Symptomatic menopausal transition and risk of subsequent stroke

Introduction

Menopausal transition is defined as the period before the final menstrual period in late reproductive life. Menopause occurs following 12 months of amenorrhea and represents the loss of ovarian follicular function; this typically occurs in women between 45 and 55 years of age. Approximately 1.5 million women experience menopausal transition every year in the United States (U.S. Census Bureau, 2005). Vasomotor symptoms, insomnia, depression, and vaginal dryness are major problems during menopausal transition (Santoro, Epperson & Mathews, 2015).

Stroke is the second leading cause of death and third leading cause of disability-adjusted life-years lost worldwide. After stroke, the survivors still have high rates of mortality, recurrent stroke, and disability for months to years (Hankey, 2017). The Chinese populations in China and Taiwan were reported to have higher stroke incidence (age-standardized annual first-ever stroke incidence: range 205–584 per 100,000 patients) than white populations (range 170–335 per 100,000 patients) (Tsai, Thomas & Sudlow, 2013). Thus, stroke prevention for populations at high risk for stroke remains an important issue in Taiwan.

Symptoms of menopausal transition, such as vasomotor symptoms and night sweats, may have a negative impact on a woman’s quality of life during her midlife (Thurston et al., 2012). The occurrence of vasomotor symptoms is associated with low serum estrogen levels (Erlik, Meldrum & Judd, 1982). In addition, vasomotor symptoms have been found to be associated with chronic insomnia (Ohayon, 2006), cardiovascular disease (Rossouw et al., 2007), and insulin resistance (Huang et al., 2017). A growing body of evidence has also shown the association of vasomotor symptoms with poor control of dyslipidemia in obese women (Thurston et al., 2012), atherosclerosis, vascular endothelial dysfunction (Thurston et al., 2017), increased prothrombotic events, smoking habits (Gallicchio et al., 2014), and disturbed sympathovagal tone (Thurston et al., 2012; Gallicchio et al., 2014). These deleterious effects accompanied by the symptomatic menopausal transition may potentially contribute to stroke as well as cardiovascular disease. However, no previous studies have assessed the direct link between symptomatic menopausal transition and the risk of stroke with a long-term follow-up. We hypothesized that symptomatic menopausal transition may have insidious adverse effects on the cerebrovascular system.

Therefore, we conducted this nationwide, large-scale, population-based cohort study to determine whether symptomatic menopausal transition is independently linked to stroke.

Materials and Methods

Study population

We used a 4-year look-back period (1996–1999) to identify subjects with newly diagnosed symptomatic menopausal transition by excluding pre-existing diagnosis (Fig. 1). Symptomatic menopausal transition was defined by at least three records made by gynecologists within a 1-year period (ICD-9-CM code 627.2: symptomatic menopausal transition). Subjects who were diagnosed with symptomatic menopausal transition during the look-back period were excluded from this study. We identified 457,996 subjects, including 41,516 with newly diagnosed symptomatic menopausal transition and 416,480 without symptomatic menopause, from the NHIRD between January 1, 1996 and December 31, 2013. The index date of the symptomatic menopausal transition group was defined as the first date of the symptomatic menopausal transition diagnosis; this date was also assigned to the matched comparison participants as the date of entry into the study. The exclusion criteria for the study and comparison groups were as follows: (1) age, <45 or >100 years, (2) history of breast cancer before index date (breast cancer treatments can cause menopausal symptoms or premature menopause) (Cusack et al., 2013), (3) history of oophorectomy before index date, (4) history of stroke before index date, (5) patients who did not survive or were followed up <30 days (a short exposure time to symptomatic menopausal transition may not be sufficient to cause an incident stroke), and (6) patients who were not matched to the comparison group. Finally, 22,058 women with newly diagnosed symptomatic menopausal transition with no history of stroke were enrolled between January 1, 2000 and December 31, 2013. Additionally, 22,058 comparison participants were selected by propensity score matching. Both groups were followed up until the date of death, developing stroke, withdrawal from the National Health Insurance, or the end of 2013. The mean follow-up period was 8.5 years (standard deviation 4.7 years).

Results

Table 1 shows the baseline characteristics of the two groups. A total of 44,116 women (22,058 women diagnosed with symptomatic menopausal transition and 22,058 matched comparison participants) with no history of stroke were enrolled in the study. The mean age of the enrolled subjects was 52.6 ± 7.4 years. The age, monthly income, comorbidities, Charlson’s comorbidity index score, clinic visit frequency, and long-term medications were balanced between the two groups, except a predilection for HRT in the symptomatic menopausal transition group.

