Common diseases as determinants of menopausal age

Introduction

Menopause marks the end of the female reproductive span, and it is defined as the permanent cessation of menstruation due to loss of ovarian activity (Greendale et al., 1999). The timing of menopause onset is determined by the critical depletion of a finite and non-renewing follicle pool, which continuously decreases since birth. In the average woman, this programmed depletion of follicles results in a fertility decline around 37–38 years of age and onset of menopause ~13 years later (McKinlay et al., 1992; Morabia and Costanza, 1998; Gold et al., 2001; Sapre and Thakur, 2014). The age of becoming menopausal is of clinical significance, as it has been shown to be an important marker for subsequent morbidity and mortality (Snowdon et al., 1989; Jacobsen et al., 2003; Mondul et al., 2005). Early age at natural menopause (ANM) has been associated with increased risks of cardiovascular disease (Mondul et al., 2005; Rahman et al., 2015), stroke (Lisabeth et al., 2009), osteoporosis (Kritz-Silverstein and Barrett-Connor, 1993), depression (Georgakis et al., 2016) and long-term negative effects on cognitive function (Ryan et al., 2014) and decreased risk of breast (Collaborative Group on Hormonal Factors in Breast, 2012) and endometrial cancers (McPherson et al., 1996).

An impact of smoking on early onset of menopause has been consistently and clearly demonstrated in many studies, with tobacco consumption accelerating ANM by 1–2 years (Gold et al., 2001; Parente et al., 2008; Sapre and Thakur, 2014). Factors such as socioeconomic status, higher education, irregular menstrual cycles, parity, moderate alcohol consumption and using contraceptives have all been suggested to delay ANM (Sapre and Thakur, 2014). Other factors such as early age at menarche, nulliparity, short-length menstrual cycles and some nutritional factors such as a high intake of fat, cholesterol and coffee accelerates ANM (Dorman et al., 2001; Ozdemir and Col, 2004; Sapre and Thakur, 2014).

Whether early menopause can be a complication of common adult diseases, rather than the reverse, has not been studied extensively. Gold et al. (2001) reported a significant association between a history of heart disease and younger ANM. However, due to the cross-sectional design of the study, it could not be shown that natural menopause preceded a diagnosis of heart disease. While most studies show a link between early menopause and increased risks of developing cardiovascular diseases following menopause, a study using the Framingham Heart Study cohort reported that an elevated cardiovascular risk profile based on cholesterol, body weight and blood pressure determines menopausal age instead (Kok et al., 2006). In one study, Type 1 diabetes has been hypothesized to accelerate ovarian aging through prolonged poor glycemic control and subsequent effects on vascular health (Dorman et al., 2001). However, in three other studies, ANM in women with diabetes were not found to experience menopause earlier than women without diabetes (Brand et al., 2013; Yarde et al., 2015) or the general population (Sjoberg et al., 2011). In addition, hypertension has also been postulated to lower ANM in a Korean population (Lim et al., 2016).

An understanding of why certain factors determine the timing of menopause may, for some women, help prevent premature loss of fecundity or delay health problems previously found to affect postmenopausal women. In the large Karolinska Mammography Project for Risk Prediction of Breast Cancer (KARMA) study, we expanded the scope of previous studies on determinants of menopausal age to include 13 common diseases: depression, anorexia, bulimia, PCOS, ovarian cyst, heart failure, myocardial infarction, angina pectoris, stroke, preeclampsia, diabetes, hypertension and hyperlipidemia.

Materials and Methods

Study participants

The KARMA (www.karmastudy.org) study is a prospective population-based cohort study of women aged 40–74 years who are offered mammography screening within the Swedish national mammography screening program. Women participating in the screening program or attending clinical mammography at one of the four mammography units in Sweden (Södersjukhuset, Helsingsborg, Landskrona and Lund hospitals) were invited to participate in KARMA. The study was initiated in January 2011 and had 70 877 participants at the end of recruitment in March, 2013.

Ethical approval

The study was approved by the regional ethical review board in Stockholm (2010/958-31/1). All participants gave informed consent and answered a detailed web-based questionnaire upon enrollment.

Exclusions

Figure 1 shows the flow diagram of how the final analytic cohort of 61 936 women was derived. We excluded women who did not respond (n = 110) or return incomplete questionnaires (n = 2263), missing information to compute menopausal status and age at last menstruation (n = 3469) and missing information to compute time-dependent exposures (n = 1580) and covariates (n = 45). As some of the non-time-dependent covariates pertain to status in adult life (i.e. menstrual cycle regularity, education, physical activity, smoking and alcohol consumption), we also excluded women with a follow-up time of <35 years (n = 1474).

Figure 1

Flow diagram of analytic cohort.

