Reproductive and hormonal factors and risk of cognitive impairment among Singapore Chinese women

Abstract

Background:

Epidemiologic studies suggest that declining estrogen level in menopause may play an important role in the pathogenesis of dementia and contribute to increased risk of cognitive impairment in women. Most previous studies have been conducted in Western population to investigate the relations of the length of reproductive periods and use of HRT with risk of cognitive function and dementia, but the findings are inconclusive. Relevant evidence among Asian populations is limited.

Objectives:

To evaluate the association between reproductive and hormonal factors, and risk of cognitive impairment in Chinese women with natural menopause.

Study Design:

The Singapore Chinese Health Study is a population-based study that recruited participants aged 45–74 years between 1993 and 1998, and the current study included 8,222 women from this cohort who had natural menopause, complete data on reproductive factors and hormonal therapies at baseline (1993–1998), follow-up 1 (1999–2004) and follow-up 2 interviews (2006–2010), and cognitive function evaluated at ages 61 to 96 years using the Singapore Modified Mini-Mental State Examination (SM-MMSE) during the follow-up 3 visits (2014–2016). Multivariable logistic regression models were used to estimate odds ratios (ORs) and 95% confidence interval (CI) for the risk of cognitive impairment.

Results:

Compared to women with menopause at 50–54 years of age, the OR (95% CI) were 1.67 (1.32, 2.11), 1.24 (1.08, 1.44), and 1.06 (0.87, 1.29) for women who experienced menopause before 45 years, at 45–49 years of age, and after 54 years, respectively. Compared to women with 35–39 reproductive years from menarche to menopause, the OR (95% CI) were 1.28 (1.11, 1.48) for women with <35 reproductive years. Furthermore, compared with women who had 1–2 children, the OR (95% CI) were 1.27 (1.04, 1.55) for women who had more than 5 children, and the risk increased significantly by 5% per child birth (OR: 1.05; 95% CI: 1.01, 1.09). Compared to those who had never used oral contraceptives, women with short-term use (≤ 5 years) of oral contraceptives had 26% lower odds of having cognitive impairment (OR: 0.74; 95% CI: 0.63, 0.87), while the association was not statistically significant for those used for more than 5 years (OR: 0.87; 95% CI: 0.68, 1.13). Women who used HRT had a 39% lower odd of getting cognitive impairment compared to non-users (OR: 0.61; 95% CI: 0.46, 0.80).

Conclusions:

Our data suggested that shorter reproductive years and higher parity were associated with higher risk of cognitive impairment in late life, while use of oral contraceptives and hormone replacement therapy were associated with decreased risk. As the population ages, understanding how these factors affect late-life cognitive function in women may help health professionals develop preventive measures targeting lifetime estrogen exposure from endogenous or exogenous sources.

Condensation:

Shorter reproductive period is associated with a higher risk of cognitive impairment in later life among Singapore Chinese women.

Introduction

Cognitive impairment is the key clinical manifestation of dementia. As the global population ages, the prevalence of dementia will dramatically increase and it is estimated that more than 152 million people will be affected by dementia by 2050.¹ Studies have shown that postmenopausal women had a higher risk of developing dementia and Alzheimer diseases than men of similar age.^2–4 A woman’s reproductive period begins with spontaneous ovarian function at menarche, and ends with a dramatic drop in estrogen levels at menopause. Epidemiologic studies suggest that declining estrogen level in menopause may play an important role in the pathogenesis of dementia and contribute to increased risk of cognitive impairment in women.^5–8 Indeed, a series of epidemiological, interventional and animal studies have demonstrated the neuroprotective and neurotrophic effects of estrogen.^9–11

In addition to endogenous exposure throughout the reproductive period, exogenous exposure can come from the use of oral contraceptives and hormone replacement therapy (HRT). A number of epidemiologic studies have investigated the association between the length of reproductive periods and use of HRT, and risk of cognitive function and dementia, but the findings are inconclusive.^{5–7, 12–15} A systematic review and meta-analysis of observational studies has reported that longer reproductive period was associated with decreased risk of cognitive decline,¹⁶ and another has reported that use of HRT was associated with decreased risk of dementia.¹⁷ However, a Cochrane systematic review of clinical trials concluded that HRT did not prevent cognitive decline in postmenopausal women,¹⁸ and some observational studies also reported null or opposite association.^{12, 13, 15}

