Factors associated with premenstrual syndrome — A survey of new female university students

Abstract

Premenstrual syndrome (PMS) can impact an individual's interpersonal relationships, social interactions, productivity, lifestyle, school performance and emotional well‐being. This study was designed to explore the factors associated with PMS in new female university students in Taiwan. The test battery included a self‐administered structured questionnaire, the five‐item brief symptoms rating scale, the Pittsburgh Sleep Quality Index and the Chinese Premenstrual Symptom Questionnaire. Additionally, details of the participants' lifestyles and family and personal histories of physical illness were recorded. Serum lipids were also measured. Of all the participants (N = 1699), 39.85% were defined as having PMS. Using logistical regression analysis, we found a positive relationship between PMS and consuming more foods containing egg yolk, greater alcohol intake, poorer sleep quality, higher likelihood of psychiatric morbidity, family history of dyslipidemia, and a higher serum cholesterol level. The results show that PMS is prevalent among new female university students and that lifestyle and nutritional/metabolic factors may play a role in this disorder.

Introduction

Premenstrual syndrome (PMS) consists of a group of psychological and somatic symptoms that are related to the menstrual cycle [1]. Sufferers are usually affected during the luteal phase of their menstrual cycle, with the symptoms disappearing by the end of menstruation.

The majority of women experience at least one PMS symptom during their menstrual cycle, but are still able to function normally at work and at home. There is, however, evidence that women with PMS experience higher levels of daily and traumatic life stress [[2], [3], [4], [5]], and some experience severe symptoms that are emotionally, behaviorally and physically disabling, particularly in the area of family and personal relationships, work productivity and social activities [6]. The extreme and predominantly psychological form of PMS is called premenstrual dysphoric disorder (PMDD).

Epidemiological surveys have estimated that the frequency of premenstrual related symptoms is quite high (80–90%) [[7], [8]], and that in about 5% of women the symptoms are so severe that they interfere with personal and social relationships or work, in many cases requiring pharmacological treatment [9]. PMS may lead to an increase in healthcare utilization [10], and there is therefore a need for the public and clinicians to have a greater understanding of this syndrome. There are still some clinicians who do not accept PMS as a serious clinical condition, however, and as a result they are unsympathetic towards affected women [11].

Although the incidence of PMS is widespread and among menstruating women of all ages, the majority of women who seek medical assistance are over 30 years of age [12], and usually report that they have had symptoms for several years. While Sternfeld et al. (2002) reported that the PMS severity score decreased with each year of age [13], few PMS studies have focused on subjects in late adolescence or young adulthood [[14], [15], [16]].

It is known that females with PMS report a poorer health‐related quality of life, and that PMS may result in increased healthcare utilization and decreased occupational productivity [10]. A poorer perceived quality of sleep has also been reported in females with PMS [17], as have a poorer lifestyle and greater psychiatric comorbidity [18]. In addition to depressed mood, it has been reported that women with PMS/PMDD experience increased appetite, expressed as increased food cravings and intake [[19], [20], [21]], as well as increased alcohol consumption. Reed et al. (2008) [22] also demonstrated that women with PMDD showed an increased desire for food items that were high in fat during the luteal phase of the menstrual cycle; however, few studies of this issue have been conducted in Asian countries [[23], [24], [25]], and very few have included dietary factors and younger women.

The aim of this study was to investigate the factors associated with PMS in new female university students. We hypothesized that associated factors related to lifestyle habits, potential psychiatric morbidity, and sleep quality may be significant for those subjects with PMS.

Methods

Study population

A total of 1917 female students (51.75% undergraduate and 48.25% postgraduate students) starting courses at one university in southern Taiwan were surveyed in September 2007, of whom 1699 fully completed the survey (response rate = 88.6%; 51.27% undergraduate and 48.73% postgraduate students). The remaining 218 participants were excluded from the study due to missing data. Participants were enrolled at the routine health examination conducted during university orientation for new students. They were invited to participate in this study after being told that they would be asked to complete a self‐administered questionnaire about their personal lifestyle habits and premenstrual symptoms. Before the study began, informed consent was obtained from all participants. As they only agreed to have their questionnaire and related examination results analyzed anonymously, any identifying information was kept confidential. The Ethical Committee for Human Research at the National Cheng Kung University Hospital approved the study protocol.