Open in new tabDownload slide

Assessment of exposure to common diseases

Exposure to the 13 common diseases was based on self-report and obtained through the KARMA questionnaire. The participants were asked whether they have ever been diagnosed with any of the common diseases by a medical doctor, and if so, at what age. Only diagnoses made before ANM or censored age were considered in this study.

Assessment of outcome (menopausal status and age at menopause)

Natural menopause is officially recognized to have occurred after 12 consecutive months of cessation of menstruation, for which there is no other obvious pathological or physiological cause. Women were asked in the questionnaire whether they had a menstrual period during the last year, and if not, the reason for not having a menstrual period and at what age they had the last menstrual period. Women who had a menstrual period during the last year were considered premenopausal. Among those with 12 or more months of amenorrhea, women who chose ‘menopause’ as the only reason for not having menstrual periods were defined as naturally postmenopausal.

Assessment of other covariates

Information on age at menarche, menstrual cycle regularity (monthly period and start of cycle predictable within 5 days earlier in life as adult), education, physical activity at age 18, BMI, smoking (ever smoked regularly for >1 year or 100 cigarettes in total) and alcohol consumption was derived from the questionnaire. Information on BMI, smoking and alcohol consumption pertains to status at time of questionnaire. Number of children was modeled as a time-dependent variable to integrate information of age at birth of each child. Premenopausal oral contraceptive use was also modeled as a binary time-dependent variable, changing from unexposed to birth to expose at the age when the women first took oral contraceptives.

Statistical analysis

The principal analytical strategy was survival analyses, which allows the inclusion of time-dependent variables and censored observations in the estimation of ANM. As age is strongly correlated with menopause onset and disease, age was chosen as the underlying timescale to control for this possible confounder (Cologne et al., 2012). Follow-up was from birth until age at questionnaire for premenopausal women and age at last menstruation for naturally postmenopausal women. Women who reported 12 or more months of amenorrhea for reasons other than natural menopause were censored at age at last menstruation. Kaplan–Meier analysis was used to estimate the median ANM and corresponding interquartile range for the entire study population and by each strata of participant characteristic. Cox proportional hazard models were used to estimate the association (hazard ratios (HR) and 95% CI) between participant characteristic and ANM. The HR represents the risk of becoming naturally menopausal at a given age. HR < 1 indicate later menopause and HR > 1 indicate earlier menopause as compared with the reference category (HR = 1).

Diagnosis of each common disease was modeled as a binary time-dependent variable, changing from unexposed to birth to expose at the self-reported age at diagnosis. Correlated common diseases were examined as aggregated variables (eating disorder (anorexia and bulimia), cardiovascular diseases (heart failure, myocardial infarction, angina pectoris and stroke) and risk factors for cardiovascular diseases (diabetes, hypertension and hyperlipidemia), see Fig. 2). The association between the premenopausal diagnosis of each common disease and ANM was examined in three consecutive models: Model 1: no covariate (unadjusted), Model 2: adjusted for reproductive factors age at menarche (<14, ≥14 or missing), adult menstrual cycle regularity (regular, irregular or missing), number of children (time-dependent variable) and premenopausal oral contraceptive use (mini and combination pill use as time-dependent variables) and risk factors of common diseases (education (elementary, intermediate, college, other or missing), BMI (kg/m², <18.5, 18.5–24.9, 25–29.9, ≥30), physical activity at age 18 (h/week, 0, <1, 1–2, 3–5, ≥5 or missing), ever smoked for 1 year or 100 cigarettes (yes, no or missing), alcohol consumption (g/day, 0, <10, 10–24, ≥25 or missing)) and Model 3: Model 2 and all common diseases examined.

Figure 2

Correlation matrix of common diseases ordered according to correlation coefficients.

Open in new tabDownload slide

To reduce misclassification of menopausal status due to gynecological (cervical or ovarian) surgery (women might have indicated menopause as the reason for cessation of menstrual bleeding on the questionnaire instead of gynecological surgery), we conducted sensitivity analyses on a subset of 41 939 women who answered no to ever undergoing such surgery. In addition, as true menopause status could be masked by hormone treatment, we further examined the associations in a subset of 32 313 women who have neither received hormone treatment nor underwent any gynecological surgery.

Results

Table I describes the study population of 61 936 women. The mean (SD) ANM was 50.8 (4.2) years, and 40.4% of the women were still menstruating. The estimated median ANM in the entire study population was 52 years. The mean age of diagnosis (years) was the highest for myocardial infarction (45.7), followed by angina pectoris (43.6), hyperlipidemia (43.5), hypertension (41.3), stroke (41.1), heart failure (40.5), depression (36.8), diabetes (36.1), ovarian cyst (32), PCOS (29.1), preeclampsia (29.0), bulimia (24.6) and anorexia (19.3).

