Connections between reproductive health and cognitive aging among women enrolled in the HCHS/SOL and SOL-INCA

Abstract

INTRODUCTION:

Reproductive health history may contribute to cognitive aging and risk for Alzheimer’s disease, but this is understudied among Hispanic/Latina women.

METHODS:

Participants included 2,126 Hispanic/Latina post-menopausal women (44–75 years) from the Study of Latinos-Investigation of Neurocognitive Aging. Survey linear regressions separately modeled the associations between reproductive health measures (age at menarche, history of oral contraceptive use, number of pregnancies, number of live births, age at menopause, female hormone use at Visit 1, and reproductive span) with cognitive outcomes at Visit 2 (performance, 7-year change, and mild cognitive impairment (MCI) prevalence).

RESULTS:

Younger age at menarche, oral contraceptive use, lower pregnancies, lower live births, and older age at menopause were associated with better cognitive performance. Older age at menarche was protective against cognitive change. Hormone use was linked to lower MCI prevalence.

DISCUSSION:

Several aspects of reproductive health appear to impact cognitive aging among Hispanic/Latina women.

1. BACKGROUND

At 45 years of age, estimated lifetime risk for Alzheimer’s disease among men is 10% but is 20% among women.¹ Aging is the strongest risk factor for Alzheimer’s disease² and the longer average lifespan in women versus men may partially explain their higher lifetime disease risk; however, evidence also suggests that sex-specific biological factors may contribute to Alzheimer’s risk.^3–5

Estrogen exerts a wide range of neuroprotective effects, and greater estrogen exposure is associated with decreased risk for Alzheimer’s disease.⁶ Various reproductive health factors, such as birth control use and number of pregnancies may modify estrogen exposure but are seldom studied in the context of cognitive aging and Alzheimer’s risk. For example, exposure to exogenous hormones (e.g., estrogen, progesterone) used in some forms of birth control or hormonal therapies may influence brain health and late-life cognitive performance, though findings are mixed.⁷ Additionally, number of pregnancies and live births may be connected to cognitive outcomes through changes to estrogen levels and also improved immunoregulation during pregnancy.^6,7 More commonly, age at menarche and menopause are studied in relation to cognitive aging and Alzheimer’s risk. For example, a study of women enrolled with Kaiser Permanente linked higher risk of Alzheimer’s disease and related dementias with several reproductive health factors, including late age at menarche, shorter reproductive span (i.e., years between menarche and menopause), earlier menopause age, and history of hysterectomy.⁸ Such relationships did not differ by racial or ethnic group, but these sample sizes were relatively small. Hispanic/Latina women made up less than 5% of the sample, and this subgroup may be unique in that they were all insured unlike a significant proportion of the Hispanic/Latino/a community.^9,10 In general, Latinos/as are at increased risk for Alzheimer’s disease compared to non-Hispanic/Latino/a Whites,² and Hispanic/Latina women have slightly higher incidence of Alzheimer’s disease compared to their male counterparts in late old age.¹¹ It is unclear if the latter is related to more severe declines in estrogen in old age or other sex-specific reproductive experiences among Hispanic/Latina women relative to their male counterparts. Therefore, it is important to examine the role of reproductive health on cognitive aging among Hispanic/Latina women.

Mexican heritage women reach menopause earlier than their non-Hispanic/Latina White counterparts on average.¹² Despite having less access to healthcare, they often times have similar, if not slightly better, pregnancy outcomes (except gestational diabetes prevalence) compared to non-Hispanic/Latina White individuals.^13,14 These factors combined with longer lifespan among Hispanic/Latina women relative to non-Hispanic/Latina White women¹⁵ warrant investigations into the connections between reproductive health and cognitive aging and impairment among diverse Hispanic/Latina women. Notably, Hispanic/Latina women residing in the United States represent various heritage groups (e.g., Central American) and immigration histories which may influence access to resources (e.g., educational, health) and reproductive health and pregnancy outcomes.^16,17 Therefore, we examined how reproductive health factors relate to cognition, cognitive change, and mild cognitive impairment (MCI) among diverse post-menopausal Hispanic/Latina women using data from the multi-site Hispanic Community Healthy Study (HCHS/SOL) and Study of Latinos-Investigation of Neurocognitive Aging (SOL-INCA). We hypothesized that later menarche, earlier menopause, and shorter reproductive span would be associated with poorer cognitive performance, greater adverse change in cognition across seven years, and higher prevalence of MCI. We also explored relationships between oral contraceptive use, number of pregnancies, number of live births, and female hormone use with these same cognitive outcomes, with no a-priori hypotheses.