Assessment of personal lifestyle habits

Through a self‐reported questionnaire, the demographic characteristics, personal medical history, family history of physical illness, lifestyle habits, and various medical problems of the participants were evaluated. The 10‐item questionnaire on personal and family histories of physical illness contained questions about the following systemic diseases and conditions: hypertension, diabetes mellitus, dyslipidemia, hyperuricemia, thyroid diseases, stroke, ischemic heart disease, anxiety disorder, depression disorder and asthma, while the personal lifestyle habits investigated included skipping breakfast and consumption of coffee, tea and alcohol. Tea and alcohol drinking were defined as consumption two or more times per week. Skipping breakfast was defined as eating breakfast less than or equal to once per week.

Habitual exercisers were defined as individuals who participated in physical activity three times or more per week, with each session lasting for more than 10 minutes to the extent of either sweating or a significant increase in heart beat. Food intake preferences were collected using a semi‐quantitative food frequency questionnaire that included several categories, such as animal fat, animal viscera, deep‐fried food, fast food, foods containing egg yolk, fruit and food with sugar (see Table 1). A preference was defined as eating a particular item three times or more per week.

Table 1.

Comparison of clinical characteristics between students with and without premenstrual syndrome.

Predictors	PMS(+)	PMS(−)	Chi‐square or t test
	(n = 677)	(n = 1022)	χ2/t/Z	p‐value
Age (years)	21.67 ± 3.80	21.51 ± 4.13	−0.79	0.437
Degree (postgraduate)	349 (51.6)	479 (46.9)	3.34	0.067
Menstrual cycle regularity (no)	192 (28.4)	258 (25.2)	2.13	0.145
Breakfast eating (<3 times/wk)	99 (14.6)	111 (10.9)	5.34	0.021
Food intake (≥3 times/wk)
Animal fat	137 (20.2)	234 (22.9)	1.68	0.195
Foods containing egg yolk	365 (53.9)	481 (47.1)	7.59	0.006
Fried food	81 (11.9)	85 (8.3)	5.74	0.017
Dessert	144 (21.2)	185 (18.1)	2.40	0.121
Drinks containing sugar	281 (41.5)	366 (35.8)	5.63	0.018
Fast food	190 (28.1)	243 (23.8)	3.90	0.048
Fruit	142 (21.0)	166 (16.2)	6.28	0.012
Drinking habit (≥3 times/wk)
Coffee	590 (87.1)	907 (88.7)	1.02	0.312
Tea	397 (58.7)	642 (62.8)	2.85	0.091
Alcohol	659 (97.3)	1014 (99.2)	9.51	0.002
Cigarette smoking
History of smoking	12 (1.8)	8 (0.8)	3.43	0.064
Current smoking	6 (0.9)	4 (0.4)	1.71	0.191
No habitual exercise (<3 times/wk)	241 (35.6)	297 (29.1)	7.58	0.006
PSQI (≥6)	410 (60.5)	416 (40.7)	60.02	<0.001
BSRS‐5 (≥6)	242 (35.7)	207 (20.3)	46.14	<0.001
Suicidal ideation	45 (6.7)	33 (3.3)	10.20	0.001
Family history of dyslipidemia	57 (8.4)	47 (4.6)	9.47	0.002
History of psychiatric disorder
Depression	4 (0.6)	1 (0.1)	3.34	0.068
Anxiety	2 (0.3)	0 (0.0)	3.07	0.084
Anthropometric and lab measurements
BMI (kg/m2)	20.57 ± 2.90	20.64 ± 2.92	−0.43	0.668
Total cholesterol (mg/dl)	175.73 ± 31.20	171.35 ± 28.10	2.91	0.004
HDL‐cholesterol (mg/dl)	64.57 ± 13.06	63.36 ± 12.66	1.84	0.065
Triglyceride (mg/dl)	65.76 ± 29.53	64.39 ± 29.50	0.91	0.365

Data expressed as number (percentage) or mean ± SD.

