This cross sectional study was conducted among 733 adolescent girls (aged 12-18 years) in January 2015. Participants were selected from several schools in the cities of Mashhad and Sabzevar, in northeastern Iran by a random cluster sampling method. Adolescents with history of autoimmune diseases, cancer, metabolic bone disease, hepatic or renal failure, cardiovascular disorders, malabsorption or thyroid, parathyroid, adrenal diseases and anorexia nervosa or bulimia were excluded from the study. In addition, participants who were included, did not receive anti-inflammatory, anti-depressant, antidiabetic, or anti-obesity drugs, vitamin D or calcium supplement, and any hormone therapy within the last 6 months. Before the start of the study, written informed consent forms confirmed by the ethical committee of Mashhad University of Medical Sciences, in Mashhad, Iran (ethical code number: 931188) were completed by all participants and their parents.
Demographic and anthropometric assessments
Demographic variables were assessed using a demographic questionnaire administered by trained interviewers. Demographic variables were age, smoking status, menstruation status, medical history, supplement intake, taking psychological treatment and chronic diseases. Participants' physical activity was assessed using a modifiable activity questionnaire (MAQ) (20). Physical activity level was calculated based on metabolic equivalent task minutes per week. Anthropometric variables such as weight, height and waist circumference (WC) were measured by the trained investigators using the standard protocols. Body mass index (BMI) computed as weight (kg) divided by square of height (m2).
Dietary intakes were estimated by a validated food-frequency questionnaire (FFQ) with 147 food items. The validity and reliability of this FFQ were confirmed previously (24). To complete the FFQ, a face-to-face interview was performed by a trained dietitian. To evaluate the frequency of food items intake during the last year, the participants answered the questions about their daily, weekly, monthly and yearly intake. The reported portion sizes in the FFQ were converted to grams using household measures, and the energy and nutrient intakes were calculated using the Nutritionist IV software (25).
Assessment of depression
Depression was assessed via a Persian version of the Beck Depression Inventory (BDI). This questionnaire has 21 items evaluating various symptoms of depression including feelings of guilt, feelings of hopelessness, sadness, crying, sleep disturbance, fear and loss of appetite over the past 2 weeks. The range of scores for the BDI was between 0 and 63 points. scores 0-13, 14-19, 20-28 and 29-63 refer to no depression, mild depression, moderate depression, and severe depression respectively. The validity and reliability of this questionnaire were confirmed by previous studies (26, 27).
Quality of life assessment
To assess health-related QOL, the SF-12v2 questionnaire was used. This questionnaire is a short form of the SF-36 questionnaire and improved version of SF-12v1 (28). The validity and reliability of this questionnaire were approved in Iran (29). The questionnaire has 12 items which evaluate 8 domains of health including physical functioning, role limitations because of physical problems, bodily pain, general health, vitality, social functioning, role limitations because of emotional problems, and mental health. The range of QoL scores are between 0 (the worst QoL) to 100 (the best QoL). The median of the QoL score is 43. The subjects were categorized as high QoL if their scores were higher than 43.
Subjects were classified into four groups across quartiles of their food group scores including whole grain, refined grain, carbonated drink and fast foods. General characteristics and nutrient intake across quartiles of food group score were expressed as means ± SDs for continuous variables, and as numbers and percentages for categorical variables. To examine the differences between quartiles, one-way-ANOVA and Chi square test were used for continuous and categorical variable respectively. To investigate the relationship between quartiles of food group scores with depression and poor QoL, multivariate regression was conducted in the crude and adjusted models. Multivariate linear regression was used to explore the association between QOL score and depression with fast food, carbonated drink, whole grain and refined grain intake. Age and energy intake were adjusted in Model I. Additionally, BMI percentile was adjusted in Model II. Finally, physical activity, age, energy intake and BMI percentile were adjusted in the model III. All statistical analyses were conducted using the SPSS version 21. P-values less than 0.05 were considered as significant.