Premenstrual Disorders, Timing of Menopause, and Severity of Vasomotor Symptoms

Key Points Question Is presence of a premenstrual disorder (PMD) associated with higher risk of early menopause and menopause-related vasomotor symptoms (VMS)? Findings In this cohort study of 3635 female participants in the US, presence of a PMD was significantly associated with higher risks of early menopause and moderate or severe VMS. Meaning These results suggest a phenotype observable during the reproductive years that may allow clinicians to target women at risk of adverse experiences during menopause transition.

In addition, we collected information on factors only related to PMDs or menopause timing/symptoms.Total breastfeeding months was surveyed from 1993.Intake of vitamin B1, vitamin B2, iron, zinc and potassium were surveyed in the SFFQ every 4 years since 1991.
Multiple imputation with chained equations was used to impute missing values in covariates 4 .Missing values were generated using predictive mean matching (for age at menarche and physical activity), logistic (for childhood abuse, maternal education level, marital status, breast feeding), and multinomial logistic regression (for smoking, BMI category, and intake of alcohol, calcium, vitamins, and minerals) using the variables menopause status, PMDs, age at diagnosis/reference year, birth year, race, and parity.
Use of oral contraceptive and hormone therapy may delay or mask onset of menopause 5 and alleviate VMS 6 .They were surveyed biennially from 1991.
Anxiety and depression are common comorbidities of PMDs 7 .We defined women with anxiety as having a Crown-Crisp Anxiety Scale items (CCI) > 6 in 1993 and 2005 8 , or ever use of minor tranquilizers surveyed in 1993 and every two years since 1997.Depression was identified by: 1) clinician-diagnosed depression, reported in NHSII questionnaire in 2003 and 2005, and PMDs assessment questionnaire; 2) antidepressant use, collected in 1993 and biennially from 1997 in NHSII, and PMDs assessment questionnaire; 3) the 5-itm Mental Health Index (MHI) scored < 60 9 , assessed in 1993, 1997 and 2001 in NHS II.

Statistical analyses
We also conducted several additional analyses.First, to test potential risk modifications, we performed stratified analyses by age at menarche, use of OC, BMI category, and smoking status at matching.Second, to address residual confounding, we further adjusted for factors related with PMDs/menopause only, including breastfeeding, and intake of vitamin B1, vitamin B2, iron, zinc, and potassium at matching.Third, to evaluate the independent role of early menopause and moderate/severe VMS, we mutually adjusted for these factors.Fourth, to evaluate the impact of imputation, we conducted a complete case analysis.Fifth, to evaluate the impact of HT use, we censored at HT use in the analysis for early menopause and excluded women who had ever used HT in the analysis for VMS.
In the analysis of early menopause, presuming that PMD diagnosis may lead to changes in lifestyle (e.g.smoking, alcohol drinking and physical activity), BMI, and marital status, we adjusted for these factors in a time-varying manner.In the analysis of moderate/severe VMS, first we only used data from current VMS and performed mixed effects logistic regression to account for correlations between repeated measurements 10 .Moreover, we excluded women who had cancer, hysterectomy, oophorectomy before menopause to rule out potential mediation to VMS.We also evaluated the association between VMS and specific premenstrual symptoms.Individuals who lacked information on any covariates (n=548 (15.1%)) were excluded.a The follow-up started from age 42, due to no events before age 42 among women with PMDs, or age at matching, whichever came later, except for analysis censored at HT use, in which the follow-up started from age 43, due to no events before age 43 among women with PMDs, or age at matching, whichever came later.b Mild VMS was not considered as an outcome event.c The estimates were adjusted for or birth year (for early menopause)/age at diagnosis or reference year (for VMS), race, maternal education level, marital status, category of body mass index, age at menarche, parity, smoking, alcohol drinking, physical activity, childhood abuse, vitamin D intake and calcium intake at matching.d The estimates were additionally adjusted for breastfeeding (ie, 0-24 months, 24 months+), intake of vitamin B1, B2, zinc, iron and potassium (in quintiles).

eTable 1. Classification
Criteria of Premenstrual Disorders and PremenstrualDysphoric Disorder Associations of Premenstrual Disorders With Risks of Early Natural Menopause and Moderate/Severe Vasomotor Symptoms in Stratified Analyses CI, confidence interval; IR, incidence rate per 1,000 person-years; N, number of events; HR, hazard ratio; OR, odds ratio; PMD, premenstrual disorder; OC, oral contraceptives; PYs, person years; Ref., reference; VMS, vasomotor symptoms.OC use, BMI category and smoking were collected at diagnosis/reference year.aThefollow-up started from age 42, due to no events before age 42 among women with PMDs, or age at matching, whichever came later.bMildVMS was not considered as an outcome event.cTheestimates were adjusted for or birth year (for early menopause)/age at diagnosis or reference year (for VMS), race, maternal education level, marital status, category of body mass index, age at menarche, parity, smoking, alcohol drinking, physical activity, childhood abuse, vitamin D intake and calcium intake at baseline.d There was no case of early menopause in the category of underweight.Associations of Premenstrual Disorders With Risks of Early Natural Menopause and Moderate/Severe Vasomotor Symptoms in Additional Analyses eTable 2. eTable 3.

eTable 4 .
Associations of Premenstrual Disorders With Risks of Early Natural Menopause in Additional Analyses PYs, person years; Ref., reference.The follow-up started from age 42, due to no events before age 42 among women with PMDs, or age at matching, whichever came later.a The estimates were adjusted for year of birth, race, maternal education level, age at menarche, parity, childhood abuse, vitamin D intake and calcium intake at matching, and time-varying marital status, body mass index category, smoking, alcohol drinking, and physical activity.eTable 5. Associations of Premenstrual Disorders With Risks of Moderate/Severe Vasomotor Symptoms in Additional Analyses CI, confidence interval; N, number; OR, odds ratio; PMD, premenstrual disorder; Ref., reference; VMS, vasomotor symptoms.Mild VMS was not considered as an outcome event.a Mixed effects logistic regression was used.b The estimates were adjusted for or age, race, maternal education level, marital status, category of body mass index, age at menarche, parity, smoking, alcohol drinking, physical activity, childhood abuse, vitamin D intake and calcium intake at diagnosis/reference year.Associations of Specific Premenstrual Symptoms With Risks of Moderate/Severe Vasomotor Symptoms CI, confidence interval; N, number; OR, odds ratio; PMD, premenstrual disorder; Ref., reference; VMS, vasomotor symptoms.Mild VMS was not considered as an outcome event.a The estimates were adjusted for or age, race, maternal education level, marital status, category of body mass index, age at menarche, parity, smoking, alcohol drinking, physical activity, childhood abuse, vitamin D intake and calcium intake at diagnosis/reference year.