologic mechanisms differ substantially by race. The observed associations of AHEI with other chronic diseases in NHS cohorts are very similar to other studies around the world.48
CONCLUSIONS
In summary, the results from this study indicate that greater long-term adherence to a healthy dietary pattern may reduce RA risk in women, particularly seropositive RA diagnosis at age 55 years or younger. Further studies are warranted to replicate our findings.
Supplementary Material
Table 2.
Pooled hazard ratios (95% CI) for incident RA according to cumulative AHEI-2010 quartile in Nurses’ Health Study (NHS, 1984–2012) and Nurses’ Health Study II (NHSII, 1991–2013)
| AHEI-2010 quartile
|
p-for trend‡ |
||||
|---|---|---|---|---|---|
| Q1 | Q2 | Q3 | Q4 | ||
| All RA | |||||
| Case/person-years | 253/917,059 | 261/919,818 | 255/920,405 | 238/920,822 | |
| Age adjusted model 1 | 1.00 | 0.97(0.82 to 1.16) | 0.96(0.81 to 1.15) | 0.86(0.72 to 1.03) | 0.11 |
| Multivariable-adjusted model 2 (main)* | 1.00 | 0.97(0.81 to 1.15) | 0.95(0.80 to 1.14) | 0.85(0.70 to 1.02) | 0.08 |
| Multivariable-adjusted model 3† | 1.00 | 0.97(0.82 to 1.16) | 0.97(0.81 to 1.15) | 0.87(0.73 to 1.05) | 0.16 |
|
| |||||
| Seropositive RA | |||||
| Case/person-years | 155/904,493 | 161/907,167 | 160/907,363 | 148/909,298 | |
| Age adjusted model 1 | 1.00 | 0.98(0.78 to 1.22) | 1.00(0.80 to 1.24) | 0.89(0.71 to 1.12) | 0.35 |
| Multivariable-adjusted model 2 (main)* | 1.00 | 0.98(0.78 to 1.23) | 1.00(0.80 to 1.25) | 0.88(0.70 to 1.12) | 0.34 |
| Multivariable-adjusted model 3† | 1.00 | 0.98(0.79 to 1.23) | 1.02(0.81 to 1.27) | 0.92(0.72 to 1.16) | 0.53 |
|
| |||||
| Seronegative RA | |||||
| Case/person-years | 97/914,049 | 100/917,055 | 95/917,425 | 90/918,075 | |
| Age adjusted model 1 | 1.00 | 0.96(0.73 to 1.28) | 0.91(0.68 to 1.20) | 0.82(0.61 to 1.09) | 0.15 |
| Multivariable-adjusted model 2 (main)* | 1.00 | 0.95(0.72 to 1.26) | 0.88(0.66 to 1.17) | 0.78(0.58 to 1.06) | 0.10 |
| Multivariable-adjusted model 3† | 1.00 | 0.95(0.72 to 1.26) | 0.89(0.66 to 1.18) | 0.81(0.60 to 1.09) | 0.14 |
Hazard ratios were calculated by using time-varying Cox proportional hazards models
Adjustment for age, cohort, smoking (never, past, current 1–14 cigarettes/d, current ≥15 cigarettes/d), age at menarche (<12, 12, >12 years), parity and breast feeding (nulliparous, parous/no breastfeeding, parous/1–12 months breastfeeding, parous/ >12 months breastfeeding), hormone use (pre-menopausal, post-menopausal with never use, current use and past use) and total energy (quintiles)
Additional adjustment for BMI (<20, 20–22.9, 23–24.9, 25–29.9, ≥30kg/m2)
p for trend was derived from tests of linear trend across categories of AHEI scores by treating the median value of each category as a continuous variable
Acknowledgments
Funding: This study was supported by the National Institutes of Health under Award Number AR061362, AR049880, AR052403, AR047782, AR059073, AR066953, AR069688, AR066109, and CA186107, CA176726, CA49449, CA67262. Dr. Sparks was also supported by the Rheumatology Research Foundation Scientific Development Award. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Footnotes
