The association between dietary inflammatory index with endometriosis: NHANES 2001-2006

Endometriosis (EM) is the presence of functional endometrial-similar tissue outside of the uterus, which may lead to the occurrence of many chronic diseases such as pelvic pain, dysmenorrhea, and infertility [ 1 ]. As a chronic estrogen-dependent disease, EM has become increasingly prevalent in women, accounting for 6–10%, during reproductive age [ 2 ]. In addition to the direct health care costs such as surgery, monitoring tests, and hospitalization, EM also exerts a substantial mental stress regarding infertility [ 3 ]. The search for new prevention and management strategies to control EM is crucial. Food intake is inextricably bound to a person’s life, and a high anti-inflammatory diet reportedly improves the body’s inflammatory status [ 4 ]. Several studies have demonstrated that dietary intakes such as red meat, soya, and alcohol are closely involved with the onset of EM [ 5 ]. In EM patients, a reasonably balanced diet may be helpful to relieve pain through its interaction with viscerosensory input [ 6 ]. The dietary inflammatory index (DII), developed by the University of South Carolina, is widely used to measure the inflammatory potential of a person’s diet [ 7 ]. By summing inflammatory index scores across the dietary parameters, an individual’s diet is categorized within a range from anti-inflammatory to pro-inflammatory [ 8 ]. Moreover, recent studies suggested that an increased DII not only affects the physical health of the patients, increasing the incidence of endometrial cancer [ 9 ], but also has a significant effect on mental health [ 10 ]. However, to our knowledge, there is limited information available regarding the correlation between DII and EM. Therefore, with the help of the public data from the National Health and Nutrition Examination Survey (NHANES), this study aimed to examine the connection between DII and EM.

A total of 4232 participants were enrolled, comprising 3936 controls and 296 cases. Table 1 displays the baseline characteristics of participants in each group. Compared to the controls, the case group was more likely to have a higher distribution of DII (pro-inflammatory diet), age, education, and PIR. In our study, the whites were more prone to suffer EM than other races. Values are presented as mean ± SD for continuous variables, and P-value was calculated by the weighted linear regression. Values are presented as percent (%) for categorical variables, and P-value was calculated by weighted chi-square test. DII: dietary inflammatory index; BMI: body mass index; PIR: poverty income ratio. Weighted generalized logistic regression analysis was performed to probe the relationship between DII and EM, and a positive correlation was identified ( Table 2 ). When compared to the lowest quartile of DII, the fourth quartile had a higher prevalence of EM in the crude model [Odds ratio (OR) = 1.43, 95% confidence interval (CI) = 1.00–2.05, P = 0.059], model 1 (OR = 1.54, 95% CI = 1.08–2.20, P = 0.022), and model 2 (OR = 1.61, 95% CI = 1.09–2.37, P = 0.018). With the escalation of DII in all models, the OR for EM increases as well (P for trend < 0.05). Crude: no covariates were adjusted; Model 1: only age and race were adjusted; Model 2: age, age at menarche, BMI, race, education, and PIR were adjusted. DII: dietary inflammatory index; EM: endometriosis; BMI: body mass index; PIR: poverty income ratio; OR: odds ratio; CI: confidence interval. To resolve the heterogeneity between groups, subgroup analysis was conducted for different covariates (age, race, education, and PIR). Participants older than 35 years were more prone to suffer EM in all models (P = 0.005, P = 0.003, P = 0.019), while those under 35 years, did not show a statistically significant association (P = 0.595, P = 0.609, P = 0.652). For participants with EM in model 2, positive and significant associations were observed in the white (OR = 1.17, 95% CI = 1.05–1.31, P = 0.007), college or above education (OR = 1.22, 95% CI = 1.05–1.41, P = 0.013) and high PIR (OR = 1.19, 95% CI = 1.01–1.39, P = 0.041) groups. In contrast, among Black (OR = 1.02, 95% CI = 0.75–1.37, P = 0.914), Mexican American (OR = 1.14, 95% CI = 0.94–1.38, P = 0.189), and other races (OR = 1.16, 95% CI = 0.65–2.06, P = 0.614), the relationship between DII and EM was not statistically significant in model 2. The same conclusion was found in participants with low education (OR = 1.19, 95% CI = 0.98–1.46, P = 0.095) or poor economic conditions (OR = 1.01, 95% CI = 0.86–1.19, P = 0.918). Meanwhile, no interaction with DII was detected in model 2 (P for interaction > 0.05), indicating that the correlation between DII and EM did not vary by age, race, education, or PIR ( Table 3 ). Crude: no covariates were adjusted; Model 1: only age and race were adjusted; Model 2: age, age at menarche, BMI, race, education, and PIR were adjusted. PIR: poverty income ratio; OR: odds ratio; CI: confidence interval. *Stratified variables themselves were also not adjusted in the subgroup analysis. After adjusting for all covariates (model 2), smoothing splines were used to visually demonstrate the relationship between DII and EM. From the overall smoothing curve, a nonlinear relationship between DII and EM (log-likelihood ratio = 0.018) was revealed. When the DII (> 1.76) increased, the odds of EM prevalence increased (OR = 1.33, 95% CI = 1.11–1.60, P = 0.002). Given the critical value of the age category, the results were then stratified by age (< 35, ≥ 35) for further analysis. For participants aged < 35 years, the association trend was not statistically significant (OR = 1.04, 95% CI = 0.90–1.21, P = 0.596). However, a significant nonlinear relationship (log-likelihood ratio = 0.020) was identified in those aged ≥ 35 years (OR = 1.11, 95% CI = 1.01–1.22, P = 0.036); the probability of EM prevalence increased (OR = 1.35, 95% CI = 1.12–1.64, P = 0.002) as per one unit change in the DII (> 1.47) ( Fig 2 and Table 4 ). (A) The red line represents the correlation between DII and EM, and the blue line stands for the 95% CI (adjusted for age, age at menarche, BMI, race, education, and PIR). (B) In age-stratified analysis, the red line represents age < 35 years while the blue line represents age ≥ 35 years (adjusted for age at menarche, BMI, race, education, and PIR). DII: dietary inflammatory index; EM: endometriosis; BMI: body mass index; PIR: poverty income ratio; CI: confidence interval. Age, age at menarche, BMI, race, education, and PIR were adjusted. DII: dietary inflammatory index; EM: endometriosis; BMI: body mass index; PIR: poverty income ratio; OR: odds ratio; CI: confidence interval.