2. METHODS

2.1. Data.

The Hispanic Community Health Study/Study of Latinos (HCHS/SOL) is a multisite prospective cohort study of N=16,415 self-identified diverse Hispanic/Latino adults (ages 18–74 at recruitment). Participants were from four major US metropolitan areas: Bronx, NY; Chicago, IL; Miami, FL; and San Diego, CA. An ancillary study of HCHS/SOL, the Study of Latinos-Investigation of Neurocognitive Aging (SOL-INCA), was conducted during the second HCHS/SOL visit to examine neurocognition among HCHS/SOL participants who were 45 years and older at the time of their initial neurocognitive testing during Visit 1.¹⁸ At Visit 2 of HCHS/SOL, N = 6,377 out of N=7,420 eligible individuals completed the SOL-INCA visit (henceforth Visit 2), approximately 7 years later. The HCHS/SOL Coordinating Center generated complex study designs and sampling procedures to obtain data of diverse Hispanic/Latino adults 50-years and older. The detailed HCHS/SOL study designs and sampling methods have been published.^19,20 Our procedures, from the data collection (e.g., population-based sampling, representation from multiple heritage groups (Central Americans, Cubans, Dominicans, Mexicans, Puerto Ricans, South Americans), testing in preferred language) to the data analysis (applying sampling weights, including sociodemographic covariates) and interpretation consider several aspects of diversity, equity, and inclusion in order to address health disparities in the Hispanic/Latino population. All participants gave informed consent, and Institutional Review Board approval was obtained at all study sites. Research complied with the Helsinki Declaration and its later amendments.

2.2. Outcomes

2.2.1. 7-year (average) performance and change in cognition.

Neurocognitive measures of Brief-Spanish English Verbal Learning Test Sum of Trials and Delayed Recall (B-SEVLT- Sum and - Recall; verbal episodic learning and memory); Word Fluency (WF; verbal fluency with letters F and A); and Digit Symbol Substitution (DSS; processing speed and executive functioning) were tested at Visit 1 and Visit 2, an average of 7 years later. At Visit 2, two measures on processing speed and executive functioning (Trails-A and -B) were additionally tested. All tests were z-scored [(X-Mean)/SD] to facilitate interpretation of results across a common metric. We calculated a global cognitive composite score by averaging the z-scores of the repeated tests (from Visit 1 and Visit 2, excluding Trails-A and -B). To examine cognitive change, we calculated a change index for the repeated cognitive measures and the global measure of cognition.²¹ We used survey linear regressions where cognitive score at Visit 2 was modeled as a function of Visit 1 cognitive score, adjusting for time between cognitive assessments. Test specific standardized measures of change and global cognitive change were calculated using (T2 − T2_pred)/RMSE, where T2 is the respondent’s cognitive score at Visit 2, T2_pred is the predicted score, and RMSE is root mean squared error. Detailed rationales on this technique can be found elsewhere.²²

2.2.2. Mild cognitive impairment (MCI) prevalence

was operationalized using NIAA criteria as defined by the SOL-INCA study and previously published.²³

2.3. Exposures.

Exposures included multiple reproductive health factors: self-reported age at menarche, history of oral contraceptive use (No, Yes), number of pregnancies, number of live births, age at menopause (we also considered a categorical operationalization ≤45 years vs. >45 years), current female hormone use (No, Yes), and reproductive span (years with menses, calculated by subtracting age at menarche from age at menopause). All exposure variables were measured at Visit 1.