Contributors: YH and BL designed the study, analyze the data and wrote the first draft. SM contributed to the data analysis. BL, EWK, KHC, JAS and FBH were involved in data collection. All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. YH and BL had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Competing interests: None declared
References
- 1.Helmick CG, Felson DT, Lawrence RC, Gabriel S, Hirsch R, Kwoh CK, et al. Estimates of the prevalence of arthritis and other rheumatic conditions in the United States. Part I. Arthritis Rheum. 2008;58(1):15–25. doi: 10.1002/art.23177. [DOI] [PubMed] [Google Scholar]
- 2.Lu B, Solomon DH, Costenbader KH, Karlson EW. Alcohol consumption and risk of incident rheumatoid arthritis in women: A prospective study. Arthritis Rheumatol (Hoboken, NJ) 2014;66(8):1998–2005. doi: 10.1002/art.38634. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Linos A, Kaklamani VG, Kaklamani E, Koumantaki Y, Giziaki E, Papazoglou S, et al. Dietary factors in relation to rheumatoid arthritis: a role for olive oil and cooked vegetables? Am J Clin Nutr. 1999;70:1077–82. doi: 10.1093/ajcn/70.6.1077. [DOI] [PubMed] [Google Scholar]
- 4.Rosell M, Wesley AM, Rydin K, Klareskog L, Alfredsson L Group E study. Dietary fish and fish oil and the risk of rheumatoid arthritis. Epidemiology. 2009;20(6):896–901. doi: 10.1097/EDE.0b013e3181b5f0ce. [DOI] [PubMed] [Google Scholar]
- 5.Pattison DJ, Silman aJ, Goodson NJ, Lunt M, Bunn D, Luben R, et al. Vitamin C and the risk of developing inflammatory polyarthritis: prospective nested case-control study. Ann Rheum Dis. 2004;63(7):843–7. doi: 10.1136/ard.2003.016097. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Cerhan JR, Saag KG, Merlino LA, Mikuls TR, Criswell LA. Antioxidant micronutrients and risk of rheumatoid arthritis in a cohort of older women. Am J Epidemiol. 2003;157:345–54. doi: 10.1093/aje/kwf205. [DOI] [PubMed] [Google Scholar]
- 7.Pattison DJ, Symmons DPM, Lunt M, Welch A, Luben R, Bingham SA, et al. Dietary risk factors for the development of inflammatory polyarthritis: evidence for a role of high level of red meat consumption. Arthritis Rheum. 2004;50(12):3804–12. doi: 10.1002/art.20731. [DOI] [PubMed] [Google Scholar]
- 8.Hu Y, Costenbader KH, Gao X, Al-Daabil M, Sparks JA, Solomon DH, et al. Sugar-sweetened soda consumption and risk of developing rheumatoid arthritis in women. Am J Clin Nutr. 2014;100:959–67. doi: 10.3945/ajcn.114.086918. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Shapiro JA, Koepsell TD, Voigt LF, Dugowson CE, Kestin M, Nelson JL. Diet and rheumatoid arthritis in women: a possible protective effect of fish consumption. Epidemiology. 1996;7:256–63. doi: 10.1097/00001648-199605000-00007. [DOI] [PubMed] [Google Scholar]
- 10.Pedersen M, Stripp C, Klarlund M, Olsen SF, Tjønneland AM, Frisch M. Diet and risk of rheumatoid arthritis in a prospective cohort. J Rheumatol. 2005;32(7):1249–52. [PubMed] [Google Scholar]
- 11.Cerhan JR, Saag KG, Criswell LA, Merlino LA, Mikuls TR. Blood transfusion, alcohol use, and anthropometric risk factors for rheumatoid arthritis in older women. J Rheumatol. 2002;29(2):246–54. [PubMed] [Google Scholar]