Most previous researches have been conducted in Western population, while some studies reported that the age at menarche of Chinese women was older than that of Caucasian women, while the age at menopause was lower than that of Caucasian women, indicating a shorter duration of reproductive periods.^19,20 Some studies revealed that estrogen concentration in Chinese women were generally lower than that in Caucasian women in both pre- and post-menopause periods.^{21, 22} Therefore, studies in Asian populations are still needed. Two cross-sectional studies in Guangdong and Zhejiang provinces reported that longer duration of reproductive period was associated with better cognitive function among Chinese postmenopausal women,^{5, 6} however, the reproductive variables were inquired at the same time of cognitive assessment and recall bias was possible.

Using prospective data from the Singapore Chinese Health Study, we aimed to evaluate the length of reproductive period, parity and the use of exogenous hormones with the risk of cognitive impairment in women who had natural menopause.

Material and Methods

Study population

The Singapore Chinese Health Study is a general population-based prospective cohort established to investigate environmental, dietary and lifestyle factors in relation to risk of chronic diseases. The detailed study design and characteristics of the study participants have been reported previously.²³ Briefly, a total of 63,257 participants (35,303 women and 27,954 men) aged 45 to 74 years was recruited from two major dialect groups (Hokkiens and Cantonese) of Chinese living in Singapore between 1993 and 1998. At baseline, trained interviewers conducted face-to-face interviews using a structured questionnaire to collect information about demographics, height, weight, dietary intake, cigarette smoking, alcohol consumption, physical activity, medical history and family history of cancer.

Three follow-up visits were conducted after baseline interview. Participants were interviewed via telephone during the follow up 1 (1999–2004) and follow up 2 (2006–2010) visits to update lifestyle factors and medical histories. In the follow up 3 visits (2014–2016), trained interviewers went to the participants’ homes to evaluate ageing related outcomes that included status of cognition, physical function, and abilities to handle activities of daily living. Lifestyle factors and medical histories were also updated at the home visits.

For the current analyses, we only included women who completed cognitive testing during the follow up 3 visits. We sent 1–2 invitation letters to the surviving participants before the home visits; some participants could not be contacted, and some were unable to participate in the follow-up visit because of serious diseases or severe cognitive impairment. Meanwhile, the third follow-up visits were only conducted until February 2016 due to limited funding. The participants in the follow up-3 visits were not purposely selected. Among the 45109 surviving individuals at the beginning of the follow-up 3 visit, 17,107 were successfully contacted and agreed to participate. We further excluded 104 participants because of inability to complete the SM-MMSE test (mute, blind, or deaf) and 55 participants of missing responses to certain items, leaving 16,948 participants (10,034 women and 6914 men) with complete data on cognitive function test. Among the 10,034women, we further excluded those with the following conditions: 327 women with missing data of ageat menopause or who started HRT before menopause, 9 women with unreasonable age at menopause (<20 or >60 years old), 23 women with missing data on age at menarche, and 72 women with missing data on HRT use. Previous studies indicated that the relation between age at menopause and cognitive decline could be modified by the reason for menopause, whether natural or due to surgery, radiotherapy or medication;^{24, 25} thus, we further excluded 1381 women who did not have natural menopause or who did not report whether menopause was natural. Finally, 8222 women with natural menopause and complete data on reproductive and hormonal factors were included in the analyses (Figure 1).

Figure 1. Flow chart of sample selection.

The study protocol was approved by the Institutional Review Board of the National University of Singapore. Written informed consent forms were obtained from all participants.

