Premenarche risk factors for future dysmenorrhoea: a prospective cohort study.

In this observational cohort study, we used data from the population-based Adolescent Brain Cognitive Development (ABCD) study, the longitudinal design of which allows us to examine the temporal relationship between demographic and symptom domains in premenarchal females at age 9–10 years and the presence and severity of dysmenorrhoea at age 12–13 years. 16 , 17 11 868 children aged 9–10 years were enrolled in the ABCD study from 21 US study sites in 2016–18. Written informed consent from parents or guardians and informed assent from participants aged 9–10 years were obtained at the baseline visit. The ABCD study conforms to the rules and procedures of each of the 21 sites’ Institutional Review Boards, and this investigation was approved by the Institutional Review Board at the University of Michigan Medical School (HUM00211215). Protocols for recruitment and assessment have been described previously. 16 , 17 Data used in this study were from the ABCD data release 6.1. To identify factors associated with new-onset dysmenorrhoea, female adolescents who had not yet reached menarche at baseline were included in this analysis. Included participants were required to have subsequently reached menarche and completed questions on dysmenorrhoea severity and interference at the 3-year follow-up visit (age 12–13 years). Participants completed the Menstrual Cycle Survey at the 3-year follow-up (ages 12–13 years). 16 , 18 Dysmenorrhoea was measured by a self-reported menstrual pain intensity item on a scale of 0–10, with 0 indicating “no pain at all” and 10 indicating “the worst pain I have ever had”. The primary outcome of dysmenorrhoea status was derived from this variable as follows: participants were first categorised as having either no dysmenorrhoea (scores ≤3) or dysmenorrhoea (scores ≥4), with scores of 4 or more considered clinically relevant. 19 The primary outcome variable dysmenorrhoea severity was then categorised as mild (scores 4–7) or severe (scores 8–10). Additionally, participants reported premenstrual symptoms, menstrual flow, bleeding duration, and menstrual pain interference (“How much does your menstrual pain stop you from doing your usual activities?”) on a scale of 0–10, with 0 indicating “not at all” and 10 indicating “stops me from doing anything”. Menstrual pain interference was the secondary outcome variable used in this study. Parents or guardians completed the Pubertal Development Scale at baseline, which includes items assessing body hair, skin changes, growth spurt, breast development, and menarche. 18 Menarche was reported as either absent (1) or completed (4), while other items were rated on a full 4-point scale from 1 (no development) to 4 (completed). Pubertal status was calculated as an average of the five items, resulting in scores ranging from 1 to 4. Participants with missing data for two or more of the five items were excluded from analysis. Parents or guardians completed the Child Behavior Checklist at baseline. The current study focused on three of the eight empirically derived behavioural symptom subscales: attention problems (internal consistency Cronbach α=0·85), anxious or depressed (α=0·67), and somatic symptoms (α=0·22). 20 These symptoms have been tied to chronic pain in both adults 21 , 22 and children. 12 , 23 The somatic symptoms subscale was split into two portions for analyses: non-painful somatic symptoms and pain items, as described previously. 12 The non-painful somatic symptoms included constipation, dizzy or lightheaded, nightmares, overtired without good reason, nausea or feels sick, problems with eyes, rashes or other skin problems, and vomiting. To assess baseline pain status, parents reported if their child currently or within the past 6 months reported the following physical problems without known medical cause: headaches, stomach aches, or other aches or pains. Responses were either often true, somewhat true, or not true. If a parent endorsed two or more items as either often true or somewhat true, the participant was coded as having multisite pain. One such item was coded as regional pain, and no such items was coded as no pain. 12 The Sleep Disturbance Scale for Children was completed by parents or guardians at baseline. It contains 26 items to evaluate sleep disorders, which are summed to create a total score of sleep disturbances. 24 , 25 We included the total sleep problems score (α=0·51) as a factor preceding dysmenorrhoea. Parents or guardians provided information on the child’s race, ethnicity, sex at birth, household income, and family size at baseline. A measure of socioeconomic status was derived using income-to-needs ratios calculated by dividing the household income by the corresponding federal poverty line for their family size. 26 A ratio value of 1 is indicative of a household at the federal poverty line, >1 indicates living above the federal poverty line, and <1 represents living below the poverty line. Asian, American Indian or Alaska Native, and Native Hawaiian or Other Pacific Islander participants were combined with Other Race for the analysis because of small numbers in these groups. We first characterised the data using descriptive statistics. The primary outcome variables of interest were the presence of dysmenorrhoea (menstrual pain severity of ≥4) and dysmenorrhoea severity (none, mild, severe). 