This study has been conducted within the framework of the Tehran Lipid and Glucose Study (TLGS), which is an ongoing prospective population-based study to prevent non-communicable diseases. The layout and other information on TLGS were provided elsewhere .
The TLGS was initiated in March 1999. A multistage, stratified cluster random sampling technique was used to enroll > 15,000 participants ≥ 3 years from district 13 of Tehran. The population of this district is representative of the urban population of Tehran, the capital city of Iran. Since 1999, the participants of TLGS underwent assessments for anthropometric measures, medication use, medical history of CVD risk factors, lifestyle factors, sociodemographic factors, socioeconomic status, and biochemical and blood pressure measurements. This information was documented through face-to-face interviews with the local research team every 3 years. Up to now, 6 phases of the examinations have been performed. Phases II, III, IV, V, and VI were prospective follow-up studies conducted during 2002–2004, 2005–2008, 2008–2011, 2012–2015, and 2016–2018, respectively.
The current study used the baseline examination data from phase III of the TLGS (2006–2008) because of the small sample size for dietary assessment in phases I and II of the study and using 24-h recall. The subjects were followed up to phase VI of TLGS (2016–2018).
During phase III of the TLGS (2006–2008), medical history and physical examination were collected for 12,523 participants. Owing to the cost, complexity, and time involved in the collection of dietary data in a large population, a representative sample of 4920 participants was randomly selected based on their age and gender. Of 4920 participants, 3462 agreed to complete a food frequency questionnaire (FFQ). The characteristics of participants who completed the FFQ were similar to those of the total population in phase III of TLGS . For the current study, of 3462 participants, 3265 adults aged 19 years or older with complete data (demographic, anthropometric, biochemical, and dietary data) were selected from phase III (2006–2008). Moreover, individuals with MetS at baseline (n = 879), women who were pregnant or lactating at baseline and during follow-up (n = 28), and subjects with under- or over-reporting of energy intakes (daily energy intake < 500 and > 4200 kcal per day) (n = 115), participants following any specific diet as a result of their hyperlipidemia, hypertension, and hyperglycemia (n = 26), and subjects with missing biochemical and anthropometric measures related to diagnosis of MetS during follow up (n = 309) were excluded from the study. Final analysis was conducted on 1915 participants until 2018, with a response rate of 66%, during the 8.9 (Interquartile range: 7.98–9.69) year follow-up. The study protocol was approved by the Ethics Committee of the Research Institute for Endocrine Sciences (RIES) of Shahid Beheshti University of Medical Sciences, Tehran, Iran. Written informed consent was obtained from all participants.
Briefly, the participants’ weight, while being minimally clothed without shoes, was recorded using a digital scale (Seca 707; range: 0–150 kg; Seca GmbH, Germany) and recorded with accuracy of 100 g. Height was also measured in a standing position, without shoes, with shoulders in neutral alignment using a stadiometer (Seca 225; Seca GmbH, Germany) and recorded to the nearest 0.5 cm. Body mass index (BMI) was calculated as weight (kg) divided by the square of height (m2). Waist circumference (WC) was measured at the umbilical level using an un-stretched tape measure without any pressure to the body surface (accuracy, 0.5 cm).
Assessment of other variables
After participants rested in a sitting position for 15 min, blood pressure was measured using a standardized mercury sphygmomanometer (calibrated by the Iranian Institute of Standards and Industrial Research) on the right arm twice, at least 30 s apart, and the average of the two measurements was reported as the participant’s blood pressure. Demographic, lifestyle (smoking status and physical activity), socioeconomic status (education and employment), medication regimen (e.g., antihypertensive, lipid-lowering, and anti-diabetes drugs), and medical history were gathered using a questionnaire.
Physical activity was assessed using a modifiable activity questionnaire (MAQ) that included a list of all three forms of activities, including leisure time, job, and household activities. The frequency and amount of time spent per week on physical activity over the last year were recorded . The physical activity levels were expressed as metabolic-equivalent (MET) hours per week (MET-h/week) . The reliability and validity of the Persian version of the MAQ have been reported .
During face-to-face interviews with expert dietitians, a validated semi-quantitative FFQ was used to determine the frequency of daily, weekly, or monthly consumption of each food item during the previous year . The Iranian food composition table (FCT) was used to calculate macro- and micronutrient intake .
From the initial number of 1915 participants at baseline, 592 participants completed all 4 FFQs (at baseline and during follow-up in phases IV, V, and VI), 804 participants completed 3 FFQs (at baseline and in two of the three phases of the follow-up study), 316 participants completed 2 FFQs (at baseline and in one of the three phases of the follow-up study), and 203 participants did not complete any FFQs during follow-up (only at baseline). To impute missing values, last observation carried forward method was used. In the present study, due to the crucial effect of recent dietary intakes on the association between diet and chronic disease, we used an alternative approach according to the Hu et al. formula . This approach, which is more important than the baseline measures, adds more weight to the recent diet, reduces within-subject variability, and evaluates the long-term diet.