Figure 3 shows the crude ANM across strata of each participant characteristic and corresponding HR and 95% CI after multivariable adjustment. Ever smoking (for 1 year or 100 cigarettes) was associated with accelerated menopause (HR [95% CI] = 1.14 [1.12–1.17]). Associated with delayed menopause were older age at menarche (0.97 [0.95–1.00]) and irregular menstrual cycles in adulthood (0.75 [0.72–0.79]). More children, higher education, higher BMI, more hours spent per week on physical activity and higher alcohol consumption were also associated with delayed menopause.

Figure 3

Associations between reproductive and lifestyle factors and age at natural menopause (ANM), interquartile range (IQR).

Open in new tabDownload slide

Figure 4 illustrates the survival curves according to the diagnosis of common diseases before menopause. The crude ANM (in years) was the highest for women with a premenopausal diagnosis of PCOS (56), followed by depression (53), eating disorder (53), preeclampsia (53), risk factors of cardiovascular diseases (53), ovarian cyst (52), cardiovascular diseases (52) and none of the above (52). In the multivariable adjusted Cox regression model (Table II, Model 3), depression (0.95 [0.91–1.00]) and PCOS (0.44 [0.28–0.71]) were associated with delayed menopause, while ovarian cyst was significantly associated with early menopause (1.12 [1.06–1.18]). None of the other common diseases were associated with ANM. In the sensitivity analysis comprising a subset of women who have not undergone any gynecological surgery, the increased risk of being menopausal associated with ovarian cyst attenuated toward the null and was no longer significant (1.02 [0.92–1.12]), while the decreased risks associated with depression and PCOS became more pronounced (0.93 [0.89–0.98] and 0.27 [0.14–0.51], respectively). The association between preeclampsia and delayed menopause also reached nominal statistical significance in this subset (0.93 [0.86–1.00]). Further excluding women who have ever received hormone treatment yielded even lower risks of becoming postmenopausal in women diagnosed with depression (0.91 [0.85–0.98]), PCOS (0.21 [0.08–0.50]) and preeclampsia (0.89 [0.81–0.98]). The results did not materially change after further restricting the data set to women with no personal cancer history (n = 31 183, data not shown).

Figure 4

Survival curves and median ANMs (IQR) according to diagnosis of common diseases before menopause.

Open in new tabDownload slide

Table II

Open in new tab

HR and corresponding 95% CI for natural menopause according to premenopausal common disease diagnosis.