19 We assessed differences in race using Fisher exact tests. Differences in ethnicity and baseline pain status were tested with χ 2 tests. Wilcoxon-Rank Sum tests were used for sleep problems, attention problems, and somatic symptoms, and ANOVA was used for pubertal status and income-to-needs ratio. Differences in continuous variables were assessed using ANOVA. Differences in menstrual flow were assessed using χ 2 tests. Our primary aim was to understand associations between dysmenorrhoea (status and severity) and select symptom domains (sleep problems, attention problems, somatic symptoms, anxious or depressive symptoms, and baseline pain status), whereas our secondary aim was to understand associations between dysmenorrhoea and participant characteristics (pubertal status, race or ethnicity, and income-to-needs ratio) that underlie social determinants of health. We used multinomial logistic regression to test associations with dysmenorrhoea severity, logistic regression to test associations with dysmenorrhoea status, and linear regression to test associations with menstrual pain interference. We used R (version 4.1.1) for all statistical analyses. Additional information regarding the participant sample derivation, measures used, and statistical testing are included in the appendix (pp 1–2). The ABCD consortium designed and implemented the study and provided data but did not participate in data analysis or writing of this report. The funder had no role in data collection, data analysis, data interpretation, writing of the report, or decision to submit for publication.

The full sample derivation and details of exclusions are shown in the figure . Clinical symptoms and demographics at baseline by menarche status are described in the appendix (p 3). Of the 2254 female adolescents included in the final analyses, clinical symptoms and demographics at baseline are described in table 1 . 1299 (57·6%) participants had dysmenorrhoea at age 12–13 years, 247 (19·0%) of whom reported severe dysmenorrhoea. Menstrual characteristics at age 12–13 years, including premenstrual symptoms occurring outside of menses, premenstrual symptoms interference with life, menstrual pain interference, menstrual bleeding duration, and menstrual flow, were significantly different across groups ( table 2 ). Associations of demographic and symptom domains with dysmenorrhoea status are shown in table 3 . Somatic symptoms were significantly associated with the presence of dysmenorrhoea (OR 1·17 [95% CI 1·04–1·32]; p=0·0070), but no significant association was found for attention problems, sleep problems, anxiety or depression, or baseline pain status. Advanced pubertal status at age 9–10 years was prospectively associated with dysmenorrhoea status (OR 1·79 [95% CI 1·47–2·17]; p<0·0001), as was lower income-to-needs ratio (0·96 [0·93–1·00]; p=0·031). Black (OR 1·38 [95% CI 1·05–1·83]; p=0·024) and Hispanic (OR 1·34 [1·03–1·68]; p=0·010) young females were more likely to report dysmenorrhoea than were White and non-Hispanic young females, respectively. Associations of demographic and symptom domains with dysmenorrhoea severity are shown in table 4 . Somatic symptoms were significantly associated with mild dysmenorrhoea (OR 1·17 [95% CI 1·04–1·31]; p=0·012). No significant association was found for sleep problems, attention problems, anxiety or depression, or baseline pain status. Advanced pubertal status at ages 9–10 years was significantly associated with developing both mild (OR 1·67 [95% CI 1·36–2·04]; p<0·0001) and severe (OR 2·43 [1·77–3·35]; p<0·0001) dysmenorrhoea 3 years later. Although both Black and Hispanic young people had significantly increased risk for developing severe dysmenorrhoea compared with White and non-Hispanic young people, respectively, only Hispanic youth were also at increased risk for developing mild dysmenorrhoea. Associations of demographic and symptom domains with menstrual pain interference are shown in table 5 . There was a significant association between sleep problems at baseline and menstrual pain interference 3 years later (β 0·16 [95% CI 0·01–0·31]; p=0·033) and a significant association between the presence of regional, but not multisite, pain at baseline and menstrual pain interference 3 years later (β 0·26 [0·01–0·52)]; p=0·042). Advanced pubertal status at baseline (β 0·51 [95% CI 0·28–0·73]; p<0·0001), Black race (0·64 [0·32–0·96]; p<0·0001), and Hispanic ethnicity (0·51 [0·25–0·78]; p=0·0001) were significantly associated with increased menstrual pain interference at age 12–13 years.