+ = with; − = without; BMI = body mass index; BSRS‐5 = Brief Symptom Rating Scale‐5; PMS = premenstrual syndrome; PSQI = Pittsburgh Sleep Quality Index.

Former and current alcohol drinkers were assigned to the alcohol group; similarly, former and current smokers were assigned to the smoking group.

Measurement of psychological symptoms and sleep quality.

The instruments in this domain included the five‐item Brief Symptoms Rating Scale (BSRS‐5) [[26], [27]], the one‐item suicide idea questionnaire, and the Chinese version of the Pittsburgh Sleep Quality Index (PSQI) [28].

The BSRS‐5 is a self‐reported scale that is modified from the symptom checklist‐90‐R and the 50‐item BSRS [29], and contains five psychopathology items [26] that were used in this work to identify possible psychiatric cases from the sample. The optimal cut‐off point of the BSRS‐5 was 5/6, and the sensitivity and specificity were 82.6% and 81.8%, respectively [[26], [30]]. The presence of suicidal ideation (one‐item) was also recorded.

The PSQI is a self‐administered questionnaire for the subjective evaluation of sleep quality during the previous month and contains 19 self‐rated questions yielding seven components: subjective quality of sleep, sleep latency, sleep duration, sleep efficiency, sleep disturbance, use of sleep medication, and daytime dysfunction. Each component is scored from 0 to 3, giving a global PSQI score between 0 and 21, with higher scores indicating a lower quality of sleep and a global PSQI score equal to 6 or more indicating a poor sleeper. The questionnaire has a diagnostic sensitivity of 90%, and specificity of 87% in distinguishing between good and poor sleepers [28].

The Chinese Premenstrual Symptoms Questionnaire (PMSQ) is a self‐rated tool used to screen for premenstrual symptoms for most cycles in the previous year. The PMSQ (Cronbach's α = 0.76) reflects the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition Text Revision (DSM‐IV‐TR) criteria for PMDD syndrome, and translates into an 11‐item symptom checklist with two categories: primary symptoms (four items) and other symptoms (seven items). Subjects with PMS are defined as those with at least five of the 11 symptoms, including at least one of the primary symptoms in category A according to the DSM‐IV‐TR criteria, and whose symptoms are noted 1 week before menstruation and begin to subside within a few days after the onset of the follicular phase and disappear after menstruation. The subjects who met these criteria were designated as the PMS group in this study, and the others were designated as the non‐PMS group.

Serum lipids and other variables

Using a direct method in an automatic biochemical analyzer (Model 7600, Hoffmann‐La Roche Inc., USA), total cholesterol (coefficient of variation [c.v.] 2.2–2.8%) and triglyceride (c.v. 2.6–3.5%) levels were determined enzymatically, and the concentration of high‐density lipoprotein cholesterol was measured enzymatically after dextran sulfate–magnesium precipitation (c.v. 2.1–2.9%). Body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared.

Statistical analyses

Group differences in categorical variables and numeric variables were analyzed using the Chi‐square test and the t test or Mann‐Whitney test, respectively. Subsequently, logistic regression was used to analyze the association between the multiple associated factors and PMS. Variables with p < 0.05 in the univariate analysis were included in the multivariate models. The two‐tailed significance level was set at 0.05. All of the analyses were carried out using SPSS software (Version 17, SPSS Inc., Chicago, IL, USA).

