Psychosocial and menstrual correlates of premenstrual disorders: A community-based study.

SI and NM designed the study. MS and NM contributed to the patient data collection. SI, NM performed the data analysis and drafted the manuscript. MS, TT, YO, KK, SN, and NM critically revised the manuscript. All authors approved the final version of the manuscript.

This study was funded by the Ministry of Health, Labor and Welfare of Japan (grant no.: 23FB0301).

Of 6966 eligible participants, 23% met the criteria for PMS and 10% for PMDD. The prevalence of PMS was highest among those aged 19 years or below, and PMDD was highest for those in their 20s, with both declining in older age categories (Figure  1 ). Prevalence of premenstrual syndrome (PMS) and premenstrual dysphoric disorder (PMDD) by age category. This figure shows the percentage of participants meeting criteria for PMS or PMDD across age categories. PMS was most prevalent among participants aged ≤19 years, while PMDD had the highest prevalence among those in their 20s. Prevalence declined with increasing age. Table  1 summarizes participant characteristics by PMS/PMDD status. The PMDD group had the highest proportion of those who were married, co‐residing with children, and current and past smokers. They had the highest number of adverse childhood experiences (ACEs) and the lowest health literacy scores. The PMS group had the lowest proportion of married participants, the highest prevalence of high‐risk drinking, and similarly elevated ACE counts and lower health literacy scores compared with controls. Participant background characteristics. Note : BMI, calculated as weight in kilograms divided by the square of height in meters. Participant sociodemographic, lifestyle, and psychosocial characteristics stratified by PMS/PMDD status. Values are presented as n (%) unless otherwise indicated. Continuous variables are shown as mean (SD). Percentages may not total 100% due to rounding or missing responses. Abbreviations: BMI, body mass index; PMDD; premenstrual dysphoric disorder; PMS, premenstrual syndrome; SD, standard deviation. In multinomial logistic models (Table  2 ), increasing age was associated with a lower relative risk of PMS and PMDD. In age‐adjusted multinomial logistic models with the control group as the reference, current and past smoking were associated with a greater relative risk of PMS and PMDD compared with never smoking. High‐risk alcohol use was associated with PMS but not with PMDD. Being married and co‐residing with a child were associated with PMDD but not with PMS. Higher ACEs and maladaptive coping scores were each associated with increased relative risk of both PMS and PMDD (per 1‐point increase in maladaptive coping: PMS RRR = 1.09, PMDD RRR = 1.14), whereas higher health literacy was associated with a lower relative risk (per 1‐point increase: PMS RRR = 0.98, PMDD RRR = 0.97). Age‐adjusted associations of background and lifestyle characteristics with PMS and PMDD status. Note : BMI, calculated as weight in kilograms divided by the square of height in meters. Separate multinomial logistic regressions for each variable with the control group as reference; estimates are RRRs with 95% CIs. All models were adjusted for age except the Age model. Continuous predictors are per one‐unit increase unless stated. N varies due to item non‐response across questionnaire waves. The bold values indicates significance ( p  < 0.05). Abbreviations: ACE, adverse childhood experiences; BMI, body mass index; CI, confidence interval; PCI, confidence interval; PMDD; premenstrual dysphoric disorder; PMS, premenstrual syndrome; RRRs, relative risk ratios. Age model is unadjusted. Table  3 shows associations between PMS/PMDD status and psychosocial characteristics. Psychological distress (K6 ≥10) and poor sleep (PSQI ≥6) were each associated with a higher relative risk of PMS and PMDD. Greater negative spillover in both work‐home and home‐work domains was associated with a higher relative risk of PMS and PMDD. Descriptive distributions are provided in Table  S1 . Age‐adjusted associations between psychosocial characteristics and PMS/PMDD status. Note : Separate multinomial logistic regressions were used for each psychosocial variable with Control as the reference outcome. Estimates are presented as RRRs with 95% CIs. All models were adjusted for age. Continuous variables were modeled per 1‐unit increase; binary variables were defined as K6 ≥10 (psychological distress) and PSQI ≥6 (poor sleep quality). The bold values indicates significance ( p  < 0.05). Abbreviations: CI, confidence interval; K6, Kessler Psychological Distress Scale; PMDD, premenstrual dysphoric disorder; PMS, premenstrual syndrome; PSQI, Pittsburgh Sleep Quality Index; RRRs, relative risk ratios. Table  4 describes menstrual cycle characteristics. The average cycle length showed a modest inverse association with PMDD (per‐day increase: RRR = 0.98), and there was no clear association with PMS. Abnormal cycles were not associated with PMS or PMDD after adjustment. Menstrual symptom burden was substantially higher in both PMS and PMDD groups. Specificity of the menstrual cycle and each subscale of menstrual symptoms is shown in Table  S2 . Adjusted associations between menstrual characteristics and PMS/PMDD status. Note : BMI, calculated as weight in kilograms divided by the square of height in meters. Separate multinomial logistic regressions for each variable with the control group as reference; estimates are RRRs with 95% CIs. All models were adjusted for age, BMI, co‐residence with a child, pregnancy intention, exercise, smoking, alcohol use, and work status. A participant was defined as having abnormal cycles if their mean CL was outside the range of 24 to 38 days based on FIGO criteria. Menstrual symptoms were assessed based on experiences reported to occur in most cycles over the past year. The bold values indicates significance ( p  < 0.05) Abbreviations: BMI, body mass index; CI, confidence interval; CL, cycle length; PMDD, premenstrual dysphoric disorder; PMS, premenstrual syndrome; RRRs, relative risk ratios.