2.4. Covariates.

All covariates were measured at Visit 1 and included age (continuous), education (less than high school degree/GED, high school or equivalent, greater than high school degree/GED), language of interview (Spanish, English), Hispanic/Latino heritage (Dominican, Central American, Cuban, Mexican, Puerto-Rican, South American), field center (Bronx, Chicago, Miami, San Diego), income (<10,000, 10,001–20,000, 20,001– 40,000, 40,001– 75,000, >75,000, or not reported), marital status (married/ cohabitating, single, separated/divorced/widowed), insurance status (uninsured, insured), nativity (born in U.S. 50 states/DC, born in U.S. territory or foreign country), and body mass index.

2.5. Statistical Analyses

2.5.1. Analytic Subpopulation.

Out of the unweighted n=6,377 individuals that completed Visit 2, n=4,110 were women. We focused on n=2,197 women who reached their menopause (either induced: hysterectomy with removal of both ovaries before the natural menopause or natural: no history of hysterectomy or hysterectomy after menopause) at 60 years or younger. More specifically, in the primary analysis, we only included women with a hysterectomy if 1) they underwent hysterectomy post-menopause or 2) they reported a bilateral oophorectomy. In the former instance, the women were placed in the natural menopause group. We excluded n=71 women (less than 5% missing) who had any missing covariates for an analytic sample of unweighted n=2,126. For analyses that examined MCI as an outcome, the unweighted sample included n=2,092 women after also excluding individuals with suspected severe cognitive impairment (n=34). See Supplemental Figure 1.

2.5.2. Analytic Approach.

First, we reported descriptive statistics for the overall analytic sample as well as by menopause type (induced, natural). Distributional differences by type of menopause were tested using survey adjusted chi-squared tests for categorical variables and t-tests for continuous variables. The survey weighted estimates are presented in Table 1. All analyses accounted for the complex survey design and survey weights.

Table 1.

Target population characteristics overall and by type of menopause in the Study of Latinos-Investigation of Neurocognitive Aging