Results

The mean age of the subjects was 21.58 ± 4.00 years. Six‐hundred‐seventy‐seven female participants were included in the PMS group (39.85%), and the characteristics of PMS and non‐PMS groups are shown in Table 1. The PMS symptom profiles of the two groups are shown in Table 2. Poor sleep quality, greater preferences for foods containing egg yolk, deep‐fried food, drinks containing sugar, fast food, fruits and alcohol, lower likelihood of habitual exercise, greater suicidal ideation, higher possible psychiatric morbidity, family history of dyslipidemia, and higher cholesterol levels were all significantly associated with possible PMS. The results of multiple logistic regressions are presented in Table 3, and these show that eating egg yolk, drinking alcohol, poorer quality of sleep, higher possible psychiatric comorbidity, family history of dyslipidemia and higher cholesterol levels were significantly associated with possible PMS.

Table 2.

Comparison of item scores of the premenstrual symptoms questionnaire between students with and without premenstrual syndrome.

Item	Total	PMS(+)	PMS(−)
	1699 (100.0)	677 (39.8)	1022 (60.2)
Depressive mood	401 (23.6)	361 (53.4)	40 (3.9)
Anxiety	658 (38.7)	533 (79.4)	125 (12.2)
Unstable mood	859 (50.6)	601 (88.9)	258 (25.2)
Easy to get angry	518 (30.5)	423 (62.6)	95 (9.3)
Loss of interest in usual exercises	438 (25.8)	313 (46.3)	125 (12.2)
Shorter attention	482 (28.4)	343 (50.7)	139 (13.6)
Fatigue easily	1110 (65.3)	614 (90.8)	496 (48.5)
Appetite changes	434 (25.5)	274 (40.5)	160 (15.7)
Changes in duration of sleep	536 (31.5)	359 (53.1)	177 (17.3)
Nearly out of control	126 (7.4)	118 (17.5)	8 (0.8)
Physical symptoms	1184 (69.7)	582 (86.1)	602 (58.9)

Data expressed as number (percentage).

+ = with; − = without; PMS = premenstrual syndrome.

Table 3.

The odds ratio and 95% confidence intervals (CI) for the factors associated with premenstrual syndrome based on logistic regression analysis.

Associated factors	Odds ratio	95% CI	p‐value
Demographic and health variables
Foods containing egg yolk (≥3 vs. <3 times/wk)	1.24	1.00–1.55	0.05
Alcohol consumption (≥3 vs. <3 times/wk)	2.85	1.18–6.84	0.02
Assessments
Pittsburgh Sleep Quality Index score (≥6 vs. <6)	1.89	1.52–2.36	<0.05
Brief Symptom Rating Scale‐5 (≥6 vs. <6)	1.87	1.46–2.40	<0.05
Family history of dyslipidemia (yes vs. no)	1.87	1.20–2.90	0.01
Lab measurements
Serum cholesterol level	1.01	1.00–1.01	<0.05

−2 log likelihood = 1947.14, Nagelkerke R square = 0.075.

Discussion

In this study, 39.85% of new female university students, including undergraduates and postgraduates, were reported as having PMS. With regard to the high prevalence of premenstrual symptoms in women [[7], [8]], the definition of PMS in this study was based on the DSM‐IV criteria for PMDD. The impaired functionality of participants was difficult to validate in this study due to using a self‐rated scale, the PMSQ. The definition of PMS in this study is therefore broader than that used in other works [[7], [8], [31]]. In addition, the prevalence rates of PMS seem to vary in different populations. Derman et al. (2004) reported that 61.4％ of Turkish adolescent girls could be characterized as suffering from PMS [14], while Tabassum et al. (2005) [16] and Nisar et al. (2008) [15] found that around 50％ of college girls in Pakistan had PMS, and Chayachinda et al. (2008) [32] demonstrated that the prevalence of PMS in Thai nurses was 25.1%. Moreover, Potter et al. (2009) interviewed 2836 French women and reported 12.2% of participants with moderate or severe PMS [31]. In this study, 39.85% of the participating Taiwanese female university students were found to have PMS. Due to the differences in the definition of PMS, target populations, social and cultural contexts, and screening tools, however, it is difficult to directly compare the findings of the various studies.