In this large community‐based study, PMS and PMDD were associated with diverse sociodemographic, psychosocial, and menstrual characteristics. PMS was most prevalent among those aged ≤19 years, while PMDD peaked in the 20s, with both prevalences declining across older age groups. Compared with controls, participants with PMDD were more likely to be married and co‐residing with children, whereas PMS was associated with high‐risk alcohol use. Both groups had higher odds of current and past smoking history, more ACEs, lower health literacy, higher maladaptive coping, greater psychological distress, poor sleep, and negative work and home interactions. Menstrual irregularity did not differ across groups, but both PMS and PMDD reported a greater number of menstrual symptoms. The prevalence of PMS (23%) and PMDD (10%) in the present study is consistent with international estimates. 1 Although retrospective tools such as the PSST have concerns of overestimating prevalence, studies using prospective tools report dropout rates of 30%–40%, 20 , 21 potentially underestimating actual cases. Moreover, prior studies have shown comparable proportions of asymptomatic individuals across both designs, suggesting the PSST's suitability for large‐scale population studies. 21 Younger age was associated with a higher relative risk of PMS and PMDD, consistent with prior reports. 4 , 6 Multiple factors may contribute to these age‐related differences, such as hormonal fluctuations, life stage‐related stress, or greater symptom awareness. An earlier menarche has also been implicated as a risk factor for premenstrual disorders. 22 However, current evidence on age remains mixed, 1 which may reflect variations in study populations, diagnostic tools, and cultural contexts when compared with previous studies. In contrast, neither condition was associated with BMI in our age‐adjusted analyses, differing from reports from Western populations that suggest a positive link between adiposity and premenstrual disorders. 4 In this population, BMI showed limited variability (mean 21.9 ± 3.8) with most participants within the normal range, consistent with national data. 23 The absence of association in this context suggests that BMI may not be a significant determinant of premenstrual disorders in populations with low prevalence of overweight and obesity. To our knowledge, this is the first study to identify an association between marital and child‐rearing status and PMDD. While childbirth is sometimes thought to alleviate menstrual symptoms, caregiving and other responsibilities may contribute to elevated stress levels, thereby influencing symptom severity. These findings suggest that life‐stage and relational factors should be considered in the clinical assessment and support of individuals with premenstrual disorders. Along with these life‐course factors, lifestyle behaviors were also implicated. Smoking was strongly associated with PMS and PMDD. While a recent meta‐analysis identified current smoking as a risk for premenstrual disorders, 24 our findings suggest that past smoking may also confer long‐term risk. PMS was also associated with high‐risk drinking, consistent with previous reports. 25 The relationship between these behaviors and premenstrual disorders may be bidirectional, with individuals experiencing premenstrual symptoms who might use smoking or alcohol as coping mechanisms for stress or discomfort, whereas these substances can, in turn, disrupt neurocircuitry and stress‐regulation pathways, potentially leading to the development or exacerbation of symptoms. 24 These findings highlight smoking and alcohol use as modifiable behaviors linked with premenstrual disorders, though causal direction cannot be inferred. Individuals with PMS and PMDD in our study reported a higher number of ACEs, consistent with prior research linking early trauma to premenstrual disorders. 5 Prior studies have shown that even moderate exposure can have dose‐dependent adverse effects on adult health and behavior, possibly through effects on epigenetic modifications, contributing to long‐term alterations in stress response systems. 26 , 27 Emotional abuse, in particular, has shown strong associations with affective disorders and may independently increase vulnerability to hormonal fluctuations. 