Risk of stroke after perimenopausal transition

During the follow-up period, 1,184 women (5.37%) of the comparison group and 2,274 (10.31%) of the symptomatic menopausal transition group developed stroke (Fig. 1). The incident rates of stroke were 8.57 (95% CI [8.08–9.06] per 1,000 person-years) and 11.17 (95% CI [10.71–11.63] per 1,000 person-years) in the comparison and symptomatic menopausal transition groups, respectively (Table 2). The cumulative incidence of stroke was higher in the symptomatic menopausal transition group than in the comparison group (Fig. 2). The crude risk of stroke in the symptomatic menopausal transition group was statistically significantly higher compared to the comparison group (SHR, 1.31; 95% CI [1.22–1.41]; P < 0.001). Additionally, two multivariable models were used to adjust the risk of subsequent stroke. After adjusting for all confounders in Table 1 (Model 2, Table 2) and the full adjustment model (Model 3, Table 2), the risk of subsequent stroke remained consistently higher in the symptomatic menopausal transition group than that of the comparison group (SHR, 1.37; 95% CI [1.27–1.48]; P < 0.001 and SHR, 1.30; 95% CI [1.21–1.40]; P < 0.001, respectively).

Table 2:

Incidence and risk of stroke in patients with symptomatic menopausal transition and their matched subjects.

Group	Event	PY	Incidencea	Model 1		Model 2		Model 3
				cSHRb (95% CI)	P-value	aSHRb (95% CI)	P-value	aSHRb (95% CI)	P-value
Comparison	1,184	138,138.72	8.57 (8.08–9.06)	Reference		Reference		Reference
Symptomatic menopausal transition	2,274	203,586.37	11.17 (10.71–11.63)	1.31 [1.22–1.41]	<0.001	1.37 [1.27–1.48]	<0.001	1.30 [1.21–1.40]	<0.001

DOI: 10.7717/peerj.7964/table-2

Notes:

Model 1: crude hazard ratio compared with the propensity-score matched comparison subjects.

Model 2: adjusted for all variables listed in Table 1.

Model 3: adjusted for all variables listed in Table 1, as well as comorbidities and medications considered as time-dependent covariates.

aSHR, adjusted subhazard ratio; cSHR, crude subhazard ratio; CI, confidence interval; PY, person-years.

a Per 1,000 person-years.

b Death before developing stroke was considered a competing risk.

Subgroup analyses

The risk of subsequent stroke was further estimated throughout the baseline clinical characteristics, including individual comorbidities and long-term medications (Fig. 3). We found that there was no statistically significant interaction between symptomatic menopausal transition and all subgroups except hypertension and use of antihypertensives (P_interaction < 0.001 and 0.003, respectively). The risk of stroke was further increased in symptomatic menopausal women without hypertension or without the use of antihypertensives (Fig. 3).

Discussion

The present study is the first and largest study to investigate the association between symptomatic menopausal transition and subsequent development of stroke during up to 14 years of follow-up. In this prognostic model, symptomatic menopausal transition has added statistically significantly prognostic power for stroke to standard risk factors. Women with symptomatic menopausal transition had a 30% increased risk of subsequent stroke compared to matched comparison participants. The effect of symptomatic menopause on stroke was consistent across all subgroups except age, comorbidity, hypertension, and the use of antihypertensives. Moreover, this effect was durable and increased when the duration of menopausal transition lasted longer.

Previous studies have shown that there is a possible link between stroke and symptomatic menopausal transition. Women experiencing vasomotor symptoms tend to have lower circulating estradiol levels and other adverse cardiovascular risk factors such as obesity, smoking, and psychosocial problems (Gast et al., 2010). Among them, low socioeconomic status, depression, and anxiety have the closest relationship between these factors and cardiovascular diseases (Thurston et al., 2012). Additionally, dysregulation of adipocyte-derived hormones, especially the higher circulating leptin levels, lower adiponectin levels, and higher leptin-to-adiponectin ratios, was reported in women transitioning through menopause (Ben Ali et al., 2011). A recent study also showed that the profile of adipokines (leptin levels, adiponectin levels, and leptin-to-adiponectin ratios) was closely related to the severity of vasomotor symptoms (Huang et al., 2017). Leptin, for example, is a widely investigated adipokine and its production is proportional to the adipose tissue mass and its secretion is pulsatile with the characteristics of diurnal rhythm. The serum leptin level is highest at night and lowest in the morning; as the leptin level rises, it increases the appetite via its effects on the brain, resulting in weight gain and obesity (Kelesidis et al., 2010). Moreover, both menopause and low levels of estrogens can cause systemic inflammation and neuroinflammation (Au et al., 2016). Therefore, a low estrogen level, dysregulation of adipokine production, high BMI, and proinflammatory status may be potential underlying pathogenic factors for stroke in women with symptomatic menopausal transition.

