Study Population
This is a cross-sectional study based on baseline data from Ravansar Non-Communicable disease (RaNCD) prospective study in western Iran in 2020. The RaNCD study is part of a prospective epidemiological research study in Iran (PERSIAN). Ravansar is a district with urban and rural areas which is located in the west of Iran and in Kermanshah province with a population of about 50,000. The initial phase data was collected in 2014, and 10,000 adults between the ages of 35 and 65, who were registered as permanent residents of Ravansar were included in this cohort study. RaNCD study methodology and design with details have been published in 2019 [22]. Participants included all subjects from the first phase of the RaNCD study. For this study, the exclusion criteria were as follows: pregnancy, chronic diseases such as hepatitis B, hepatitis C, cancer, thyroid disorder and the case that their information was incomplete.
Data Collection
Data collection and all measurements of anthropometry and biochemical were conducted and assessed in the RaNCD cohort site. Participants were invited to the cohort center and the questionnaires were completed by trained experts. Demographic information and personal habits were completed face-to-face in the digital cohort questionnaire.
Anthropometry Measurements
We used the Bio-Impedance Analyzer BIA (Inbody 770, Inbody Co, Seoul, Korea) to measure body weight with a precision of 0.5 kg and BSM 370 (Biospace Co, Seoul, Korea) to measure height with the precision of 0.1 cm. Body composition components including BMI, Body fat mass (BFM), visceral fat area (VFA), Waist hip ratio (WHR) and WC were measured by BIA. WC also was measured with a flexible measuring tape at a level midway between the lowest rib margin and the iliac crest.
Biochemical Factors
For collection of the blood sample, serum concentrations of liver enzymes and lipid profiles were measured after 8–12 hours of fasting.
Blood Pressure
We measured blood pressure using a manometer cuff and stethoscope from both arm in the seated position and after 10 minutes of rest for two times from each arm with an interval of 5 minutes. Then a mean of both systolic and diastolic blood pressure was reported.
Fatty Liver Index
FLI was first introduced by Bedogni et al. in 2006 using the bootstrapped stepwise logistic regression analysis [14], with thirteen variables (including gender, age, ethanol intake, Alanine aminotransferase (ALT), Aspartate transaminase (AST), GGT, BMI, WC, sum of four skinfolds, glucose, insulin, TG and cholesterol) that 4 variables remained as predictors in the equation:
FLI= [e 0.953×log(e) (TG) + 0.139× BMI+ 0.718× log (e) (GGT) + 0.053 × WC – 15.745] / [ 1+ e 0.953× log(e) (TG) + 0.139× BMI+ 0.718× log (e) (GGT) + 0.053 × WC – 15.745] × 100.
The accuracy measured with area under the receiver operator characteristic curve (AUROC) of the FLI was 0.83 (95% CI: 0.825 to 0.842) in detecting fatty liver [20]. The FLI ranges from 0 to 100. Thus FLI scores of <30 and FLI≥ 60 indicated the absence or presence, respectively, in fatty liver with a good diagnostic accuracy [14].
Assessment of DII
Dietary information derived from the Food Frequency Questionnaire (FFQ) was used to calculate DII scores for subjects. Shivappa et al. found that 45 foods items were associated with one or more of the inflammatory including Interleukin-1b (IL-1b), Interleukin- 6 (IL-6), Tumor Necrosis Factor-a (TNF-a) or C-reactive protein (CRP) or anti-inflammatory markers including Interleukin-4 (IL-4) and Interleukin-10 (IL-10). They scored the inflammatory potential for each food parameter according to whether it increased inflammatory or decreased anti-inflammatory markers (+1), or it decreased inflammatory or increased anti-inflammatory markers (-1), or had no effect (0) on the level of inflammatory or anti-inflammatory markers. They calculated global mean and standard deviation for each of the 45 food parameters based on 11 data sets from 11 countries in different parts of the world [9].
In the present study, according to the food parameters in the Iranian questionnaire, we calculated DII score based on 31 food parameters, foods and nutrients that we were able to use in this study as follow: vitamin A, vitamin B6, vitamin B12, vitamin C, vitamin D, vitamin E, folic acid, niacin, iron, zinc, selenium, magnesium, beta-carotene, caffeine, thiamin, riboflavin, onion, garlic, tea, omega-3, omega-6, trans fat, saturated fats (SFAs), cholesterol, mono-unsaturated fatty acids (MUFA), poly-unsaturated fatty acids (PUFA), fiber, protein, total fat, carbohydrate and energy.
To calculate DII score for each subject, we subtracted the "standard global average" from the value consumed per person and then divided by the "global standard deviation" to obtain the Zscore for each food parameter. We used the global means ± SDs from the Shivappa et al. study [9]. Then we converted these values to a centered percentiles score to minimize the risk of skewness. The inflammatory score for each of the food parameters was calculated by this method, and then the inflammatory score of all parameters was summed to calculate the overall DII score; that this score could be positive or negative. The more positive DII scores indicate more pro-inflammatory diets and more negative scores imply more inti-inflammatory diets [9, 23].
Statistical Analysis
All analyzes in this study were performed using Stata version 14.1 software (Stata Corp, College Station, TX, USA). General characteristics, anthropometric indices and biochemical factors of participants across quartiles of DII score were reported as mean± standard deviation for continuous variables and as percentages for qualitative variables. The normality was checked using the Kolmogorov–Smirnov test. To comparisons of differences across DII quartiles, we used one-way ANOVA test. Analysis of linear regression was used to determine associations between DII score and FLI, anthropometric indices adjusted for the following confounding factors: age, sex, total energy, smoking, alcohol use and physical activity. Variables with p-value <0.2 in univariable analysis were entered into multivariable linear model. For statistical analyses, a p-value of <0.05 with 95% confidence intervals (CIs) was considered significant.