Major dietary patterns in relation to menstrual pain: a nested case control study

A nested case control study was conducted from April to July 2016 in a medical college located in Urmia, in the West Azarbaijan province of Iran. This University comprises 7 faculties (Medicine, Pharmacy, Paramedical, Health, Nursing and Midwifery, Dentistry and International Branch) and administers 5 teaching hospitals. The sample size was calculated by the following formula:

(n = the minimum sample size, N = population size (1972), z = 1.96 for 95% confidence level, p (the prevalence of dysmenorrhea [11]) = 0.74, q = 0.26 and d (the degree of precision) = 0.05). However, taking into account the possible sample loss, a total of 386 samples were estimated. Students were selected by a cluster sampling method. The research team distributed self-administered questionnaires among participants. Students with energy intakes outside the range of 500–5000 kcal/day [12] and those who did not answer more than 20% of Food Frequency Questionnaire (FFQ) items were excluded from the study (n = 93) and dietary patterns of 293 students were extracted. Out of 293 students, 60 subjects without dysmenorrhea and 60 subjects with 2nd to 3rd grade dysmenorrhea were randomly assigned to each of control and case groups, respectively. The inclusion criteria in this section was the presence of menstrual periods, being single, not having any chronic diseases, no history of abdominal surgery, no history of alcohol intake and cigarette smoking, no administration of OCPs, dietary supplements and herbal remedies. Body weight and height of both case and control groups were measured and a gynecologist performed abdominal examination and trans-abdominal ultrasonography using an ultrasound scanner (SIUI, CTS-5500, Guangdong, China) to rule out subjects with genital diseases associated with dysmenorrhea (e.g., congenital malformation of the mullerian system, cervical stenosis and ovarian cysts [1]) (n = 20). Flow chart of sampling procedure for the study is shown in Fig. 1.

The study questionnaire included five sections. In the first section, students were asked to state their demographic information. In the second section, they were asked about their menstrual characteristics over the last 6 months, including severity of dysmenorrhea, age at menarche, bleeding length, length and regularity of menstrual cycle, severity of menstrual flow and family history of dysmenorrhea. A verbal multidimensional scoring system [13] was used to assess the severity of dysmenorrhea. Based on the pain intensity and its effect on daily activity, symptoms, and the need for analgesics, dysmenorrhea was categorized into four grades: zero (none), 1 (mild), 2 (moderate) or 3 (severe). For the other menstrual characteristics, the questions of the Women’s Health Symptom Survey [14] were used. In the third section, physical activity level was assessed by the short form of International Physical Activity Questionnaire [15, 16]. In this instrument, students were asked to report the frequency and duration of their last 7 days vigorous and moderate activities, walking and sitting. According to the scores, students were categorized into three levels of physical activity: “low” if participants had low activities or did not meet moderate or high category criteria, “moderate” if they had vigorous activities for 3 days or more (at least 20 min per day), a combination of moderate activities and/or walking for 5 days or more (at least 30 min per day) or a combination of walking, moderate or vigorous activities for at least 5 days with MET (Metabolic Equivalent of Task)-minutes/week ≥ 600, and “high” if they had vigorous activities for at least 3 days with MET-minutes/week≥1500 or a combination of walking, moderate or vigorous activities for at least 7 days with MET-minutes/week≥3000. In the fourth section, participants’ depression and anxiety levels over the last 2 weeks were assessed by Patient Health Questionnaire-4 [17]. This instrument had 4 items: the first and second questions measured depression and the third and fourth questions measured anxiety. The response options were “not at all”, “several days”, “more than half days”, and “nearly every day”, which were scored as 0, 1, 2 and 3, respectively. Based on their total scores, students were categorised into four groups: normal (0–2), mild (3–5), moderate (6–7) and severe (9–12). In the fifth section, the participants’ dietary intakes for the last 6 months were assessed by a 115-item semi quantitative FFQ. This questionnaire was adapted from Block design [18] and Tehran Lipid and Glucose Study [19]. The frequency of food intakes options were “2-3 times per day”, “daily”, “4-6 times per week”, “2-3 times per week”, “weekly”, “2-3 times per month”, “monthly”, “1-5 times during the previous 6 month” and “never”. A picture of different portion sizes was given for some food items and for other items, standard household measures (for example “cup” for drinks) or standard units were used. Daily intake of food items (gram/day) was calculated using both frequencies and portion sizes. Total energy intake (kcal/day) of subjects was calculated according to the data available at USDA (United States Department of Agriculture) national nutrient database [20] or other sources [21, 22]. Then, food items were categorized into 30 separate groups based on the previous studies [23, 24] or similarities in their nutrient profile. The FFQ was validated with two 24/h dietary recalls and the correlation coefficients ranged from 0.08 (low fat dairy products) and 0.64 (for high fat dairy products) for food groups.

Body weight was measured (without shoes and in light clothing) to the nearest 0.01 kg using a digital scale (Beurer BF18 Digital Scale). Height was measured in the standing position without shoes to the nearest 0.1 cm. BMI was calculated by dividing weight in kilogram by their height in meters squared.

All statistical analyses were performed using IBM SPSS Software Package for Windows (version 20.0, Armonk, NY: IBM Corp). The Kolmogorov–Smirnov test was used to determine whether continuous variables follow a normal distribution. Independent-samples t-test, Mann-Whitney U test, Pearson chi squared test and Fisher’s exact test were used to compare continuous or categorical variables between case and control groups. To identify dietary patterns, principal component analysis was carried out and the factors were rotated by varimax rotation. According to factor eigenvalue greater than 1.8, the break point of the scree plot and factor interpretability, factors (dietary patterns) were noted. These factors were nominated based on food groups with factor loading> 0.2. After computing factor scores for each individual, we categorized participants by tertiles of scores and the first tertile was considered as a reference. We used logistic regression to calculate the crude and adjusted odds ratios (OR) and the 95% confidence intervals to interpret the association between dietary patterns and risk of dysmenorrhea. Adjusted ORs were calculated by adjusting for family history of dysmenorrhea. A P value of < 0.05 was considered statistically significant.