	Induced menopause n=282	Natural menopause n=1,844	Overall n=2,126	P-value
Covariates
Age in years (mean, [SD])	58.6 (8.1)	59.6 (7.8)	59.5 (7.8)	0.161
Education (%, [SE])
< High School	33.3 (3.7)	41.2 (1.8)	40.1 (1.7)	0.063
High School	17.3 (2.8)	19.3 (1.4)	19.0 (1.3)
> High School	49.3 (4.3)	39.6 (1.7)	40.9 (1.6)
Language (%, [SE])
Spanish	92.8 (2.6)	89.5 (1.0)	90.0 (0.9)	0.311
English	7.2 (2.6)	10.5 (1.0)	10.0 (0.9)
Center (%, [SE])
Bronx	21.1 (3.3)	25.7 (2.0)	25.1 (1.9)	0.368
Chicago	8.8 (1.7)	10.8 (0.9)	10.6 (0.8)
Miami	45.8 (4.6)	42.5 (2.8)	43.0 (2.6)
San Diego	24.3 (3.8)	20.9 (1.8)	21.4 (1.7)
Income (%, [SE])
<$10,000	17.1 (2.9)	20.3 (1.5)	19.8 (1.3)	0.647
$10,001-$20,000	31.0 (3.7)	31.0 (1.5)	31.0 (1.4)
$20,001-$40,000	30.7 (4.0)	24.9 (1.5)	25.7 (1.4)
$40,001-$75,000	8.6 (2.1)	8.6 (0.9)	8.6 (0.8)
>$75,000	1.2 (0.7)	1.9 (0.4)	1.8 (0.3)
Not Reported	11.4 (2.6)	13.3 (1.2)	13.0 (1.1)
Marital status (%, [SE])
Married/cohabitating	47.1 (4.2)	44.5 (2.0)	44.9 (1.8)	0.819
Single	15.5 (2.6)	16.1 (1.2)	16.0 (1.1)
Separated/Divorced/Widowed	37.4 (3.9)	39.4 (1.8)	39.1 (1.7)
Insurance status (%, [SE])
Uninsured	40.0 (4.2)	41.6 (1.8)	41.4 (1.7)	0.708
Insured/Medicaid	60.0 (4.2)	58.4 (1.8)	58.6 (1.7)
Nativity (%, [SE])
Not born in the US	83.0 (2.8)	81.6 (1.4)	81.8 (1.2)	0.655
Born in the US	17.0 (2.8)	18.4 (1.4)	18.2 (1.2)
BMI (mean, [SD])	29.8 (6.0)	30.4 (6.1)	30.3 (6.1)	0.181
Exposures
Female hormone use (%, [SE])
No	90.3 (2.3)	97.2 (0.6)	96.2 (0.6)	<0.001
Yes	9.7 (2.3)	2.8 (0.6)	3.8 (0.6)
History of oral contraceptive use (%, [SE])
No	48.5 (4.3)	44.2 (1.9)	44.8 (1.7)	0.380
Yes	51.5 (4.3)	55.8 (1.9)	55.2 (1.7)
Age at menarche (mean, [SD])	12.7 (1.9)	12.8 (1.9)	12.7 (1.9)	0.520
Number of live births (mean, [SD])	2.6 (1.8)	2.9 (2.0)	2.9 (2.0)	0.031
Number of pregnancies (mean, [SD])	3.4 (2.3)	3.9 (2.5)	3.8 (2.5)	0.012
Age at menopause (mean, [SD])	42.1 (7.7)	48.9 (4.7)	47.9 (5.8)	<0.001
Menopause age (>45 years; %, [SE])
≤45 years	67.4 (3.7)	25.5 (1.5)	31.3 (1.5)	<0.001
>45 years	32.6 (3.7)	74.5 (1.5)	68.7 (1.5)
Reproductive span (mean, [SD])	29.6 (7.8)	36.1 (4.9)	35.2 (5.9)	<0.001

Note. Sample size is unweighted; all other reported values are weighted to represent the target population.

Abbreviations: BMI = body mass index; SD = standard deviation; SE = standard error

Second, we examined the hypothesized associations between the reproductive health exposures and our cognitive outcomes including: cognitive performance at Visit 2 (on average 7-years from the Visit 1), cognitive change, and MCI. We fit a series of survey linear and logistic regression models sequentially adjusting for covariables: (1) crude, (2) age and education adjusted, (3) full covariates adjustment (see covariables above section 2.4). The estimated beta coefficients and their standard errors (for cognitive performance at Visit 2 and cognitive change) or odds ratios and their 95% confidence intervals (for MCI) are presented in Table 2 and Table 3, respectively. In post-hoc analyses, we calculated average marginal means and probabilities for significant associations and plotted these with their 95% confidence intervals to facilitate interpretation (Figures 1–3).

Table 2.

Associations of each reproductive history factor with Visit 2 (7-years post-baseline on average) cognition and 7-year cognitive change (n=2,126) in the Study of Latinos-Investigation of Neurocognitive Aging