Assessment of reproductive and hormonal factors and other exposures

Detailed information on reproductive and hormonal factors were collected during the baseline interview, and including were age at menarche (<11, 11–12, 13–14, 15–16, and ≥17 years old), age at first livebirth (<15, 15–17, 18–20, 21–25, 26–30, 31–35, and ≥36 years old), number of children, use of oral contraceptive for one month or longer and years of use, age at menopause (<40, 40–44, 45–49, 50–54, and ≥55 years old), type of menopause (natural, surgery, radiotherapy or medication) and HRT use and types of use. Data on age at menarche (exact age) were updated at follow up 1 visits and data on menopausal status, age at menopause (exact age), type of menopause and HRT use were updated at both follow-up 1 and follow-up 2 visits. Since inquiry of some variables were repeated multiple times, we have established a priori data usage principle: continuous variable first if available rather than categorical variable, data in baseline or early follow-up visit if available to reduce the probability of false recall. Therefore, data on age at first livebirth, use of oral contraceptive and number of children were derived from the baseline interviews, and data on the age at menarche, menopausal status, age at menopause, type of menopause and HRT usage were derived from the follow-up 1 visits, and if not available, from the follow-up 2 visits. Reproductive period was calculated as the period extending from age at menarche to age at menopause in years.

The physical activity level was measured by hours per week spent on moderate activities, strenuous sports and vigorous work, and then categorized as <0.5 h/week, 0.5–3.9 h/week and ≥4.0 h/week. Body mass index (BMI) was calculated as weight divided by the square of height (kg/m²) and categorized using the World Health Organization categories for weight recommendation in Asians: BMI <18.5 kg/m² as underweight, 18.5–22.9 kg/m² as normal, 23.0–27.4 kg/m² as overweight and ≥27.5 kg/m² as obesity.²⁶ Total energy intake and the alternate Mediterranean diet pattern score were computed using the validated food frequency questionnaire in this cohort at recruitment and the Singapore Food Composition Database developed specifically for this cohort.²⁷ We have recently demonstrated that the Mediterranean diet pattern was associated with a lower risk of cognitive impairment in this cohort.²⁸ Depressive symptoms were assessed at the follow-up 3 visits by using the 15-item geriatric depression scale (GDS-15), and depression was defined as a GDS-15 score ≥5.²⁹

Assessment of cognitive function

At the follow-up 3 visit, the Singapore Modified Mini-Mental State Examination (SM-MMSE) was used to assess cognitive function. The SM-MMSE is a modified version of the Mini-Mental State Examination (MMSE).³⁰ The test is composed of 30 items to assess orientation, immediate and delayed recall, attention, language and visuospatial ability, and it has been validated in Singapore population.³¹ The score of SM-MMSE ranges from 0 to 30 and higher scores indicate better cognitive function. A cut-off of 23/24 for MMSE was commonly used to define cognitive impairment in Western countries.³² However, previous study found that MMSE was significantly associated with education level.^33,34 Hence, in our study, we used education-specific cut-off points that originated from the Shanghai Dementia Survey due to the similar education levels between our study participants and their participants.³⁵ The cut-off points were 17/18 for subjects with no formal education, 20/21 for subjects with primary school education and 24/25 for those with secondary school or higher education.

The SM-MMSE testing was performed through face-to-face interviews by trained interviewers in a quiet environment. All interviewers were systematically trained by an experienced geriatric epidemiologist in our team (L. Feng) before the study. All interviews were recorded and 20% of recordings were randomly selected for quality control. If the interviewers were found not to be compliant with the protocol, they were re-trained and re-assessed before they were allowed to conduct further interviews with participants.

Statistical analysis

Demographic factors of participants by cognitive status and use of HRT were presented by proportions for categorical variables and means with standard deviations (SD) for continuous variables. Multivariable logistic regression models were used to calculate odds ratios (ORs) and 95% confidence intervals (CIs) for the associations. We adjusted for potential confounders in sequential steps. Model 1 adjusted for demographic factors: age, year of recruitment (1993–1995, 1996–1998), dialect group (Hokkiens, Cantonese), marital status (married, widowed, separated/divorced, never married), education level (no formal education, primary school, secondary school or higher). Model 2 further adjusted for dietary and lifestyle factors: cigarette smoking (never, former, current smoker), tea intake (none, monthly, weekly, daily), coffee intake (none/less than daily, 1 cup/day, sleep duration (≤5, 6–8, ≥9 hours/day), physical activity (<0.5 h/week, 0.5–3.9 h/week and ≥4.0 h/week, BMI (<18.5, 18.5–22.9, 23–27.4 and ≥27.5 kg/m²), total energy intake (kcal/day), and the alternate Mediterranean diet pattern score (including alcohol consumption, in quartiles). Model 3 additionally adjusted for baseline history of hypertension, diabetes, cardiovascular disease (CVD) and cancer, and mutually adjusted for the reproductive and hormonal variables. Linear trends were analyzed by using the median values of the exposure categories as continuous variables.