To our knowledge, this is the first study to examine risk factors present before the onset of menses and their association with the development of dysmenorrhoea. Using data from the ABCD Study, we examined whether symptom domains previously associated with the risk of developing pain and demographic characteristics associated with social determinants of health in premenarchal female adolescents were associated with dysmenorrhoea presence and severity and with menstrual pain interference 3 years later. The presence of non-painful somatic symptoms at age 9–10 years was strongly prospectively associated with future dysmenorrhoea, and sleep problems were strongly associated with the future burden of menstrual pain. Demographic characteristics and advanced pubertal status were strongly prospectively associated with both dysmenorrhoea severity and interference. These findings raise the possibility of identifying at-risk individuals before pain becomes a lifelong issue. We found that non-painful somatic symptoms were positively associated with the development of dysmenorrhoea 3 years later. Complementary to external sensory sensitivity (sensitivity to external or environmental stimuli such as bright lights), somatic symptoms reflect a form of internal sensory sensitivity (increased sensitivity to internal sensations and symptoms). Heightened internal and external sensory sensitivity are noted in patients with nociplastic pain. 22 In both adults and adolescents, increased somatic symptoms often precede new-onset pain (chronic widespread pain, facial pain, and head pain). 27 – 29 Although women with dysmenorrhoea often present with augmented sensory and somatic sensitivity, these findings suggest that increased somatic sensitivity precedes and is associated with the development of dysmenorrhoea. Our findings complement these studies showcasing an underappreciated relationship between internal sensory sensitivity and chronic pain in both children and adults. Similarly, previous work has shown that non-painful somatic symptoms are predictive of future development of multisite pain in children. 12 Although psychological interventions to address interoceptive awareness have been used to treat chronic pain in adolescents, 30 this approach might exacerbate over-awareness and sensitivity to interoceptive processes in young women with existing sensitivity to somatic symptoms. Internal and external sensory sensitivity might be an early sign of the future development of nociplastic pain. Screening for existing sensitivity to interoceptive processes could be used to help identify individuals who would most likely benefit from interoceptive interventions focused on attenuating overawareness or sensitivity to somatic symptoms. Sleep disturbances and pain have a well-documented relationship due to a high rate of co-occurrence and bidirectional reinforcement. Self-reported disturbances in sleep quality exacerbate existing pain and can predict new-onset pain in adults 21 and adolescents. 31 , 32 Sleep disturbances often co-occur with menstruation and dysmenorrhoea, 33 probably due to changes in the hormonal and inflammatory environment during menses. Our findings indicate that sleep disturbances present before menarche are a risk factor for worsened menstrual pain interference—the extent to which menstrual pain interferes with performing usual activities. A recent cross-sectional study of the 3-year follow-up (age 12–13 years) in the ABCD study found a robust relationship between sleep disturbances and menstrual pain intensity and effect on daily life. 34 Existing problems with sleep quality might be exacerbated during menstruation and increase the burden of menstrual pain. Although predictive of pain interference, we did not find sleep disturbances to be associated with dysmenorrhoea severity. The lack of association between sleep and pain severity contrasts with previous findings that sleep disturbances predate new-onset pain in adolescents. 31 , 35 For example, sleep disturbances have previously been shown to predict new-onset multisite pain in children 1 year later. This discrepancy might be due to the time difference between assessment of sleep quality and pain, as the current study spans a much longer timeframe (3 years) than previous work (1 year). 12 Addressing sleep disturbances present before menarche might be an important target for reducing the burden of dysmenorrhoea on young females. For example, cognitive behavioural therapy targeting sleep disturbances in adolescents with chronic pain is a promising intervention for addressing both sleep disturbances and pain interference in young people. 36 Advanced pubertal status at age 9–10 years (premenarche) was a significant risk factor for the development of dysmenorrhoea and increased menstrual pain interference 3 years later. By this time, 1310 female youth in the ABCD study population had not yet reached menarche, and the sample for the present analyses was enriched for female adolescents who experienced early or on-time menarche. Further analyses of the relationship between pubertal status and dysmenorrhoea in female adolescents who experienced menarche after age 12 years will deepen our understanding of how age at menarche and other markers of pubertal development contribute to the onset of dysmenorrhoea. Cultural, environmental, and social factors that contribute to differences in health outcomes (known as social determinants of health) 37 are associated with health disparities in both chronic pain and gynaecological disorders. 