5 Both PMS and PMDD were associated with lower health literacy scale and greater maladaptive coping. These characteristics may limit individuals' ability to interpret bodily changes, seek medical care, or adopt effective symptom management strategies. Previous digital interventions have demonstrated that app‐based education can enhance health literacy and reduce symptom burden, 28 suggesting that such approaches could be valuable for premenstrual disorder management. Psychological distress, poor sleep, and negative work and home spillover were also associated with a higher relative risk of PMS and PMDD. Although defined by symptoms limited to the luteal phase, the burdens of these disorders appear to extend throughout the cycle. Such associations may reflect shared vulnerabilities, including heightened stress reactivity and serotonergic dysfunctions. Neurocognitive studies have indicated altered top‐down control in individuals with premenstrual disorders, which may help explain the heightened burdens observed in affected individuals. 29 Menstrual cycle characteristics were also examined. Associations with abnormal cycles were insignificant for both groups, and only marginal differences in average cycle length were observed for the PMDD group. Based on real‐world cycle data, these findings contrast with previous studies using self‐reported menstrual irregularity, which reported greater irregularities among those with premenstrual symptoms. 8 This discrepancy may reflect differences in symptom perception, as individuals with premenstrual disorders are reported to hold more negative attitudes toward menstruation. 30 Because stable menstruation is often indicative of regularly occurring ovulation, and premenstrual disorders are primarily linked to ovulatory progesterone fluctuations, 1 these findings may underscore the role of regular ovulation and hormonal cycling in the development of premenstrual disorders. The observation that symptoms often subside during pregnancy or after menopause and do not occur before menarche 1 further supports the role of ovulatory processes in their etiology. Although premenstrual disorders are characterized by symptoms that resolve with the onset of menstruation, our findings show that affected individuals also experience greater symptom burden during menstruation itself. The total number of symptoms during menstruation and all subcategories of symptoms were associated with PMS and PMDD. While these conditions should be distinguished for their distinct etiology (menstrual pain is generally linked to prostaglandin activity, and premenstrual symptoms to progesterone fluctuations), those affected may experience only a brief symptom‐free interval each cycle, contributing to the substantial psychosocial burden observed in this population. Other gynecologic conditions, such as polycystic ovary syndrome (PCOS) and endometriosis, may also be associated with premenstrual disorders through distinct biological pathways. PCOS, characterized by anovulatory or irregular cycles, involves altered ovarian steroidogenesis and neurosteroid imbalances such as allopregnanolone, 31 which have been implicated in mood symptoms and may also contribute to premenstrual symptoms. 1 In contrast, endometriosis is primarily associated with inflammatory mechanisms and pain‐related mechanisms that may overlap with premenstrual symptoms. Some risk factors, such as smoking, have also been associated with endometriosis, 32 suggesting potential overlaps between these conditions. Although these conditions were not assessed in the present study, future analyses examining their overlap with PMS and PMDD could help elucidate the distinct pathways contributing to premenstrual disorders. This study has several limitations. First, the use of a retrospective screening tool may lead to overestimation of prevalence compared to provisional diagnosis. Second, due to the cross‐sectional design, causal inferences between outcomes cannot be established. Third, all measures were self‐reported, potentially leading to recall or reporting bias. Fourth, although the prevalence of smartphone users in Japan was 74.3% in 2021, 33 and is estimated to be even higher in the younger population, the participants were users of a menstrual tracking app, which may not fully represent the general population. Fifth, despite adjustments for key covariates, unmeasured confounding cannot be excluded. Finally, some models had a reduced sample size due to incomplete questionnaire data, which may have affected statistical power.