We conducted an additional analysis to explore the association between type of menopause (surgical vs. natural) and risk of cognitive impairment, and in this analysis, 1324 participants with surgical menopause were included. To test the robustness of our findings, we conducted a series of sensitivity analyses: 1) we excluding those who had cancer or CVD at baseline; 2) we used the 23/24 as cut-off point for SM-MMSE to define cognitive impairment; 3) we further adjusted for depressive symptoms as a covariate in the multivariate models or excluded those with depression at follow-up 3 visits, because previous studies indicated that depressive symptoms could be affected by reproductive factors and on the other hand depression could affect the later-life cognitive function;^{36, 37} 4) we repeated our analyses by excluding participants with inconsistent data on age at menarche and age at menopause, or using the categorical variables of age at menarche and age at menopause collected from baseline interviews.

We also compared the baseline characteristics of women included in and excluded from the analyses, as well as the reproducibility of the self-reported exposure data across the multiple interviews. All analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC), and the statistical significance was two-sided P values <0.05.

Data availability

A de-identified dataset can be shared with researchers who meet the criteria and fulfil justification for access to the data for verification of our findings.

Results

The participants had a mean age (standard deviation, SD) of 53.4 (6.4) years at enrollment and 73.6 (6.6) years at SM-MMSE measurement. Among the 8222 postmenopausal women, 1332 (16.2%) were assessed to have cognitive impairment. Compared to those with normal SM-MMSE scores, women with cognitive impairment were older, had lower educational level, less likely to be married with living spouses at recruitment, but more likely to be smokers and more likely to have chronic diseases (Table 1). Compared with HRT users, non-users were generally younger, more likely to be married, to have lower BMI levels, to have a higher education level, to be more active, to eat healthier, but less likely to be current smokers, but similar prevalence of baseline comorbidities (Table A-1).

Table 1.

Baseline characteristics of study participants according to cognitive status^a

	Overall	Cognitive impairment		P Valueb
	(n=8222)	Yes (n=1332)	No (n=6890)
Age at baseline, mean (SD), y	53.4 (6.4)	57.5 (6.7)	52.6(6.1)	<0.001
Age at SM-MMSE measurement, mean (SD), y	73.6 (6.6)	78.0 (6.7)	72.8 (6.3)	<0.001
Body mass index, mean (SD), kg/m2	23.2 (3.3)	23.5 (3.4)	23.1 (3.3)	<0.001
Married with living spouse	6852 (83.3)	1035 (77.7)	5817 (84.4)	<0.001
Dialect group				0.01
Cantonese	4096 (49.8)	705 (52.9)	3391 (49.2)
Hokkien	4126 (50.2)	627 (47.1)	3499 (50.8)
Education level				<0.001
No education	2470 (30.0)	504 (37.8)	1966 (28.5)
Primary school	3600 (43.8)	527 (39.6)	3073 (44.6)
Secondary school or above	2152 (26.2)	301 (22.6)	1851 (26.9)
Current smokers	270 (3.3)	71 (5.3)	199 (2.9)	<0.001
Daily tea drinkers	1467(17.8)	224 (16.8)	1243 (18.0)	<0.001
Daily coffee drinkers	5758 (70.0)	946 (72.0)	4812 (69.8)	0.37
Physical activity				0.05
None	5952 (72.4)	980 (73.6)	4972 (72.2)
0.5–3.9 hours/week	1513 (18.4)	216 (16.2)	1297 (18.8)
≥4 hours/week	757 (9.2)	136 (10.2)	621 (9.0)
Sleep duration				0.006
≤5 hours/day	813 (9.9)	163 (12.2)	650 (9.4)
6–8 hours/day	6931 (84.3)	1091 (81.9)	5840 (84.8)
≥9 hours/day	478 (5.8)	78 (5.9)	400 (5.8)
Alternate Mediterranean diet score, mean (SD)	4.2 (1.7)	4.0 (1.7)	4.2 (1.7)	<0.001
History of disease at baseline
Hypertension	1557 (18.9)	318 (23.9)	1239 (18.0)	<0.001
Diabetes	384 (4.7)	87 (6.5)	297 (4.3)	<0.001
Cardiovascular disease	175 (2.1)	53 (4.0)	122 (1.8)	<0.001
Cancer	165 (2.0)	27 (2.0)	138 (2.0)	0.95

^aData are presented as frequency (percentage) unless otherwise indicated.

