Study population and measurements

The present study was conducted within the framework of the Tehran Lipid and Glucose Study (TLGS), an ongoing prospective study initiated in 1998 to evaluate non-communicable disease risk factors among 15 005 Tehranian adults aged ≥ 3 years^{(Reference Azizi, Zadeh-Vakili and Takyar20)}. In the third phase of the TLGS (2006–2008), there were 10 091 adults aged ≥ 19 years with a complete medical history and physical examination data. Among those, we excluded participants who had uncompleted dietary data (n 7036), participants with missing data on anthropometric measurements and plasma glucose (n 330), participants with under- or over-reports of energy intake (< 800 kcal/d or > 4200 kcal/d, respectively) (n 146), participants with pre-diabetes (n 382) or T2DM (n 284) at baseline and participants who lost to follow-up. The final analyses were conducted on data of 2042 adults free of pre-diabetes and 2295 adults free of T2DM (Fig. 1). The eligible participants were followed up to the sixth phase of the TLGS (2014–2017). The median periods of follow-up were 7·37 and 8·47 years from baseline, for pre-diabetes and T2DM, respectively.

Figure 1.

Flow chart of the study.

Anthropometric and demographic assessments

Data on anthropometric, demographic and biochemical variables were collected by trained interviewers of TLGS at baseline (2006–2008). The body weight and height of participants were measured to the nearest 100 g and 0·5 cm, using digital scales (Seca) and a tape metre, respectively. BMI was calculated as weight (kg) divided by the square of height (m²). Waist circumference (WC) was measured to the nearest of 0·1 cm, from midway between the lower border of the ribs and the iliac crest at the widest portion, using a soft measuring tape. To measure systolic and diastolic blood pressures, we used a standard mercury sphygmomanometer calibrated by the Iranian Institute of Standards and Industrial Research^{(Reference Askari, Asghari and Ghanbarian21)}. The blood pressure of participants was measured twice on the right arm with a 30-s interval between two measurements. The final blood pressure of participants was considered as the mean of the two measurements. The usual physical activity of participants was evaluated using the modifiable activity questionnaire (MAQ). Physical activity levels were expressed as metabolic equivalent minutes per week (MET-min/week) and categorised as low physical activity (scores ≤ 600 MET-min/week) and moderate and high physical activity (scores > 600 MET-min/week)^{(Reference Ainsworth, Haskell and Leon22)}. The reliability and validity of the Persian version of the MAQ were previously investigated^{(Reference Momenan, Delshad and Sarbazi23)}.

Biochemical measurements

A blood sample was taken after overnight fasting between 07.00 and 09.00 from all participants. All blood analysis was done at the research laboratory of the TLGS, using Pars Azmoon kits (Pars Azmoon Inc.) and a Selectra 2 auto-analyser (VitalScientific). Fasting serum glucose and 2-h serum glucose were measured by the enzymatic colorimetric method using glucose oxidase. Serum TAG was measured by the enzymatic colorimetric method using glycerol phosphate oxidase. HDL-cholesterol was measured after precipitation of the apo B-containing lipoproteins with phosphotungstic acid. Both inter- and intra-assay CV at baseline and follow-up phase were less than 5 %.

Dietary assessment

The dietary intake of participants over the previous year was assessed using a validated 168-item FFQ at baseline^{(Reference Mirmiran, Esfahani and Mehrabi24)}. Participants were asked to designate their intake frequency for each food item consumed during the past year on a daily, weekly or monthly basis. The frequencies were then converted to daily intakes, and portion sizes, reported in household measures, were converted to grams^{(Reference Hosseini-Esfahani, Jessri and Mirmiran25)}. Since the Iranian Food Composition Table is incomplete and has limited data on the nutrient content of raw foods and beverages, the United States Department of Agriculture (USDA) food composition table was used to obtain the amount of energy and macronutrients per gram of each type of food.