Common disease .	No (Reference) .	Yes .	Model 1 .	Model 2 .	Model 3 .
Overall (n = 61 936)
Depression	55 316	6620	0.94 (0.90–0.98)	0.95 (0.91–0.99)	0.95 (0.91–1.00)
Eating disorder	61 473	464	0.90 (0.77–1.04)	0.91 (0.78–1.05)	0.91 (0.78–1.06)
Anorexia	61 473	463	0.90 (0.77–1.04)	0.91 (0.78–1.05)	0.94 (0.80–1.09)
Bulimia	61 643	293	0.85 (0.66–1.10)	0.84 (0.65–1.09)	0.89 (0.68–1.16)
PCOS	1730	206	0.39 (0.25–0.60)	0.45 (0.28–0.72)	0.44 (0.28–0.71)
Ovarian cyst	58 637	3299	1.12 (1.06–1.18)	1.12 (1.06–1.18)	1.12 (1.06–1.18)
Cardiovascular disease	61 444	492	0.95 (0.81–1.11)	0.96 (0.82–1.13)	0.98 (0.83–1.15)
Heart failure	61 852	84	1.12 (0.80–1.57)	1.12 (0.80–1.59)	1.18 (0.85–1.65)
Myocardial infarction	61 830	106	0.86 (0.66–1.12)	0.87 (0.67–1.13)	0.90 (0.68–1.17)
Angina pectoris	61 781	155	0.89 (0.72–1.10)	0.91 (0.74–1.13)	0.94 (0.76–1.18)
Stroke	61 735	201	1.07 (0.87–1.31)	1.05 (0.86–1.30)	1.06 (0.86–1.30)
Preeclampsia	59 171	2765	0.92 (0.86–0.97)	0.95 (0.89–1.01)	0.96 (0.90.-1.02)
Cardiovascular disease risk factor	55 086	6850	0.95 (0.92–0.99)	0.96 (0.93–1.00)	0.96 (0.93–1.00)
Diabetes	61 192	744	0.88 (0.79–0.99)	0.90 (0.80–1.01)	0.91 (0.81–1.02)
Hypertension	56 738	5198	0.96 (0.92–1.00)	0.97 (0.93–1.01)	0.98 (0.94–1.02)
Hyperlipidemia	59 977	1959	0.99 (0.93–1.06)	1.00 (0.93–1.07)	1.01 (0.94–1.09)
Subset of women who answered No to Did you ever have cervical or ovarian surgery? (n=41 939)
Depression	37 559	4380	0.92 (0.87–0.97)	0.93 (0.88–0.98)	0.93 (0.89–0.99)
Eating disorder	41 624	316	0.82 (0.69–0.98)	0.84 (0.7–1.01)	0.85 (0.71–1.02)
PCOS	41 797	142	0.23 (0.12–0.42)	0.27 (0.14–0.51)	0.27 (0.14–0.51)
Ovarian cyst	40 877	1062	1.01 (0.92–1.11)	1.01 (0.91–1.11)	1.02 (0.92–1.12)
Cardiovascular disease	41 620	319	1.01 (0.82–1.23)	1.02 (0.83–1.24)	1.03 (0.84–1.26)
Preeclampsia	40 166	1773	0.90 (0.83–0.96)	0.92 (0.86–1.00)	0.93 (0.86–1.00)
Cardiovascular disease risk factor	37 374	4565	0.96 (0.92–1.00)	0.96 (0.92–1.01)	0.97 (0.93–1.01)
Subset of women who never received hormone treatment and answered No to Did you ever have cervical or ovarian surgery? (n=32 313)
Depression	29 004	3309	0.9 (0.84–0.97)	0.91 (0.85–0.97)	0.91 (0.85–0.98)
Eating disorder	32 074	239	0.83 (0.69–1.01)	0.86 (0.71–1.04)	0.87 (0.72–1.06)
PCOS	32 243	70	0.19 (0.08–0.47)	0.21 (0.08–0.50)	0.21 (0.08–0.50)
Ovarian cyst	31 565	748	0.97 (0.86–1.1)	0.97 (0.85–1.1)	0.97 (0.86–1.11)
Cardiovascular disease	32 081	232	1.02 (0.77–1.35)	1.01 (0.77–1.33)	1.02 (0.78–1.35)
Preeclampsia	30 951	1362	0.86 (0.78–0.94)	0.89 (0.81–0.98)	0.89 (0.81–0.98)
Cardiovascular disease risk factor	28 946	3367	0.97 (0.92–1.03)	0.97 (0.92–1.03)	0.98 (0.93–1.04)