^bTwo-sided P values were derived from ANOVA for continuous variables and Chi-square test for categorical variables.

After adjusting for potential confounders, compared with women who experienced natural menopause between the age of 50 and 54 years old, the OR (95% CI) were 1.67 (1.32, 2.11), 1.24 (1.08, 1.44), and 1.06 (0.87, 1.29) for women who experienced menopause before 45 years, between the age of 45 and 49 years, and after 54 years, respectively. Women with <35 reproductive years had 28% (OR: 1.28; 95% CI: 1.11, 1.48) higher odds of cognitive impairment compared to women with 35–39 reproductive years. The risk of cognitive impairment was significantly reduced by 3% for every one-year increment in age at menopause (OR: 0.97; 95% CI: 0.96, 0.99), as well as for duration of reproductive period (OR: 0.97; 95% CI: 0.96, 0.98) (Table 2). Compared with women who had 1–2 children, the OR (95% CI) were 1.27 (1.04, 1.55) for women who had more than 5 children, and the risk increased significantly by 5% per child birth (OR: 1.05; 95% CI: 1.01, 1.09). No significant association was found between age at menarche or age at first livebirth and risk of cognitive impairment.

Table 2.

Relations of reproductive and hormonal factors with risk of cognitive impairment (n=8222)

	Case/N	Model la	Model 2b	Model 3c
		OR (95% Cl)	OR (95% Cl)	OR (95% Cl)
Age at menarche
<13 years	260/2034	1.00	1.00	1.00
13–14	459/3046	1.01 (0.85,1.20)	1.01 (0.85,1.20)	1.02 (0.86,1.22)
15–16	430/2325	1.08 (0.90,1.30)	1.09 (0.91,1.31)	1.13 (0.94,1.36)
>16 years	183/817	1.16(0.92,1.46)	1.14(0.90,1.44)	1.18 (0.93,1.49)
P for trend		0.15	0.18	0.09
Per 1-year increase in age at menarche		1.03 (0.99,1.07)	1.03 (0.99,1.06)	1.04 (1.00,1.07)
Age at first livebirthd
<20 years	249/1180	1.00	1.00	1.00
21–25 years	534/3001	1.02 (0.85,1.22)	1.03 (0.86,1.24)	1.06 (0.88,1.28)
26–30 years	338/2465	0.84 (0.69,1.02)	0.88 (0.72,1.07)	0.92 (0.74,1.15)
>30 years	125/945	0.79 (0.61,1.02)	0.82 (0.63,1.06)	0.85 (0.64,1.13)
P for trend		0.009	0.03	0.14
Per 5-year increase in age at first livebirth		0.91 (0.84,0.98)	0.92 (0.85,0.99)	0.94 (0.86,1.02)
Number of children
0	84/624	1.09 (0.76,1.58)	1.08 (0.75,1.56)	1.02 (0.70,1.48)
1–2	324/2586	1.00	1.00	1.00
3–4	495/3457	1.04 (0.88,1.22)	1.02 (0.87,1.20)	1.04 (0.88,1.25)
≥5	429/1555	1.33 (1.09,1.61)	1.26(1.03,1.53)	1.27(1.04,1.55)
Per 1 increase in number of children		1.06(1.02,1.09)	1.05 (1.01,1.08)	1.05 (1.01,1.09)
Age at menopause
<45 years	131/508	1.69(1.34,2.13)	1.69(1.34,2.13)	1.67(1.32,2.11)
45–49 years	419/2291	1.22(1.06,1.41)	1.24(1.07,1.43)	1.24(1.08,1.44)
50–54 years	600/4361	1.00	1.00	1.00
≥54 years	182/1062	1.11 (0.91,1.34)	1.10(0.90,1.33)	1.06 (0.87,1.29)
Per 1-year increase in age at natural menopause		0.97 (0.96,0.99)	0.97 (0.96,0.99)	0.97 (0.96,0.99)
Duration of reproductive period
<35 years	477/2193	1.29(1.12,1.48)	1.29(1.12,1.49)	1.28 (1.11,1.48)
35–39 years	604/4155	1.00	1.00	1.00
>39 years	251/1874	0.95 (0.81,1.13)	0.96 (0.81,1.13)	0.94 (0.79,1.11)
Per 1-year increase in reproductive period		0.97 (0.96,0.98)	0.97 (0.96,0.98)	0.97 (0.96,0.98)
Use of oral contraceptives
Never	1013/5626	1.00	1.00	1.00
≤5 years	233/2029	0.74 (0.63,0.87)	0.73 (0.62,0.86)	0.74 (0.63,0.87)
>5 years	86/567	0.89 (0.69,1.14)	0.89 (0.69,1.14)	0.87 (0.68,1.13)
Hormone replacement therapy
No	1273/7489	1.00	1.00	1.00
Yes	59/733	0.56 (0.42,0.74)	0.59 (0.44,0.78)	0.61 (0.46,0.80)
Estrogen	27/290	0.60 (0.40,0.91)	0.62 (0.41,0.94)	0.63 (0.42,0.95)
Estrogen plus progesterone	32/443	0.52 (0.36,0.76)	0.56(0.39,0.82)	0.59 (0.40,0.86)