In this work, PMS was found to be associated with poorer quality of sleep and higher possible psychiatric morbidity. Some of the findings were consistent with those of other studies of psychiatric morbidity in college students with PMS [[33], [34]]. Although sleep disturbance is listed as one of the defining criteria for a diagnosis of PMS in the DSM‐IV, there has been minimal research on the nature and severity of these premenstrual sleep disturbances. Based on retrospective reports, Mauri et al. (1988) found that women with PMS reported poorer sleep quality in the luteal phase than in the follicular phase [35]; however, Parry et al. (1995) found no differences between the late luteal phase and the follicular phase in terms of the duration of sleep or time spent awake. Barker et al. (2007) also demonstrated that although perceived poor sleep quality is a characteristic of severe PMS, sleep composition as assessed by polysomnographic measures and quantitative electroencephalographic analysis does not differ in association with PMS expression in the late luteal phase. Although no changes in sleep architecture have been observed in women with more severe PMS, the poorer the quality of sleep, the greater the severity of PMS would be. Borenstein et al. (2003) also claimed that PMS significantly affects health‐related quality of life, of which quality of sleep is an important item [10]. Moreover, PMS may cause increased utilization of healthcare and lead to decreased occupational productivity.

The association between alcohol intake and PMS observed in this work was also reported in the study carried out by Gold et al. (2007), which showed that alcohol intake is positively associated with premenstrual anxiety and mood changes in individuals with PMS [36]. This contrasts with the results reported by Bertone‐Johnson et al. (2009) showing that alcohol use is not strongly associated with the development of PMS [37]. Interestingly, a family history of dyslipidemia and higher cholesterol levels in the PMS group were observed in this study. Nagata et al. (2004) reported that high intake of fats may be associated with PMS [25], and Reed et al. (2008) recently reported that women with PMDD had an increased desire for food items high in fat and ate more calories/fat in their luteal phase compared to women without PMDD [22]. The finding of this study that there is an association between PMDD and greater consumption of foods containing egg yolk, a higher plasma cholesterol level and a family history of dyslipidemia seems plausible, because a higher cholesterol level may be due to an increased desire for food items high in calories/fat in females with PMS, and the family history of dyslipidemia may be due to sharing similar genes or a similar environment. Meanwhile, an association between PMS and hypertension has also been reported [38]. There was no significant difference in BMI between the PMS and non‐PMS subgroups, so further studies are needed in order to confirm and validate the relationships between nutritional/metabolic factors and PMS [39].

There are several limitations to this study, which suggest some avenues for future research. First, we did not conduct individual interviews to confirm the diagnoses of PMS and psychiatric morbidity, and the severity of PMS was not analyzed. Moreover, the validity of the PMSQ was not tested. Caution should therefore be taken with regard to generalizing this result. Second, we only identified the associated factors in this cross‐sectional study, and thus no causal relationship can be confirmed. Third, excluding new female university students who did not complete all of the questionnaires for analysis might have influenced the validity of the findings. Fourth, it is well known that symptoms of PMS occur during the luteal phase. The cross‐sectional design in this study and the use of a self‐administered questionnaire for the PMSQ mean that the PMS subgroups may have been confounded by mismatching participants who had similar discomforts in other phases. Finally, very few of the participants in either group were categorized as drinking alcohol, and this may reduce the validity of the finding that a significantly higher proportion of the PMS group consumed alcohol than the non‐PMS group.

Conclusion

In conclusion, our findings indicate that PMS is prevalent in new female university students, and that poor quality of sleep, psychological impairment, greater alcohol intake, more consumption of foods containing egg yolk, a family history of dyslipidemia, and higher cholesterol levels are related to PMS. Nutritional/metabolic factors may also play a role in PMS.

Acknowledgments

The authors wish to thank Professor Bih‐Ching Shu from the Department of Nursing, Miss Ching Lin Chu and Linda J Chang, and the students of the National Cheng Kung University who participated in the study.