Common disease .	No (Reference) .	Yes .	Model 1 .	Model 2 .	Model 3 .
Overall (n = 61 936)
Depression	55 316	6620	0.94 (0.90–0.98)	0.95 (0.91–0.99)	0.95 (0.91–1.00)
Eating disorder	61 473	464	0.90 (0.77–1.04)	0.91 (0.78–1.05)	0.91 (0.78–1.06)
Anorexia	61 473	463	0.90 (0.77–1.04)	0.91 (0.78–1.05)	0.94 (0.80–1.09)
Bulimia	61 643	293	0.85 (0.66–1.10)	0.84 (0.65–1.09)	0.89 (0.68–1.16)
PCOS	1730	206	0.39 (0.25–0.60)	0.45 (0.28–0.72)	0.44 (0.28–0.71)
Ovarian cyst	58 637	3299	1.12 (1.06–1.18)	1.12 (1.06–1.18)	1.12 (1.06–1.18)
Cardiovascular disease	61 444	492	0.95 (0.81–1.11)	0.96 (0.82–1.13)	0.98 (0.83–1.15)
Heart failure	61 852	84	1.12 (0.80–1.57)	1.12 (0.80–1.59)	1.18 (0.85–1.65)
Myocardial infarction	61 830	106	0.86 (0.66–1.12)	0.87 (0.67–1.13)	0.90 (0.68–1.17)
Angina pectoris	61 781	155	0.89 (0.72–1.10)	0.91 (0.74–1.13)	0.94 (0.76–1.18)
Stroke	61 735	201	1.07 (0.87–1.31)	1.05 (0.86–1.30)	1.06 (0.86–1.30)
Preeclampsia	59 171	2765	0.92 (0.86–0.97)	0.95 (0.89–1.01)	0.96 (0.90.-1.02)
Cardiovascular disease risk factor	55 086	6850	0.95 (0.92–0.99)	0.96 (0.93–1.00)	0.96 (0.93–1.00)
Diabetes	61 192	744	0.88 (0.79–0.99)	0.90 (0.80–1.01)	0.91 (0.81–1.02)
Hypertension	56 738	5198	0.96 (0.92–1.00)	0.97 (0.93–1.01)	0.98 (0.94–1.02)
Hyperlipidemia	59 977	1959	0.99 (0.93–1.06)	1.00 (0.93–1.07)	1.01 (0.94–1.09)
Subset of women who answered No to Did you ever have cervical or ovarian surgery? (n=41 939)
Depression	37 559	4380	0.92 (0.87–0.97)	0.93 (0.88–0.98)	0.93 (0.89–0.99)
Eating disorder	41 624	316	0.82 (0.69–0.98)	0.84 (0.7–1.01)	0.85 (0.71–1.02)
PCOS	41 797	142	0.23 (0.12–0.42)	0.27 (0.14–0.51)	0.27 (0.14–0.51)
Ovarian cyst	40 877	1062	1.01 (0.92–1.11)	1.01 (0.91–1.11)	1.02 (0.92–1.12)
Cardiovascular disease	41 620	319	1.01 (0.82–1.23)	1.02 (0.83–1.24)	1.03 (0.84–1.26)
Preeclampsia	40 166	1773	0.90 (0.83–0.96)	0.92 (0.86–1.00)	0.93 (0.86–1.00)
Cardiovascular disease risk factor	37 374	4565	0.96 (0.92–1.00)	0.96 (0.92–1.01)	0.97 (0.93–1.01)
Subset of women who never received hormone treatment and answered No to Did you ever have cervical or ovarian surgery? (n=32 313)
Depression	29 004	3309	0.9 (0.84–0.97)	0.91 (0.85–0.97)	0.91 (0.85–0.98)
Eating disorder	32 074	239	0.83 (0.69–1.01)	0.86 (0.71–1.04)	0.87 (0.72–1.06)
PCOS	32 243	70	0.19 (0.08–0.47)	0.21 (0.08–0.50)	0.21 (0.08–0.50)
Ovarian cyst	31 565	748	0.97 (0.86–1.1)	0.97 (0.85–1.1)	0.97 (0.86–1.11)
Cardiovascular disease	32 081	232	1.02 (0.77–1.35)	1.01 (0.77–1.33)	1.02 (0.78–1.35)
Preeclampsia	30 951	1362	0.86 (0.78–0.94)	0.89 (0.81–0.98)	0.89 (0.81–0.98)
Cardiovascular disease risk factor	28 946	3367	0.97 (0.92–1.03)	0.97 (0.92–1.03)	0.98 (0.93–1.04)