Abbreviations: CI, confidence interval; OR, odds ratio.

^aModel 1: adjusted for age at MMSE measurement, year of baseline interview, dialect group, marital status, and education level.

^bModel 2: model 1 plus smoking status, tea intake, coffee intake, sleep duration, physical activity, body mass index, total energy intake, alternate Mediterranean dietary pattern score.

^cModel 3: model 2 plus baseline history of hypertension, diabetes, cardiovascular disease, and cancer. Age at menarche, number of children, use of oral contraceptives, age at menopause and use of hormone replacement therapy were mutually adjusted in the final model as well. For analysis on duration of reproductive period, we further adjusted for number of children, use of oral contraceptives and use of hormone replacement therapy.

^dThe following participants were excluded from the analysis: 575 nulliparous women, 49 without livebirths and 7 women without information on age at firstbirth.

Compared to those who had never used oral contraceptives, women with short-term use (≤5 years) of oral contraceptives had 26% lower odds of having cognitive impairment (OR: 0.74; 95% CI: 0.63, 0.87), while the association was not statistically significant for those used for more than 5 years (OR: 0.87; 95% CI: 0.68, 1.13). Women who used HRT had a 39% lower odd of getting cognitive impairment compared to non-users (OR: 0.61; 95% CI: 0.46, 0.80), and similar association was found for use of estrogen only or estrogen plus progesterone (Table 2).

We did not find a significant association between type of menopause and risk of cognitive impairment (surgical vs. natural, OR: 0.89; 95% CI: 0.72, 1.10; Table A-2).

The results were generally robust in various sensitivity analyses after excluding those who had cancer or CVD at baseline (Table A-3), or using the 23/24 as cut-off point for definition of cognitive impairment (Table A-4), or further adjusting for depressive symptoms in the final models or excluding those with depression at follow-up 3 visits (Table A-5).

Age at menarche was asked as a categorical variable at baseline (n=8222) while a continuous variable at follow-up 1 visits, 62.1% of participants provided consistent data with a Kappa value of 0.59 (Table A-6). Age at menopause was asked as a categorical variable at baseline (n=4924) while a continuous variable at follow-up 1 and 2 visits, 65.4% of participants in follow-up 1 (Table A-7) and 59.3% of participants in follow-up 2 visits (Table A-8) provided consistent data with baseline interview with Kappa values of 0.52 and 0.46, respectively. Reproducibility of age at menopause was also determined in 7124 women who reported exact age at natural menopause at follow-up 1 and 2 visit. A total of 73.9% women reported age at menopause with a maximum deviation of ±2 years, and 27.8% provided the same age at both interviews. We also compared the age at menopause follow-up 1 and 2 visits as a categorical variable, 58.3% of participants provided consistent data with a Kappa value of 0.54 (Table A-9).