This study highlights the multifaceted burden associated with PMS and PMDD in a large community‐based cohort. Individuals with premenstrual disorders exhibited a higher relative risk associated with smoking, ACEs, maladaptive coping, and low health literacy. PMDD was additionally associated with marital and child‐rearing status, while PMS was linked to high‐risk drinking. Both groups showed higher psychological distress, poor sleep quality, negative work‐home interactions, and higher menstrual symptom burdens. These findings underscore the need for integrated interventions addressing broader psychosocial and behavioral factors.

Premenstrual disorders, including premenstrual syndrome (PMS) and premenstrual dysphoric disorder (PMDD), are characterized by physical, emotional, and behavioral symptoms that occur in the luteal phase of the menstrual cycle and subside with menstruation. 1 PMS is reported to affect approximately 20%–30% of women of reproductive age, while PMDD affects 2%–8%. 2 , 3 Despite their prevalence, these conditions are often underrecognized and undertreated, leading to impairments in daily functioning and reduced quality of life. Understanding their broader determinants is therefore critical for effective prevention and management strategies. Previous studies have suggested a range of risk factors, including higher body mass index (BMI, calculated as weight in kilograms divided by the square of height in meters), early life trauma, nicotine use, and comorbid psychological conditions. 1 , 4 , 5 While some studies suggest higher symptoms among younger individuals, 4 , 6 others report no apparent age‐related differences. 1 Evidence on broader background and psychosocial characteristics remains limited, and prior research has primarily focused on reduced work or academic outcomes. 7 Broader psychosocial domains, such as sleep disturbances, coping skills, and family or home‐life stress, have been less explored in the general population. Additionally, although prior studies reported increased menstrual irregularity among those with premenstrual symptoms, 8 few have used subjective, real‐world data to assess cycle characteristics of those with premenstrual disorders. The study examined background, psychosocial, and menstrual correlates associated with PMS and PMDD in a large community‐based sample. Using questionnaire responses and menstrual cycle logs from a widely used smartphone application, we aimed to identify factors associated with the likelihood of experiencing PMS or PMDD and to provide a comprehensive understanding of their broader health and societal implications.

The authors have no interests relevant to this article to disclose.

This cohort study utilized data from LunaLuna, a widely used menstrual tracking application in Japan with over 16 million downloads. Periodic questionnaires (Wave 1–6) were distributed via the app and linked to individual cycle logs (Figure  S1 ). All active users were informed of the research objectives, data collection methods, and participation procedures between January 23 and March 25, 2020. The study protocol was approved by the institutional review board of the National Center for Child Health and Development (approval no. 1985, July 15, 2020). Participation was voluntary, and informed consent was obtained electronically via the app interface. The Japanese version of the premenstrual symptoms screening tool (PSST), 9 , 10 was administered in Wave 2. The PSST includes 12 symptom items and five functional impairment items rated on a four‐point scale (“none” to “severe”). Classification followed the DSM‐IV criteria, and participants were categorized as PMDD (≥1 core symptom rated as “severe,” ≥4 symptoms rated “moderate” or “severe,” and ≥1 functional impairment rated as “severe”), PMS (meeting similar but less severe criteria), or control (all others). Background factors and psychosocial measures were obtained from multiple questionnaire waves (see Figure  S1 ). High‐risk drinking was defined as consuming an average of more than 20 g of pure ethanol per day. 11 Health literacy was measured using a validated 17‐item scale with higher scores indicating higher literacy. 12 Psychological distress was assessed using the K6 scale (scores ≥10 indicating distress). 13 Sleep quality was evaluated using the Pittsburgh Sleep Quality Index (PSQI), with scores ≥5.5 indicating poor sleep. 14 Coping strategies were assessed using the Brief COPE inventory. 15 Work‐home interactions were assessed using the validated Japanese version of the Survey Work–Home Interaction NijmeGen (J‐SWING), 16 , 17 which comprises four subscales (work‐to‐family and family‐to‐work positive/negative spillover). Cycle data were collected from January 2019 through March 2021. Cycle length (CL) was the interval between consecutive menstrual start dates. The mean and standard deviation (SD) of CL were calculated for each participant. Abnormal cycle length was defined as an individual mean CL outside the range of 24 to 38 days. 18 Symptoms experienced during menstruation were assessed using 31 items with six domains (pain, concentration, behavioral, autonomic, water retention, and negative affect). Detailed definitions of each domain are provided in Table  S2 . Among 23 850 users who completed Wave 1, those with fewer than three logged cycles were excluded (Figure  S1 ). Cycles ±4 SD from the log‐transformed mean were excluded as outliers. 19 Of 9325 participants completing Wave 2, 9081 (97.4%) completed the PSST. Participants were further excluded if they were aged over 45 years, had missing age data, were using hormonal contraception, were undergoing infertility treatment, were currently pregnant, or had a history of psychiatric disorders. The final analytic sample comprised 6966 participants and 151 833 cycles. Because variables such as health literacy, sleep, and Brief COPE were collected in different waves, analytic subsamples varied according to data availability for each measure. Descriptive statistics are summarized by PMS/PMDD status. Separate multinomial logistic regression models were used to examine associations between each background, psychosocial, and menstrual variable with PMS/PMDD status, using the control group as the reference. Results are presented as relative risk ratios (RRR) with 95% confidence intervals (CI). All models were adjusted for age, except for the model where age was the exposure. Menstrual outcomes were further adjusted for BMI, co‐residence with children, pregnancy intention, exercise, smoking, alcohol use, and work schedule. Statistical significance was defined as a P value less than 0.05. All analyses were performed with STATA/SE 18.0.

Figure S1. Table S1. Table S2.