Model 1: no covariate (unadjusted), Model 2: adjusted for reproductive factors (age at menarche (<14, ≥14 or missing), adult menstrual cycle regularity (regular, irregular or missing), number of children (time-dependent variable) and premenopausal oral contraceptive use (mini and combination pill use as time-dependent variables) and risk factors of common diseases (education (elementary, intermediate, college, other or missing), BMI (kg/m², <18.5, 18.5–24.9, 25–29.9, ≥30), physical activity at age 18 (h/week, 0, <1, 1–2, 3–5, ≥5 or missing), ever smoked for 1 year or 100 cigarettes (yes, no or missing), alcohol consumption (g/day, 0, <10, 10–24, ≥25 or missing))Model 3: Model 2 and all common diseases examined. The hazard ratios (HR) represent the risk of becoming naturally menopausal at a given age. HR < 1 indicate later menopause and HR > 1 indicate earlier menopause as compared with the reference category. Significant associations are denoted in bold.

Table II

Open in new tab

HR and corresponding 95% CI for natural menopause according to premenopausal common disease diagnosis.

Common disease .	No (Reference) .	Yes .	Model 1 .	Model 2 .	Model 3 .
Overall (n = 61 936)
Depression	55 316	6620	0.94 (0.90–0.98)	0.95 (0.91–0.99)	0.95 (0.91–1.00)
Eating disorder	61 473	464	0.90 (0.77–1.04)	0.91 (0.78–1.05)	0.91 (0.78–1.06)
Anorexia	61 473	463	0.90 (0.77–1.04)	0.91 (0.78–1.05)	0.94 (0.80–1.09)
Bulimia	61 643	293	0.85 (0.66–1.10)	0.84 (0.65–1.09)	0.89 (0.68–1.16)
PCOS	1730	206	0.39 (0.25–0.60)	0.45 (0.28–0.72)	0.44 (0.28–0.71)
Ovarian cyst	58 637	3299	1.12 (1.06–1.18)	1.12 (1.06–1.18)	1.12 (1.06–1.18)
Cardiovascular disease	61 444	492	0.95 (0.81–1.11)	0.96 (0.82–1.13)	0.98 (0.83–1.15)
Heart failure	61 852	84	1.12 (0.80–1.57)	1.12 (0.80–1.59)	1.18 (0.85–1.65)
Myocardial infarction	61 830	106	0.86 (0.66–1.12)	0.87 (0.67–1.13)	0.90 (0.68–1.17)
Angina pectoris	61 781	155	0.89 (0.72–1.10)	0.91 (0.74–1.13)	0.94 (0.76–1.18)
Stroke	61 735	201	1.07 (0.87–1.31)	1.05 (0.86–1.30)	1.06 (0.86–1.30)
Preeclampsia	59 171	2765	0.92 (0.86–0.97)	0.95 (0.89–1.01)	0.96 (0.90.-1.02)
Cardiovascular disease risk factor	55 086	6850	0.95 (0.92–0.99)	0.96 (0.93–1.00)	0.96 (0.93–1.00)
Diabetes	61 192	744	0.88 (0.79–0.99)	0.90 (0.80–1.01)	0.91 (0.81–1.02)
Hypertension	56 738	5198	0.96 (0.92–1.00)	0.97 (0.93–1.01)	0.98 (0.94–1.02)
Hyperlipidemia	59 977	1959	0.99 (0.93–1.06)	1.00 (0.93–1.07)	1.01 (0.94–1.09)
Subset of women who answered No to Did you ever have cervical or ovarian surgery? (n=41 939)
Depression	37 559	4380	0.92 (0.87–0.97)	0.93 (0.88–0.98)	0.93 (0.89–0.99)
Eating disorder	41 624	316	0.82 (0.69–0.98)	0.84 (0.7–1.01)	0.85 (0.71–1.02)
PCOS	41 797	142	0.23 (0.12–0.42)	0.27 (0.14–0.51)	0.27 (0.14–0.51)
Ovarian cyst	40 877	1062	1.01 (0.92–1.11)	1.01 (0.91–1.11)	1.02 (0.92–1.12)
Cardiovascular disease	41 620	319	1.01 (0.82–1.23)	1.02 (0.83–1.24)	1.03 (0.84–1.26)
Preeclampsia	40 166	1773	0.90 (0.83–0.96)	0.92 (0.86–1.00)	0.93 (0.86–1.00)
Cardiovascular disease risk factor	37 374	4565	0.96 (0.92–1.00)	0.96 (0.92–1.01)	0.97 (0.93–1.01)
Subset of women who never received hormone treatment and answered No to Did you ever have cervical or ovarian surgery? (n=32 313)
Depression	29 004	3309	0.9 (0.84–0.97)	0.91 (0.85–0.97)	0.91 (0.85–0.98)
Eating disorder	32 074	239	0.83 (0.69–1.01)	0.86 (0.71–1.04)	0.87 (0.72–1.06)
PCOS	32 243	70	0.19 (0.08–0.47)	0.21 (0.08–0.50)	0.21 (0.08–0.50)
Ovarian cyst	31 565	748	0.97 (0.86–1.1)	0.97 (0.85–1.1)	0.97 (0.86–1.11)
Cardiovascular disease	32 081	232	1.02 (0.77–1.35)	1.01 (0.77–1.33)	1.02 (0.78–1.35)
Preeclampsia	30 951	1362	0.86 (0.78–0.94)	0.89 (0.81–0.98)	0.89 (0.81–0.98)
Cardiovascular disease risk factor	28 946	3367	0.97 (0.92–1.03)	0.97 (0.92–1.03)	0.98 (0.93–1.04)