The results remained unchanged if we excluding participants with inconsistent data on age at menarche and age at menopause (Table A-10), or used the baseline data of age at menarche and age at menopause as exposures in the analyses (Table A-11).

Compared with women who were excluded from our analyses, those who were included in the analyses were generally younger, more likely to be married, to have a higher education level, to be more active, and to have lower prevalence of baseline comorbidities (Table A-12).

Discussion

Principal Findings

In this large population-based cohort study of women with natural menopause, we found that shorter reproductive period was associated with higher risk of cognitive impairment at later life, while exogenous estrogen use (oral contraceptives or HRT) was associated with lower risk. Among the reproductive factors, early age at menopause and higher parity were was associated with increased risk of cognitive impairment.

In relation to other studies

Our findings are generally consistent with previous studies that have examined the relations of menopause age and reproductive period on cognitive function.^{16, 25, 38, 39} A recent meta-analysis of 13 observational studies reported that later age at menopause and longer reproductive period was generally associated with better cognitive performance or lower cognitive decline.¹⁶ Some studies were conducted among naturally postmenopausal women,^{25, 38, 39} while some studies did not mention the type of menopause.^{7, 14, 40} To our knowledge, only two cross-sectional studies have examined the relation of reproductive period with cognitive function among Chinese women, and both reported that longer reproductive period was significantly associated with better cognitive function.^{5, 6} A longer duration of reproductive period indicates a higher lifetime exposure to endogenous estrogen, thus our study supports the hypothesis for a protective function in the neurocognitive effects of estrogen.

Several studies have examined the relations of age at menarche and parity with cognitive function, but the results were inconclusive.^{5–7, 12, 14, 24, 41} A French cohort study in 996 women found that later age at menarche was associated with poor cognitive performance, while no significant association was found for the number of livebirths.⁷ A cross-sectional study of 6604 women found that later age at menarche and having had five or more children were associated with increased risk of cognitive impairment.¹⁴ Two cross-sectional studies in China also found that higher parity was associated with worse cognitive function.^{5, 6} Hence, our findings were generally consistent with these studies. However, some studies did not find significant relations of age at menarche and number of children with cognition function.^{12, 24, 41} Age at menarche marks the establishment of regulatory ovulatory cycles,⁴² and an earlier age at menarche is significantly associated with higher estradiol levels in adolescents and young adults.^{33, 44} Previous studies have found that parous women have shorter menstrual cycles and lower level of estradiol than nulliparous women, and thus higher parity could lead to an overall lower levels of lifetime estrogen exposure.^45,46 The association between age at first livebirth and cognitive function was not widely studied and existing results were inconsistent.^{5, 7} A cohort study conducted in French women reported that women who had their first child at younger ages (≤20 years old) had poorer cognitive performance compared to women who at their first child at older ages,⁷ while a Chinese study found that older age of first pregnancy was associated with worse cognitive function,⁵ but our study did not find significant association.

The association between use of oral contraceptive and cognitive function has been controversial in previous studies. A systematic review of 22 studies reported an association between the use of oral contraceptive in premenopausal women and improved verbal memory.⁴⁷ Some cross-sectional studies, including a study in Chinese women,⁶ found that oral contraceptive use was associated with better cognitive function; while some studies reported marginally increased risk among long-term users (>10 years)⁴⁸ or shorter-term users (<5 years);⁴⁹ while some studies found no significant association.^{7, 39, 40} We found significant association between short-term use (≤5 years) of oral contraceptive and decreased risk of cognitive impairment, but null association among long-term users (>5 years). The null association among long-term users may be explained by insufficient statistical power due to small number of women in this group in our cohort, but still the results should be interpreted cautiously.