Fat quality indices

AI is the sum of SFA divided by the sum of MUFA and n-3 and n-6 PUFA.

$$AI ={{[ ( C12:0 + ( 4 \times C14:0 ) + C16:0 ) ]}\over{( \sum MUFA + \;\sum n - 6 + \;\sum n - 3 )}}$$

TI is attributed to the less anti-atherogenic effect of MUFA and n-6 PUFA, compared with the n-3 PUFA.

$$\small{\eqalign{& TI =\cr& {{{\left( {C14:0 + C16:0 + C18:0} \right)}}\over{{[\left( {0\cdot5 \times \sum MUFA} \right) + \left( {0\cdot5 \times \sum n - 6 + \left( {3 \times \sum n - 3} \right) + \left( {\sum n - 3/\sum n - 6} \right)} \right]}}}}}$$

HPI is the sum of the unsaturated fats divided by the saturated fats.

$$HPI = {{{\left( {\sum MUFA + \;\sum PUFA} \right)}}\over{{\left( {C12:0 + \left( {4 \times C14:0} \right) + C16:0} \right)}}}$$

Ratio of PUFA and SFA is the sum of PUFA divided by the sum of SFA.

$$ PUFA/SFA= {{{\sum PUFA}}\over{{\sum SFA}}}$$

Ratio of hypo- and hypercholesterolaemia (h:H) is the sum of hypocholesterolemic PUFA divided by the sum of myristic acid and palmitic acid.

$${\small{\eqalign{& (h/H)=\cr & {(C18:1\ n-9+C18:3\ n-6+C18:3\ n-3+C20:5\ n-3+C22:6\ n-3)\over(C14:0+C16:0)}}}}$$

Definition of terms and outcomes

Pre-diabetes was defined as impaired fasting glucose (fasting serum glucose levels of 5.56–6.95 mmol/l) or impaired glucose tolerance (2h-serum glucose values of 7.78–11.06 mmol/l). T2DM was defined as fasting serum glucose ≥ 7.00 mmol/l, 2h-serum glucose ≥ 11.11 mmol/l or self-reported taking of anti-diabetic medications⁽²⁶⁾. Having at least one parent or sibling with T2DM is considered a positive family history of T2DM. The diabetes risk score (DRS) was calculated as follows: systolic blood pressure (mm Hg) < 120 (0 point), 120 < systolic blood pressure < 140 (3 points) and systolic blood pressure ≥ 140 (7 points); family history of diabetes (5 points); WC:height ratio: < 0·54 (0 point), 0·54–0·59 (6 points) and ≥ 0·59 (11 points); TAG:HDL-cholesterol ratio: < 3·5 (0 point) and ≥ 3·5 (3 points); fasting serum glucose (mmol/l): < 5 (0 point), 5–5.5 (12 points) and 5.56–6.89 (33 points)^{(Reference Bozorgmanesh, Hadaegh and Ghaffari27)}.

Statistical analyses

The incidence of pre-diabetes and T2DM over the follow-up period were considered as dichotomous variables (yes/no) in the models. Mean (sd) values and frequencies (%) of baseline characteristics of participants were compared according to the incidence of pre-diabetes and T2DM, using ANOVA and χ ² test, respectively. Dietary FQI were categorised into tertiles. The first tertile of each category was considered as a reference.

Cox proportional hazard regression models with person-years as the underlying time metric were used to estimate hazard ratios (HR) and 95 % CI for the association between dietary FQI (AI, TI, HPI, PUFA:SFA and h:H) and risk of pre-diabetes and T2DM. The event date for cases was defined as the middle-time between the date of the follow-up visit at which pre-diabetes/T2DM was detected for the first time and the most recent follow-up visit preceding the diagnosis; the follow-up time was drawn from the difference between the calculated mid-time date and the date at which the subjects entered the study. The survival time was the interval between the first and the last observation dates for the censored and lost to follow-up subjects. The proportional hazard assumption of the multivariable Cox model was assessed using Schoenfeld’s global test of residuals. Sensitivity analysis was conducted after the exclusion of subjects with pre-diabetes at baseline in T2DM models. This approach was taken to assess if the results may have been influenced by the pre-diabetes subjects.

Potential confounders were determined using univariate analysis. Variables with P_E less than 0·2 in the univariate analyses, which were selected as confounders and adjusted in the Cox models, included sex (male/female), age (years), BMI (kg/m²), smoking (yes/no), physical activity level (low/moderate and high), DRS (continuous), socio-economic status (marital status, employment status and education level), total energy (kcal/d) and total fibre intake (g/d). Adjustment of DRS, as a continuous potential risk factor of T2DM, improved the stability of our models due to the limited number of outcomes.

All statistical analyses were performed using the Statistical Package for Social Science (version 20; IBM Corp.), with P-values < 0·05 being considered significant.