- 12.Hu FB, Hu FB. Dietary pattern analysis: a new direction in nutritional epidemiology. Curr Opin Lipidol. 2002;13(1):3–9. doi: 10.1097/00041433-200202000-00002. [DOI] [PubMed] [Google Scholar]
- 13.Chiuve SE, Fung TT, Rimm EB, Hu FB, McCullough ML, Wang M, et al. Alternative Dietary Indices Both Strongly Predict Risk of Chronic Disease. J Nutr. 2012;142(C):1009–18. doi: 10.3945/jn.111.157222. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Akbaraly TN, Ferrie JE, Berr C, Brunner EJ, Head J, Marmot MG, et al. Alternative healthy eating index and mortality over 18 y of follow-up: Results from the Whitehall II cohort. Am J Clin Nutr. 2011;94(1):247–53. doi: 10.3945/ajcn.111.013128. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Fung TT, Hu FB, McCullough ML, Newby PK, Willett WC, Holmes MD. Diet quality is associated with the risk of estrogen receptor-negative breast cancer in postmenopausal women. J Nutr. 2006;136(2):466–72. doi: 10.1093/jn/136.2.466. [DOI] [PubMed] [Google Scholar]
- 16.Reedy J, Mitrou PN, Krebs-Smith SM, Wirfält E, Flood A, Kipnis V, et al. Index-based dietary patterns and risk of colorectal cancer: the NIH-AARP Diet and Health Study. Am J Epidemiol. 2008;168(1):38–48. doi: 10.1093/aje/kwn097. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Belin RJ, Greenland P, Allison M, Martin L, Shikany JM, Larson J, et al. Diet quality and the risk of cardiovascular disease: the Women’s Health Initiative (WHI) Am J Clin Nutr. 2011;94(1):49–57. doi: 10.3945/ajcn.110.011221. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Fung TT, McCullough M, van Dam RM, Hu FB. A prospective study of overall diet quality and risk of type 2 diabetes in women. Diabetes Care. 2007;30(7):1753–7. doi: 10.2337/dc06-2581. [DOI] [PubMed] [Google Scholar]
- 19.McCullough ML, Feskanich D, Stampfer MJ, Giovannucci EL, Rimm EB, Hu FB, et al. Diet quality and major chronic disease risk in men and women: Moving toward improved dietary guidance. Am J Clin Nutr. 2002;76(6):1261–71. doi: 10.1093/ajcn/76.6.1261. [DOI] [PubMed] [Google Scholar]
- 20.Cohen L, Curhan GC, Forman JP. Influence of age on the association between lifestyle factors and risk of hypertension. J Am Soc Hypertens. 2012;6(4):284–90. doi: 10.1016/j.jash.2012.06.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Janssen I. Influence of age on the relation between waist circumference and cardiometabolic risk markers. NutrMetab Cardiovasc. 2009;19:163–9. doi: 10.1016/j.numecd.2008.06.013. (1590–3729 (Electronic)) [DOI] [PubMed] [Google Scholar]
- 22.Tate RB, Manfreda J, Cuddy TE. Ann Epidemiol. 7. Vol. 8. Elsevier; 2016. Jul, The Effect of Age on Risk Factors for Ischemic Heart Disease; pp. 415–21. [DOI] [PubMed] [Google Scholar]
- 23.Esler MD, Thompson JM, Kaye DM, Turner AG, Jennings GL, Cox HS, et al. Effects of Aging on the Responsiveness of the Human Cardiac Sympathetic Nerves to Stressors. Circulation. 1995;91(2):351–8. doi: 10.1161/01.cir.91.2.351. [DOI] [PubMed] [Google Scholar]
- 24.Hogikyan RV, Supiano Ma. Arterial alpha-adrenergic responsiveness is decreased and SNS activity is increased in older humans. Am J Physiol. 1994;266(5 Pt 1):E717–24. doi: 10.1152/ajpendo.1994.266.5.E717. [DOI] [PubMed] [Google Scholar]