38 One potential reason is that female adults and adolescents in lower-income groups are less likely to have their dysmenorrhoea treated and thus might not report dysmenorrhoea until the pain becomes intolerable. 3 , 14 This might in part be due to normalisation of dysmenorrhoea, cultural biases related to menstruation (eg, stigma), and limited health-care access. 2 Previous studies have identified low socioeconomic status as a risk factor for pain in children. 39 Furthermore, Black and Hispanic female adolescents were significantly more likely to develop dysmenorrhoea than White and non-Hispanic female adolescents, respectively. There are several possible explanations for these findings, including earlier pubertal development and sociocultural stressors. 40 Further examination of the interplay between race, ethnicity, and other social determinants of health utilising upcoming timepoints from the ABCD study will enable a more nuanced examination of these relationships as more adolescents progress through puberty and reach menarche. We assessed risk factors for dysmenorrhoea present in girls before menarche. We excluded female adolescents who already experienced menarche at baseline, had not yet reached menarche at follow-up, or did not provide information on their menarche status. A limitation of this approach is that our sample excludes those who develop early, with menarche before age 9–10 years, and those who develop late, with menarche after ages 12–13 years. The positive relationship between baseline pubertal status and dysmenorrhoea severity indicates that advanced pubertal status, assessed before menarche, is an important risk factor for later development of dysmenorrhoea. An important caveat is that the included sample has more advanced pubertal status than the expected average for this age group and might be missing some participants who were premenarchal at baseline but whose parents did not complete questions relating to menarche. 40 Future data releases in the ABCD study and analyses that include pubertal data beyond the baseline assessment will help to contextualise the risk of early maturation. Contrary to our hypothesis, we did not find that baseline multisite pain was associated with the onset of dysmenorrhoea. This might reflect a key distinction between the developmental pathways of dysmenorrhoea and multisite pain or might be due to limitations in methodology. Surveys regarding pain severity and location (ie, body map, pain intensity, and interference) will enable more detailed, nuanced assessment of the relationship between adolescent pain and dysmenorrhoea in future studies. We did find that regional pain at baseline—endorsement of pain at one of the three sites—was associated with increased menstrual pain interference. Although we expected to see this association with both multisite and regional pain, the presence of regional pain at baseline is likely to augment the burden of existing menstrual pain. The temporal relationship between dysmenorrhoea and other pain conditions remains an avenue for identifying those most at risk for developing nociplastic pain. Depressive and anxious symptoms at baseline were not associated with future dysmenorrhoea in this sample. Although highly related in cross-sectional research on dysmenorrhoea and more broadly with paediatric chronic or recurrent pain, depressive or anxious symptoms might not predate pain symptoms. These findings, along with data from other longitudinal studies, 5 , 12 , 23 suggest that depressive or anxious symptoms occur concurrently or follow recurrent and chronic pain. Despite not being predictive of future dysmenorrhoea, screening and treatment for mental health symptoms in adolescents with dysmenorrhoea are an important aspect of reducing the burden of pain in female adolescents. A limitation of the study is the use of parent report for baseline measures and adolescent self-report for outcome measures. Because parents often underreport symptoms such as sleep disturbances and pain, findings might underestimate youth experiences. Additionally, pain assessment at baseline included only three body sites and a binary response for pain without known medical cause. Although we excluded individuals who reported a condition associated with pain at baseline, there might have been young people who did not report pain on this measure due to a known cause or because their pain was located outside of the predefined regions (eg, musculoskeletal pain). Finally, this study population excluded participants who had not reached menarche at the 3-year follow-up, which might limit the generalisability of the findings for late-maturing female adolescents. As a period of intense biological and social change, adolescence represents a window of vulnerability for the development of chronic pain and marks the emergence of sex differences in pain conditions. As such, it also represents an under-explored crucial period for the prevention of menstrual pain, owing to challenges in the early identification of those most at risk. This study helps close that gap by identifying early risk factors. Furthermore, the variables identified in this study (somatic symptoms, sleep disturbances, and advanced puberty) have been noted as risk factors for paediatric pain more broadly. When left untreated, dysmenorrhoea might increase the likelihood of chronic pain later in life. Identifying risk factors for dysmenorrhoea, especially severe dysmenorrhoea, has the potential to improve outcomes for female adolescents through early and more targeted treatment. Future studies should examine whether early interventions—especially those focused on sleep and somatic symptoms—can reduce the effect of dysmenorrhoea.