Definition of DIL and DII
Food insulin index (FII) refers to the incremental insulin area under the curve over 2 h in response to the consumption of a 1000-kJ (239 kcal) portion of the test food divided by the area under the curve after the ingestion of a 1000-kJ (239 kcal) portion of the reference food. The insulin index for 68 food items was obtained from studies by Bao et al.  (50 items), Bell et al.  (13 items), and Holt et al.  (5 items). The insulin index for three food items, including tea, coffee, and salt, was considered 0 because these foods’ energy, carbohydrate, protein, and fat content were close to 0. For the remaining 49 food items that were not available in the food lists of the mentioned studies, the FII of similar food items was used based on the correlation between their energy, fiber, carbohydrate, protein, and fat content. For example, both dates and raisins are dried fruits. The energy, carbohydrate, fat, protein, and fiber content of both fruits are similar. Therefore, we used the insulin index of raisins for dates. The 120 items of the FFQ, the source of the FII, and the FII value are presented in DII and DIL in Relation to MetS: The Shahedieh Cohort Study (available at www.jandonline.org). To determine DIL, first, the insulin load of each food was calculated using the following formula: Insulin load of a given food = insulin index of that food × energy content per 1 g of that food amount consumed (g/d). By summing the insulin load of each food, DIL was obtained for each person. DII was then calculated for each participant by dividing DIL by total energy intake.
For biochemical measurements, after 12–14 h of overnight fasting, venous blood samples were collected in vacutainer tubes and centrifuged within 30–45 min of collection. The fasting plasma glucose (FPG), high-density lipoprotein-cholesterol (HDL-C), and triglyceride (TG) levels were measured in the TLGS research laboratory on the day of sample collection, using a Selectra 2 autoanalyzer (Vital Scientific, Spankeren, the Netherlands) and commercial kits (Pars Azmoon Inc., Tehran, Iran). FPG level was measured using an enzymatic colorimetric method with the glucose oxidase technique. The inter- and intra-assay coefficients of variation (CV) at baseline and after follow-up were both below 2.3%. TG was also assayed using an enzymatic colorimetric method with glycerol phosphate oxidase. HDL-C was measured after the precipitation of apolipoprotein B-containing lipoproteins with phosphotungstic acid. In baseline and follow-up assays, both intra- and inter-assay CVs were below 2.1% and 3.0% for TG and HDL-C. All samples were analyzed when the internal quality control met the acceptable criteria.
Definition of MetS
According to the Joint Interim Statement, a MetS diagnosis requires the presence of three or more criteria , including (1) elevated glucose concentrations (FPG concentration ≥ 100 mg/dL) or treatment with anti-hyperglycemic medications; (2) elevated serum TG concentration (≥ 150 mg/dL) or treatment with anti-hypertriglyceridemia medications; (3) reduced serum HDL-C (< 50 mg/dL in women and < 40 mg/dL in men); (4) elevated blood pressure (≥ 130/85 mmHg) or treatment with anti-hypertensive medications; and (5) enlarged abdominal circumference (≥ 95 cm according to the population- and country-specific cut-off points for Iranian adults of both genders .
Definition of weight change
Percentage weight change was calculated by subtracting the baseline weight from the follow-up one and multiplying it by 100. Participants were categorized as those who lost weight (≥ 3%), those with weight stability (± 3%), and those who gained weight (≥ 3%) .
Data are reported as mean (SD) and median (25th and 75th percentiles) for continuous variables or percentages for categorical variables. DII and DIL were categorized into tertiles. Baseline characteristics and energy-adjusted dietary variables were described across the tertiles of DII and DIL, using the general linear model and Chi-square test for continuous and categorical variables, respectively. Moreover, Cox proportional-hazards regression models were used to estimate the hazard ratios (HRs) and their 95% confidence intervals (CIs) for the incidence of MetS and weight gain ≥ 3% across the tertiles of DII and DIL. The first model was crude, while the second model was adjusted for age, gender, smoking, physical activity, education levels, occupation status, total energy intake, and family history of diabetes, dietary fiber and dietary cholesterol (all variable that adjusted was at baseline). The third model was additionally adjusted for BMI at baseline. The linearity of trends was determined by integrating the median values of tertiles as continuous variables into the Cox regression models. Based on the multivariable Cox regression model, by joint classification, we estimated the HRs and 95% CIs for MetS, according to the weight changes, sex, physical activity levels and smoking status. All statistical analyses were performed in SPSS version 15.0 (SPSS Inc., Chicago, IL, USA), and P-values less than 0.05 were considered statistically significant.