- 25.Lu B, Hiraki LT, Sparks JA, Malspeis S, Chen CY, Awosogba JA, et al. Being overweight or obese and risk of developing rheumatoid arthritis among women: a prospective cohort study. Ann Rheum Dis. 2014;73(11):1914–22. doi: 10.1136/annrheumdis-2014-205459. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Camacho EM, Lunt M, Farragher TM, Verstappen SMM, Bunn DK, Symmons DPM. The relationship between oral contraceptive use and functional outcome in women with recent-onset inflammatory polyarthritis: Results from the Norfolk Arthritis Register. Arthritis Rheum. 2011;63(8):2183–91. doi: 10.1002/art.30416. [DOI] [PubMed] [Google Scholar]
- 27.Ignacio J, Villa-Blanco Jaime C-A. Elderly onset rheumatoid arthritis: Differential diagnosis and choice of first-line and subsequent therapy. Drugs and Aging. 2009;26(9):739–50. doi: 10.2165/11316740-000000000-00000. [DOI] [PubMed] [Google Scholar]
- 28.Salvini S, Hunter DJ, Sampson L, Stampfer MJ, Colditz GA, Rosner B, et al. Food-based validation of a dietary questionnaire: the effects of week-to-week variation in food consumption. Int J Epidemiol. 1989;18(4):858–67. doi: 10.1093/ije/18.4.858. [DOI] [PubMed] [Google Scholar]
- 29.Fung TT, Malik V, Rexrode KM, Manson JE, Willett WC, Hu FB. Am J Clin Nutr. 4. Vol. 89. American Society for Nutrition; 2009. Sweetened beverage consumption and risk of coronary heart disease in women; pp. 1037–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Schulze MB, Manson JE, Ludwig DS, Colditz GA, Stampfer MJ, Willett WC, et al. Sugar-sweetened beverages, weight gain, and incidence of type 2 diabetes in young and middle-aged women. JAMA. 2004;292:927–34. doi: 10.1001/jama.292.8.927. [DOI] [PubMed] [Google Scholar]
- 31.Feskanich D, Rimm EB, Giovannucci EL, Colditz GA, Stampfer MJ, Litin LB, et al. Reproducibility and validity of food intake measurements from a semiquantitative food frequency questionnaire. J Am Diet Assoc. 1993;93:790–6. doi: 10.1016/0002-8223(93)91754-e. [DOI] [PubMed] [Google Scholar]
- 32.Giovannucci E, Stampfer MJ, Colditz G, Rimm EB, Willett WC. Relationship of diet to risk of colorectal adenoma in men. J Natl Cancer Inst. 1992;84(2):91–8. doi: 10.1093/jnci/84.2.91. [DOI] [PubMed] [Google Scholar]
- 33.Hu Y, Costenbader KH, Gao X, Hu FB, Karlson EW, Lu B. Mediterranean diet and incidence of rheumatoid arthritis in women. Arthritis Care Res (Hoboken) 2014;67(5):597–606. doi: 10.1002/acr.22481. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Karlson EW, Chibnik LB, Tworoger SS, Lee I-M, Buring JE, Shadick Na, et al. Biomarkers of inflammation and development of rheumatoid arthritis in women from two prospective cohort studies. Arthritis Rheum. 2009 Mar;60(3):641–52. doi: 10.1002/art.24350. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF, Cooper NS, et al. The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum. 1988;31(3):315–24. doi: 10.1002/art.1780310302. [DOI] [PubMed] [Google Scholar]
- 36.Karlson EW, Mandl LA, Hankinson SE, Grodstein F. Do breast-feeding and other reproductive factors influence future risk of rheumatoid arthritis? Results from the Nurses’ Health Study. Arthritis Rheum. 2004;50(11):3458–67. doi: 10.1002/art.20621. [DOI] [PubMed] [Google Scholar]
