Participants: This cross-sectional study was a part of the mental and physical health assessment of university student (MEPHASOUS) project which was performed in 2012-2013. This project was aimed to determine the contributing factors to the health problems and unhealthy behaviors of Iranian university students. Details on the study design, sampling methods, and data gathering in the MEPHASOUS project have been published previously [21,22]. In the MEPHASOUS project, all students from 74 governmental universities (in the provinces of Iran), related to the Ministry of Science and Technology (MST), were asked to take part in the study. Inclusion criteria were the registration in a governmental university and being in an age range of ≥18 years. The data collection was conducted in the health centers of universities. All required data on demographic variables, physical activity, anthropometric measures, medical history, and dietary intakes were gathered from each student via pre-tested questionnaires. After merging the data and removing the students with missing information, 83,463 students remained for the final analysis. All participants signed the informed consent document. The ethics committee of the Ministry of Science and Technology, Tehran, Iran approved the whole project.
Physical activity assessment: Because of the high sample size of the MEPHASOUS study, we used a short question for the assessment of physical activity. Participants were asked how many times per week they do exercise with a moderate-to-high severity for 30 minutes. The response option for this question was: “rarely”, “1-2 times/wk”, “3-4 times/wk”, “>5 times/wk”. We considered students who had physical activity ≥3 times/wk as physically active and those with a rare physical activity as inactive. Also, having physical activity 1-2 times/wk was considered a moderate physical activity.
Assessment of primary headaches: Primary headaches were evaluated using a self-reported question by which students were asked whether they had experienced primary headaches (including migraine, tension-type headache, and non-classifiable headache) during the last 12 months. They were asked to not report the headaches related to cold, fever or any other type of illness. Students could select two options (yes/no). If yes, they were referred to a general practitioner, who was experienced in terms of neurovascular diseases, for further examination. Primary headaches were defined according to the criteria introduced by the International Classification of Headache Disorders-3 (ICHD-3) with the exceptions that the number of attacks and the duration of headaches were not included .
Dietary habits: In order to collect information on dietary intake of selected food groups and also breakfast consumption patterns, a self-administered validated dietary habits questionnaire was used. A previously published article revealed the reliability and validity of the questionnaire . Students were asked to report their dietary intakes for fresh and dried fruits, vegetables, dairy products, fast foods, sugar-sweetened beverages (SSBs), and sweets during the last year. In the questionnaire, the response categories were different for each food item based on its usual intake among the Iranian population. For example, the response categories for fruits which are frequently consumed by Iranians were in daily format (<1 serv/day, 1 serv/day, 2-3 serv/day, ≥4 serv/day), while these categories for sweets which are infrequently consumed were in weekly format (<1 serv/wk, 1 serv/wk, 2-3 serv/wk, ≥4 serv/wk). In addition to dietary intakes, students were asked to report the frequency of breakfast consumption in a week by the use of these options: <1 day/wk, 1-2 days/wk, 3-4 days/wk, ≥5 days/wk. Breakfast skipper was defined as individuals who consumed breakfast ≤4 days/wk.
The reliability of the dietary habits questionnaire was examined in a separate study which was done on a subgroup of 70 students in each center of the MEPHASOUS project (total: 1960 students) . In that study, a test-retest reliability process was applied in order to estimate the reliability of the questionnaire. Participants were asked to fill the dietary habits questionnaire two times with a 2–3 weeks’ interval. The correlation coefficients for all dietary habits between the two times were more than 0.60 indicating sufficient reliability of this questionnaire. The reliability and validity of the questionnaire were also confirmed by previous studies which applied this questionnaire for the assessment of dietary habits [22,24,25].
Assessment of other variables: We used a self-reported pre-tested questionnaire to gather data on age, gender (male/female), education (advanced diploma/bachelor of science (BSc)/master of science (MSc)/medical science (MD)/philosophy of doctor (Ph.D.)), marital status (single/married), occupation (having/not-having), health insurance (having/not-having), smoking (non-smoker/ex-smoker or current smoker), and current use of nutritional supplements (including Fe, Ca, vitamins and other nutritional supplements) (yes/no). Students who were in the advanced diploma, BSc and MD courses were defined as the under-graduate students and those students in the MSc and Ph.D. courses were considered as the graduate students. Since health insurance in Iran can cost a lot for people, it was considered as an index for the evaluation of economic status. Therefore, we considered students who had health insurance as economically “good” and those who did not have any type of health insurance as economically “weak”. In addition, sleep pattern was assessed by the two questions: “how is your pattern of sleeping and awaking?” and “how many hours do you sleep in a day?” The response options for the first question were “regular”, “irregular” and for the second question were: “<6 hours/day”, “6-8 hours/day”, “8-10 hours/day”, “>10 hours/day. Due to the probable influence of internet addiction on physical activity and maybe primary headaches, we assessed the time that each student spend for the use of internet-connected devices including computer, cell phone, and notebook in a day. They should answer the question by these choices: “rarely”, “<2 hours/day”, “2-4 hours/day”, and “>4 hours/day”. Using of these devices >4 hours/day were considered as frequent use. In order to measure anthropometric indices, we employed a standard procedure to measure weight and height. We calculated the body mass index (BMI) as weight (kg)/height (in square meters). Overweight and obesity were considered as the BMI of 25-30 and ≥30 kg/m2, respectively [26,27]. Blood pressure was measured in a seated position after a 5-min rest two times with a 20-minute interval. The average of two measurements was considered as the final systolic and diastolic blood pressure. Hypertension was defined as systolic blood pressure (SBP) ≥ 140 mmHg and diastolic blood pressure (DBP) ≥90 mmHg .
Statistical analysis: We first classified students according to three categories of physical activity (inactive, moderate, and active). One-way analysis of variance (ANOVA) was employed to assess differences in continuous variables among the levels of physical activity. To evaluate the distribution of categorical variables across the categories of physical activity, the Chi-square test was used. Binary logistic regression in multivariable-adjusted models was used to obtain odds ratios (ORs) of primary headaches across the levels of physical activity. This analysis was performed among the whole population and separately by gender and BMI status (<25/≥25 kg/m2). For creating the adjusted models, we first included age and gender (not included in the sex-stratified analysis) into the model (first model). Then, we included other variables including marital status, education, occupation, economic status, smoking, the use of internet-connected devices, sleep pattern, hypertension, supplement use, and breakfast skipping in the second model in addition to variables belong to model 1. Further adjustment was made for the consumption of fruits, vegetables, dairy products, fast foods, SSBs, and sweets in model 3. In the last model, we additionally controlled for BMI to obtain an obesity-independent association between physical activity and primary headaches. In this analysis, the reference group was considered those students who were physically inactive. Also, to compute the P-value for the trend of odds ratios across the levels of physical activity, we considered these levels as an ordinal variable in binary logistic regression. All statistical analyses were done using SPSS software (version 19.0; SPSS Inc, Chicago IL). P values were considered significant at <0.05.