In the present study, the risk of stroke was especially high in women with early onset of symptomatic menopause, not using antihypertensives, in the absence of comorbidity, and without hypertension. Lisabeth et al. (2009) reported that natural menopause before age 42 was associated with a higher risk of ischemic stroke. Another study also suggested that early menopause is associated with an increased risk of ischemic stroke (Rocca et al., 2012). Early menopause may expose women to longer duration of systemic- or neuro-inflammation contributing to a higher risk of subsequent stroke. We also observed an interaction between symptomatic menopausal transition and comorbidity whereby the risk of stroke was highest in women without any comorbidity and the effect of symptomatic menopausal transition on stroke was significantly reduced in women with a comorbid condition. Furthermore, since blood pressure control is one of the most important risk factors for stroke, especially ischemic stroke in women, we evaluated the interaction between symptomatic menopausal transition and hypertension or use of antihypertensives (Gorgui et al., 2014). Our results revealed that the risk of stroke was much higher in women without or without control of hypertension. Because hypertension is already a strong risk factor for stroke, the impact of symptomatic menopausal transition on subsequent stroke becomes smaller in the hypertension or with antihypertensives subgroups than in the non-hypertension or without antihypertensives subgroups. Our findings suggest that these subgroups are at the highest risk for stroke and that the prevention of subsequent stroke should be considered in addition to traditional management of symptomatic menopause.

Vasomotor symptoms are some of the major problems leading middle-aged women to seek medical advice and treatment for symptomatic menopause. Although HRT plays a key role in treating symptomatic menopausal transition (Erlik, Meldrum & Judd, 1982; Huang et al., 2008), HRT has been reported to increase the risk of breast cancer (Russo & Russo, 2006), cerebral (Rossouw et al., 2002) or cardiac vascular events (Wilson, Garrison & Castelli, 1985), and new-onset atrial fibrillation (Tsai et al., 2016; Bretler et al., 2012). We found that the risk of stroke remained higher in women with symptomatic menopausal transition after adjustment for HRT. Furthermore, there was no statistically significant interaction between symptomatic menopausal transition and HRT.

The strength of our study is its large sample size. A large sample size is more likely to produce sufficiently narrow CIs. In this case, the narrow CIs of our results (e.g., 95% CI of SHR [1.21–1.40]; Table 2) indicate adequate precision. We did not conduct ex ante power calculations because our study utilized medical claims retrospectively retrieved from the NHIRD. However, the retrospective power of this study calculated using an observed effect size was determined entirely by the P-value which was already observed. Thus, post hoc sample size calculation or power analysis provides no additional information beyond the effect size. Additionally, the number of events observed may be important when modeling survival data. Previous studies have shown that 10–50 events per variable may be necessary to assure accurate estimation of regression coefficients, standard errors, and CIs (Austin, Allignol & Fine, 2017; Peduzzi et al., 1995). A 14-year follow-up was more than appropriate to detect stroke events in middle-aged women (3,458 stroke events were observed during the study period). However, there were some limitations to this study. First, the NHIRD did not include information relevant to stroke such as smoking history, BMI, physical activity, daily salt intake, blood pressure, and blood glucose and lipid profiles. These unmeasured covariates could have an impact on the outcome, even after balancing the baseline clinical characteristics with propensity score matching. Second, the diagnosis of symptomatic menopausal transition was mainly based on the ICD-9-CM codes; this could lead to the underestimation of cases with less severe symptoms. However, the risk of stroke may be underestimated if the cases were misclassified as matched comparison participants. Third, we cannot exclude the “healthy user bias”; patients who seek medical advice might be compliant with healthy lifestyles or treatments that could reduce the occurrence of adverse events (Brookhart et al., 2007). Finally, since most of the population in Taiwan are Han Chinese, further studies are required to clarify whether these findings can be generalized to other ethnicities.

Conclusion

This large-scale long-term cohort study demonstrated that symptomatic menopausal transition was statistically significantly associated with an increased risk of subsequent stroke. Women with early menopausal transition, without comorbid condition, without hypertension, or using antihypertensives are at a higher risk of stroke. In order to fully elucidate whether stroke prevention improves cerebrovascular outcomes in the care of symptomatic menopausal transition further studies are required.

Supplemental Information