Common disease .	No (Reference) .	Yes .	Model 1 .	Model 2 .	Model 3 .
Overall (n = 61 936)
Depression	55 316	6620	0.94 (0.90–0.98)	0.95 (0.91–0.99)	0.95 (0.91–1.00)
Eating disorder	61 473	464	0.90 (0.77–1.04)	0.91 (0.78–1.05)	0.91 (0.78–1.06)
Anorexia	61 473	463	0.90 (0.77–1.04)	0.91 (0.78–1.05)	0.94 (0.80–1.09)
Bulimia	61 643	293	0.85 (0.66–1.10)	0.84 (0.65–1.09)	0.89 (0.68–1.16)
PCOS	1730	206	0.39 (0.25–0.60)	0.45 (0.28–0.72)	0.44 (0.28–0.71)
Ovarian cyst	58 637	3299	1.12 (1.06–1.18)	1.12 (1.06–1.18)	1.12 (1.06–1.18)
Cardiovascular disease	61 444	492	0.95 (0.81–1.11)	0.96 (0.82–1.13)	0.98 (0.83–1.15)
Heart failure	61 852	84	1.12 (0.80–1.57)	1.12 (0.80–1.59)	1.18 (0.85–1.65)
Myocardial infarction	61 830	106	0.86 (0.66–1.12)	0.87 (0.67–1.13)	0.90 (0.68–1.17)
Angina pectoris	61 781	155	0.89 (0.72–1.10)	0.91 (0.74–1.13)	0.94 (0.76–1.18)
Stroke	61 735	201	1.07 (0.87–1.31)	1.05 (0.86–1.30)	1.06 (0.86–1.30)
Preeclampsia	59 171	2765	0.92 (0.86–0.97)	0.95 (0.89–1.01)	0.96 (0.90.-1.02)
Cardiovascular disease risk factor	55 086	6850	0.95 (0.92–0.99)	0.96 (0.93–1.00)	0.96 (0.93–1.00)
Diabetes	61 192	744	0.88 (0.79–0.99)	0.90 (0.80–1.01)	0.91 (0.81–1.02)
Hypertension	56 738	5198	0.96 (0.92–1.00)	0.97 (0.93–1.01)	0.98 (0.94–1.02)
Hyperlipidemia	59 977	1959	0.99 (0.93–1.06)	1.00 (0.93–1.07)	1.01 (0.94–1.09)
Subset of women who answered No to Did you ever have cervical or ovarian surgery? (n=41 939)
Depression	37 559	4380	0.92 (0.87–0.97)	0.93 (0.88–0.98)	0.93 (0.89–0.99)
Eating disorder	41 624	316	0.82 (0.69–0.98)	0.84 (0.7–1.01)	0.85 (0.71–1.02)
PCOS	41 797	142	0.23 (0.12–0.42)	0.27 (0.14–0.51)	0.27 (0.14–0.51)
Ovarian cyst	40 877	1062	1.01 (0.92–1.11)	1.01 (0.91–1.11)	1.02 (0.92–1.12)
Cardiovascular disease	41 620	319	1.01 (0.82–1.23)	1.02 (0.83–1.24)	1.03 (0.84–1.26)
Preeclampsia	40 166	1773	0.90 (0.83–0.96)	0.92 (0.86–1.00)	0.93 (0.86–1.00)
Cardiovascular disease risk factor	37 374	4565	0.96 (0.92–1.00)	0.96 (0.92–1.01)	0.97 (0.93–1.01)
Subset of women who never received hormone treatment and answered No to Did you ever have cervical or ovarian surgery? (n=32 313)
Depression	29 004	3309	0.9 (0.84–0.97)	0.91 (0.85–0.97)	0.91 (0.85–0.98)
Eating disorder	32 074	239	0.83 (0.69–1.01)	0.86 (0.71–1.04)	0.87 (0.72–1.06)
PCOS	32 243	70	0.19 (0.08–0.47)	0.21 (0.08–0.50)	0.21 (0.08–0.50)
Ovarian cyst	31 565	748	0.97 (0.86–1.1)	0.97 (0.85–1.1)	0.97 (0.86–1.11)
Cardiovascular disease	32 081	232	1.02 (0.77–1.35)	1.01 (0.77–1.33)	1.02 (0.78–1.35)
Preeclampsia	30 951	1362	0.86 (0.78–0.94)	0.89 (0.81–0.98)	0.89 (0.81–0.98)
Cardiovascular disease risk factor	28 946	3367	0.97 (0.92–1.03)	0.97 (0.92–1.03)	0.98 (0.93–1.04)

Model 1: no covariate (unadjusted), Model 2: adjusted for reproductive factors (age at menarche (<14, ≥14 or missing), adult menstrual cycle regularity (regular, irregular or missing), number of children (time-dependent variable) and premenopausal oral contraceptive use (mini and combination pill use as time-dependent variables) and risk factors of common diseases (education (elementary, intermediate, college, other or missing), BMI (kg/m², <18.5, 18.5–24.9, 25–29.9, ≥30), physical activity at age 18 (h/week, 0, <1, 1–2, 3–5, ≥5 or missing), ever smoked for 1 year or 100 cigarettes (yes, no or missing), alcohol consumption (g/day, 0, <10, 10–24, ≥25 or missing))Model 3: Model 2 and all common diseases examined. The hazard ratios (HR) represent the risk of becoming naturally menopausal at a given age. HR < 1 indicate later menopause and HR > 1 indicate earlier menopause as compared with the reference category. Significant associations are denoted in bold.

Discussion

Among the 13 common diseases studied, we found that PCOS strongly delays menopause, while depression and preeclampsia also decelerates menopause to a modest extent. We found no statistically significant overall association between cardiovascular diseases, risk factors of cardiovascular diseases, eating disorders, ovarian cyst and ANM.

‘PCOS’ affects ~10% of women worldwide and can be diagnosed at any age before ‘menopause’ (Alsamarai et al., 2009; Daniilidis and Dinas, 2009; Welt and Carmina, 2013). A main underlying problem with PCOS is a hormonal imbalance, which typically leads to menstrual irregularities and infertility. In a previous study of 85 women with PCOS who were followed-up, Tehrani et al. reported that the reproductive life span of PCOS women was found to be on average 2 years longer than women who ovulated normally (Tehrani et al., 2010). The longer reproductive life span in women with PCOS compared to that of women without PCOS has been partly attributed to higher adrenal and ovarian androgen levels. A higher number of follicles has also been found to be contained within the ovaries of women with PCOS, compared with control women at any age, which may be explanatory (Pigny et al., 2003; Broekmans et al., 2004; Mulders et al., 2004; Nikolaou and Gilling-Smith, 2004; Schmidt et al., 2011).