- 37.Wolf AM, Hunter DJ, Colditz GA, Manson JE, Stampfer MJ, Corsano KA, et al. Reproducibility and validity of a self-administered physical activity questionnaire. Int J Epidemiol. 1994;23:991–9. doi: 10.1093/ije/23.5.991. [DOI] [PubMed] [Google Scholar]
- 38.Benito-Garcia E, Feskanich D, Hu FB, Mandl La, Karlson EW. Protein, iron, and meat consumption and risk for rheumatoid arthritis: a prospective cohort study. Arthritis Res Ther. 2007;9(1):R16. doi: 10.1186/ar2123. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Di Giuseppe D, Wallin A, Bottai M, Askling J, Wolk A. Long-term intake of dietary long-chain n-3 polyunsaturated fatty acids and risk of rheumatoid arthritis: a prospective cohort study of women. Ann Rheum Dis. 2013:1–5. doi: 10.1136/annrheumdis-2013-203338. [DOI] [PubMed] [Google Scholar]
- 40.Moghaddami M, James M, Proudman S, Cleland LG. Synovial fluid and plasma n3 long chain polyunsaturated fatty acids in patients with inflammatory arthritis. Prostaglandins Leukot Essent Fatty Acids. 2015;97:7–12. doi: 10.1016/j.plefa.2015.02.005. [DOI] [PubMed] [Google Scholar]
- 41.Salgado E, Bes-Rastrollo M, de Irala J, Carmona L, Gómez-Reino JJ. High Sodium Intake Is Associated With Self-Reported Rheumatoid Arthritis: A Cross Sectional and Case Control Analysis Within the SUN Cohort. Medicine (Baltimore) 2015;94(37):e0924. doi: 10.1097/MD.0000000000000924. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Sundström B, Johansson I, Rantapää-Dahlqvist S. Interaction between dietary sodium and smoking increases the risk for rheumatoid arthritis: results from a nested case-control study. Rheumatology. 2014:1–7. doi: 10.1093/rheumatology/keu330. [DOI] [PubMed] [Google Scholar]
- 43.Binger KJ, Gebhard M, Heinig M, Rintisch C, Schroeder A, Neuhofer W. High salt reduces the activation of IL-4 and IL-13 stimulated macrophages. J Clin Invest. 2015;(26):1–3. doi: 10.1172/JCI80919. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Binger KJ, Linker RA, Muller DN, Kleinewietfeld M. Sodium chloride, SGK1, and Th17 activation. Pflugers Archiv European Journal of Physiology. 2015:543–50. doi: 10.1007/s00424-014-1659-z. [DOI] [PubMed] [Google Scholar]
- 45.Woodworth T, Ranganath V, Furst DE. Aging health. 2. Vol. 9. Future Medicine; 2013. Apr 1, Rheumatoid arthritis in the elderly: recent advances in understanding the pathogenesis, risk factors, comorbidities and risk–benefit of treatments; pp. 167–78. [Google Scholar]
- 46.Bravata DM, Olkin I. Simple pooling versus combining in meta-analysis. Eval Health Prof. 2001;24(2):218–30. doi: 10.1177/01632780122034885. [DOI] [PubMed] [Google Scholar]
- 47.Rücker G, Schumacher M. Simpson’s paradox visualized: the example of the rosiglitazone meta-analysis. BMC Med Res Methodol. 2008;8:34. doi: 10.1186/1471-2288-8-34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Schwingshackl L, Hoffmann G. J Acad Nutr Diet. 5. Vol. 115. Elsevier; 2016. Nov 5, Diet Quality as Assessed by the Healthy Eating Index, the Alternate Healthy Eating Index, the Dietary Approaches to Stop Hypertension Score, and Health Outcomes: A Systematic Review and Meta-Analysis of Cohort Studies; pp. 780–800.e5. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.