A recent systematic review reported that HRT was associated with better cognitive function in observational studies, but mixed findings were reported in randomized clinical trials.⁵⁰ Some trials conducted in older postmenopausal women (>65 years) have shown increased risk of dementia and cognitive outcomes for HRT users,^{15, 51,52} while some trials conducted in young postmenopausal women (mean age of 53 years) reported null effect.^{53, 54} In observational studies, HRT users were more likely to be better educated and were generally healthier,⁵⁵ as also shown in our study, and thus the healthy-user bias may be an explanation for the inconsistent results between observational studies and clinical trials. Another explanation could be due to differences in the age at which hormone initiation occurred.⁵⁶ While HRT use is most often initiated during or shortly after the menopausal transition in women enrolled in observational studies, the clinical trials generally started in older postmenopausal women of age 65 years or older. Thus, it is possible that estrogen therapy could be more effective only if used within a window of opportunity occurring shortly after menopause.⁵⁰ Due to the potential systematic bias (particularly the healthy-user bias) and residual confounding in the observational studies, further studies, particularly randomized controlled trials, are needed to determine whether HRTs have protective effects on women’s neurocognitive outcomes, and the timing of treatment that gives greatest neuroprotective benefits to postmenopausal women.

Strengths and Limitations

The main strength of our study was the availability of detailed information on a wide range of reproductive and hormonal factors which were collected multiple times and many years before the assessment of cognitive function. In addition, the collection of information on participants’ demographic, dietary, lifestyle factors and medical history allowed for the adjustment for a wide range of potential confounders in the models. Nevertheless, several limitations should be noted. First, misclassification may exist due to the use of self-reported data. However, although reproductive history typically includes events that occurred many years ago, previous studies have suggested that recalled data on age at menarche, parity, age at menopause and use of HRT could still be reliable over many years,^{57, 58} thus self-reported reproductive history was widely used in epidemiological studies. Second, cognitive function was measured by SM-MMSE only once during the follow-up 3 visit, thus we were unable to evaluate the decline of cognitive function over time, and we did not exclude people with cognitive impairment at recruitment. It is reported that women with dementia or cognitive impairment were less likely to report HRT thus reverse causation was possible.⁵⁹ However, we deemed that the impact could be minimal given that majority of the participants had good cognitive function at enrollment since they could answer comprehensive questionnaires (including 165-item food frequency questionnaire), thus it was unlikely for those with considerable impaired cognitive function to have been included in this cohort. Also, SM-MMSE questionnaire is only a global cognitive screening test but not a formal neuropsychological assessment, and cognitive impairment defined by SM-MMSE is not a clinical diagnosis of dementia. Third, despite extensive adjustment of covariates, unmeasured confounding factors that contributed to late menarche or early menopause may be still possible to explain the associations. For example, puberty is delayed in circumstances of low nutrient availability or severe stress, and late menopause is also associated with overall better health parameters.^{60, 61} As for HRT use, a healthy-user bias was possible that women who were HRT users were in general healthier than those who did not use HRT. Finally, our study population included Chinese women with relatively low education level, and the generalizability of our results to other populations remains to be verified.

Conclusions

Hormonal and reproductive factors (age at menopause, reproductive period, parity, the use of oral contraceptives and HRT) were significantly associated with risk of cognitive impairment later in life among Singapore Chinese women. As the population ages, understanding how these factors affect late-life cognitive function in women may help health professionals develop preventive measures targeting lifetime estrogen exposure from endogenous or exogenous sources.

AJOG at a Glance.

1. Why was the study conducted? Previous studies have investigated the association between the length of reproductive periods and use of hormone replacement therapy (HRT), and risk of cognitive function and dementia, but the findings are inconclusive. Most previous researches have been conducted in Western population, and relevant evidence among Asian populations is lacking.

2. What are the key findings? We found short reproductive years was associated with an increased risk of cognitive impairment, while use of oral contraceptives and HRT was associated with a lower risk of cognitive impairment. Specifically, early age at menopause and higher parity was associated with a higher risk of cognitive impairment.

3. What does this study add to what is already known? Our findings indicate that shorter reproductive period is associated with a higher risk of cognitive impairment among Singapore Chinese women, but the lower risk related to external hormone use (HRT and oral contraceptives) requires further validation.

Abbreviations:

BMI

Body mass index

CI

confidence interval

CVD

cardiovascular disease

HRT

hormone replacement therapy

OR

odd ratio

SD

standard deviation

SM-MMSE

the Singapore Modified Mini-Mental State Examination