Although there are no observational studies directly examining depression as a determinant of ANM, our result on the independent association between depression and late menopause is not in agreement with studies suggesting the opposite direction. Since the highest prevalence of major depression in women has been observed in the age window during the transit to menopause (35–45 years), depression was reviewed by (Harlow and Signorello 2000) as a precursor to the premature cessation of menstrual cycles. The role of depression as a cause of early menopause was suggested to act through the pharmacological effects of treatment, as there is indication that the use of antidepressants or tranquilizers hampers ovulation by disrupting hormonal mechanisms (Grodstein et al., 1993). Alternatively, depression could also act as a marker for a premature decline in ovarian function; declining estrogen levels precede and predispose women to depression rather than being the result of its presence or treatment (Harlow and Signorello, 2000). However, depression is common among women with PCOS. It is possible that the same biological mechanisms that drive delayed menopause in women with PCOS also exert an influence on women with depression.

The association of preeclampsia with delayed ANM is novel and has not been previously described. Preeclampsia is a relatively common pregnancy-related hypertensive disorder that negatively impacts vascular health. In a Dutch study, women with a history of preeclampsia were found to have significantly lower ovarian reserve status estimated by serum levels of anti-Mullerian hormone (AMH) when compared to those of women with normotensive pregnancies, suggesting that vascular mechanisms are involved in the acceleration of ovarian aging in women with preeclampsia (Yarde et al., 2014). Although our findings on preeclampsia and later ANM were not in agreement, it should be noted that the degree of preeclampsia was not characterized in our study, whereas the Dutch cohort was well phenotyped. In addition, although reduced, AMH levels in women of the Dutch cohort were still within the normal range for reproductive years, and none of them had yet reached menopause when the study was conducted (Yarde et al., 2014). As women with PCOS have been associated with a 3-fold or higher risk of developing preeclampsia (Boomsma et al., 2006), it is plausible that delayed menopause in women with PCOS could have influenced our finding on delayed menopause in women with preeclampsia.

The association between PCOS, depression, preeclampsia and ANM became more pronounced in the sensitivity analyses, strengthening our conclusions that these common diseases delay menopause. On the contrary, the initial observation that ANM may be lower in women with ovarian cyst was completely diminished in the subset of women who have never undergone gynecological surgery, suggesting that there could be some misclassification bias (some women could have selected natural menopause as the reason for the cessation of menstrual bleeding instead of gynecological surgery). Previous prospective studies have reported a critical reduction in the ovarian reserve of young women who undergo conservative fertility-sparing surgery for ovarian cysts, although there was no apparent removal of ovarian tissue at the time of surgery (Lind et al., 2015). In addition, the surgical removal of one ovary has been previously associated with acceleration of menopause by ~1 year (Yasui et al., 2012; Bjelland et al., 2014). Substantial reduction of ovarian reserve has been observed also in young women who preserve both ovaries when undergoing hysterectomy, which suggests that vascular impairment in the ovarian blood supply can be induced by the surgical procedures themselves (Trabuco et al., 2016).

The main strength of our study is the large sample size. We have expanded the scope of traditional risk factors studied in connection to early menopause to include premenopausal diagnoses of 13 common diseases. All the data used in this study were self-reported, and so some recall bias is possible, but it is unlikely that any bias was different in the strata of different factors considered. Furthermore, any misclassification of age at menopause would—if anything—tend to reduce the observed differences (Rodstrom et al., 2005). The use of self-reported disease history in this study is an advantage as primary care is not covered by the Swedish national registers. The Swedish Inpatient Register was launched in 1964 to collect information regarding inpatients at public hospitals (psychiatric diagnoses from 1973) but complete coverage did not begin until 1987. Outpatient visits were included in the register later from 2001. Some conditions such as PCOS which develops in early adult life may thus not have been properly captured in the registers. Although the mean ages of diagnoses for our significant results were relatively young, risk of misclassification is still possible in women for whom the common diseases were diagnosed at approximately the same age as ANM (i.e. reverse causation), as the transition to menopause could have begun many years prior to menopause (Harlow and Paramsothy, 2011).

We consider the information derived from the KARMA study to be a valuable source regarding common diseases and female reproductive health. The identification of factors that predict ANM can help in individual's risk assessment and opportunity to improve current treatments. Each woman has a unique biological clock. Since age at menopause itself is a risk factor for a number of disorders, understanding the determinants of menopausal age is important and should be pursued.

Acknowledgements

We thank the two anonymous reviewers whose comments and suggestions helped improve and clarify this manuscript.

Authors’ roles

J.L. and K.A.R.-W. were involved with every stage of the study, including overseeing and contributing to the research design, execution, analysis and results, manuscript drafting and editing. M.E., K.C. and P.H. assisted with every stage of the research design, execution, discussion of analysis and results, manuscript drafting and editing.

Funding

KARMA was financed by the Märit and Hans Rausing's Initiative Against Breast Cancer. K.R.W. is supported by the Swedish Society of Medicine and by Stockholm County Council. J.L. is a recipient of an Alex and Eva Wallström Foundation award.

Conflict of Interest

None declared.

References