In middle-aged and older women (aged ≥ 35 years), exists a non-linear relationship between DII and EM. This may provide new insight into the role of diet in the prevention and treatment of EM. However, further studies are needed to identify the causal relationship and intrinsic mechanism between DII and EM.

The NHANES ( https://www.cdc.gov/nchs/nhanes ) is an ongoing cross-sectional survey that evaluates the diet and health of the noninstitutionalized US civilian population [ 11 ]. Using a complex multistage probability sampling design, a representative sample of the U.S. population was selected for the NHANES. Participants were interviewed in their homes, followed by a physical examination and an interview in a mobile examination center. Data for this study were obtained from three consecutive 2-year NHANES cycles (2001–2002, 2003–2004, 2005–2006). Since a complete assessment was not available for all participants, only those who met the predefined criteria were enrolled in our study. The inclusion criteria included complete information on: 1) demographics (including age, race, education, and poverty) and body mass index (BMI) variables 2) raw dietary components 3) EM. A total number of 4232 participants were included in this study ( Fig 1 ). This investigation was granted by the National Health and Nutrition Examination Survey Ethics Review Board, and no external ethic approval was required for this study. After screening, a total number of 4232 participants were enrolled including 3936 controls (without EM) and 296 cases (with EM). NHANES: National Health and Nutrition Examination Survey. With reference to recent related studies [ 12 , 13 ], the following variables were introduced as covariates: age, age at menarche, education (categorized as less high school, high school, and college or above), race (categorized as White, Mexican American, Black and other races), poverty income ratio (PIR), and BMI. Age at menarche was obtained via using the question “How old were you when you had your first menstrual period?”. Demographic data (age, education, race, and PIR) were obtained through a household interview survey. PIR was categorized into three groups: low (< 1.35), medium (1.35 to < 3.0), and high (≥ 3.0) [ 14 ]. BMI information was measured in the mobile examination center, and recoded into four categories: underweight ( 30 kg/m 2 ). As a literature-based scoring algorithm, the DII is now widely used as a special parameter for evaluating overall dietary inflammation. Based on the effect of 45 different food parameters on inflammatory biomarkers, an individual diet’s overall inflammatory potential score was ranked from the most anti-inflammatory to the most pro-inflammatory [ 15 ]. Due to the inherent restrictions of the NHANES data, dietary intake was only available for 28 out of the 45 food parameters [ 16 ]. However, because vitamin D intake data collection began in 2007, only 27 dietary parameters were enrolled in this study including: carbohydrates; protein; total fat; alcohol; fiber; cholesterol; saturated fat; monounsaturated and polyunsaturated fatty acids; n-3 and n-6 polyunsaturated fatty acids; niacin; vitamins A, B1, B2, B6, B12, C, D, E; iron; magnesium; zinc; selenium; folic acid; beta carotene; caffeine; energy. Through conducting a 24-h dietary recall interview at the mobile examination center, 27 components of each participant consumed in a 24-h period were acquired. DII contained in the diet was acquired using an inbuilt function in the ‘nhanesR’ package. By subtracting the standard mean of each food parameter and dividing it by the standard deviation, the individual’s Z-score value was measured and converted to centered proportions. They were multiplied by the responding inflammatory effect index and summed up, yielding a total DII score for each participant [ 17 ]. For further analyses, the DII was equally divided into four groups, ranging from the lowest quartile (maximally anti-inflammatory) to the highest quartile (maximally pro-inflammatory). EM was diagnosed based on a questionnaire-based survey asking: “Has a doctor or other health professional ever told you that you had endometriosis (age at interview 20–54 years)?” Those participants who answered yes were considered cases. Participants who answered no were considered controls. All data were analyzed via R statistical (version 4.1.1) and EmpowerStats software ( http://www.empowerstats.net ). For an impartial evaluation, weighted data were analyzed based on the available guidelines of NHANES. For continuous variables, statistics were calculated using weighted logistic regression to generate P values. For categorical variables, P values were determined by a weighted chi-square test. Three multiple weighted logistic regression models were developed to explore the relationship between DII and EM: the crude model was not adjusted for any covariate; model 1 was adjusted for age and race; model 2 was adjusted for all covariates. A smoothing curve was performed to visually estimate the association between DII and EM. P < 0.05 was considered statistically significant for all analyses.