Dysmenorrhoea, also known as menstrual pain, is prevalent among female adolescents (approximately 70%, with up to 20% of cases being severe) and can have a significant impact on daily activities. 1 Because dysmenorrhoea is so common among young women, menstrual pain is often dismissed as an inevitable, if unfortunate, feature of life for female adolescents. 2 , 3 Coupled with a general bias to view paediatric pain as benign, most female adolescents do not receive a thorough assessment of their pain until it is severe and disabling. 3 , 4 These cultural and medical biases have negative consequences for young women and girls: according to a large 2019 systematic review and meta-analysis, dysmenorrhoea contributes to about 20% of school absences and negatively affects academic performance and participation in extracurricular activities, 1 and survey data from Australia highlight that dysmenorrhoea in adolescence is associated with poorer mental health. 5 Moreover, dysmenorrhoea often persists into adulthood with debilitating effects such as chronic pelvic pain. 6 Improved understanding of risk factors for dysmenorrhoea (especially severe dysmenorrhoea) in adolescence would support early diagnosis and intervention to reduce the disease burden and potentially prevent the lifelong course of chronic pain that occurs in some individuals with dysmenorrhoea. Dysmenorrhoea is categorised as primary (pain with menses in the absence of peripheral pathology) or secondary (pain with underlying pathology such as endometriosis). 7 In both cases, nociplastic pain mechanisms might contribute to altered pain processing in some individuals. 8 , 9 Nociplastic pain has been defined as “pain that arises from altered nociception despite no clear evidence of actual or threatened tissue damage causing the activation of peripheral nociceptors or evidence for disease or lesion of the somatosensory system causing the pain”. 10 Dysmenorrhoea in young females is an important risk factor for the development of other chronic pain disorders, including pain in multiple body regions (multisite pain), 8 and the presence of chronic pain conditions is associated with worse disease burden of dysmenorrhoea. 9 , 11 Hallmark features of nociplastic pain are the presence of multisite pain and symptoms such as sleep disturbances, heightened somatic awareness, and cognitive problems. 10 We previously identified risk factors in pain-free male and female youth for the subsequent development of multisite pain; notably, female sex was a significant predictor for the development of new multisite pain. 12 The symptom domains that predicted new multisite pain were non-painful somatic symptoms, attentional issues, and sleep problems, suggesting a CNS vulnerability to pain development. Risk factors for adolescent dysmenorrhoea include social determinants of health, lifestyle factors, maternal dysmenorrhoea, age at menarche, and menstrual cycle patterns. 1 , 13 However, there is little consensus across studies due to differences in study methodology, inconsistent use of validated pain measures, and sociodemographic cohort differences. The few longitudinal studies on adolescent dysmenorrhoea have focused on late adolescence after menarche onset, 14 , 15 which is problematic because this limits opportunities for early diagnosis and intervention. A fundamental impediment to identifying risk for dysmenorrhoea is a lack of research into risk factors that might occur before menarche. In this study, we investigated whether non-painful somatic symptoms, attentional issues, sleep disturbances, anxious or depressive symptoms, and multisite pain (indicative of nociplastic pain vulnerabilities 12 ) are prospectively associated with the incidence and severity of dysmenorrhoea within 3 years. We also examined the relationship between social determinants of health and dysmenorrhoea outcomes in adolescence.