7-year cognition
	B-SEVLT Sum			B-SEVLT Recall			WF			DSS
	M1	M2	M3	M1	M2	M3	M1	M2	M3	M1	M2	M3
	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)
Female hormone use
No	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref
Yes	0.238** (0.083)	0.065 (0.083)	0.032 (0.091)	0.244* (0.120)	0.104 (0.105)	0.086 (0.116)	0.229* (0.114)	0.098 (0.124)	−0.020 (0.111)	0.298* (0.133)	0.067 (0.127)	0.006 (0.104)
Contraceptive use
No	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref
Yes	0.302*** (0.065)	0.189** (0.060)	0.175** (0.059)	0.214*** (0.062)	0.120* (0.061)	0.092 (0.063)	0.237*** (0.071)	0.172** (0.063)	0.090 (0.065)	0.384*** (0.062)	0.239*** (0.052)	0.145** (0.049)
Age at menarche	0.002 (0.020)	0.030 (0.019)	0.018 (0.018)	0.001 (0.019)	0.022 (0.018)	0.011 (0.018)	−0.019 (0.017)	0.008 (0.016)	−0.004 (0.016)	−0.090*** (0.016)	−0.050*** (0.014)	−0.042*** (0.012)
Number of live births	−0.041* (0.018)	0.019 (0.017)	0.015 (0.016)	−0.026 (0.018)	0.018 (0.018)	0.006 (0.017)	−0.066*** (0.019)	0.004 (0.019)	0.008 (0.020)	−0.139*** (0.014)	−0.054*** (0.014)	−0.044** (0.014)
Number of pregnancies	−0.017 (0.014)	0.014 (0.014)	0.015 (0.013)	−0.005 (0.015)	0.019 (0.015)	0.016 (0.014)	−0.042** (0.014)	−0.005 (0.013)	0.005 (0.013)	−0.080*** (0.014)	−0.034** (0.012)	−0.023* (0.011)
Age at menopause	0.005 (0.006)	0.012* (0.006)	0.009 (0.005)	0.008 (0.007)	0.015* (0.006)	0.013* (0.006)	0.008 (0.007)	0.004 (0.006)	0.000 (0.006)	−0.001 (0.006)	0.004 (0.005)	−0.000 (0.004)
Menopause age (>45 years)
≤45 years	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref
>45 years	0.135 (0.075)	0.133 (0.072)	0.112 (0.066)	0.125 (0.078)	0.139 (0.078)	0.129 (0.071)	0.065 (0.081)	−0.025 (0.078)	−0.063 (0.074)	0.093 (0.067)	0.057 (0.059)	0.024 (0.054)
Reproductive span	0.005 (0.006)	0.008 (0.006)	0.007 (0.005)	0.007 (0.007)	0.012 (0.007)	0.011 (0.006)	0.009 (0.007)	0.002 (0.006)	0.000 (0.006)	0.007 (0.006)	0.008 (0.004)	0.003 (0.004)

	Trails A			Trails B			Global cognition
	M1	M2	M3	M1	M2	M3	M1	M2	M3
	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)
Female hormone use
No	ref	ref	ref	ref	ref	ref	ref	ref	ref
Yes	−0.206 (0.125)	−0.023 (0.129)	−0.017 (0.118)	−0.099 (0.138)	0.060 (0.139)	0.106 (0.125)	0.253*** (0.073)	0.084 (0.065)	0.026 (0.062)
Contraceptive use
No	ref	ref	ref	ref	ref	ref	ref	ref	ref
Yes	−0.362*** (0.085)	−0.237** (0.075)	−0.125 (0.070)	−0.261*** (0.065)	−0.169** (0.060)	−0.082 (0.058)	0.284*** (0.049)	0.179*** (0.040)	0.124** (0.040)
Age at menarche	0.115*** (0.024)	0.083*** (0.021)	0.068*** (0.020)	0.055*** (0.015)	0.033* (0.014)	0.028 (0.014)	−0.026 (0.013)	0.003 (0.012)	−0.003 (0.011)
Number of live births	0.091*** (0.020)	0.015 (0.021)	0.009 (0.021)	0.112*** (0.015)	0.041** (0.015)	0.029 (0.015)	−0.068*** (0.013)	−0.003 (0.012)	−0.004 (0.012)
Number of pregnancies	0.040* (0.017)	−0.001 (0.017)	−0.006 (0.016)	0.068*** (0.013)	0.029* (0.011)	0.015 (0.011)	−0.036*** (0.011)	−0.001 (0.010)	0.003 (0.009)
Age at menopause	0.002 (0.008)	−0.001 (0.008)	−0.000 (0.006)	0.002 (0.006)	−0.001 (0.005)	0.001 (0.005)	0.005 (0.005)	0.008* (0.004)	0.005 (0.004)
Menopause age (>45 years)
≤45 years	ref	ref	ref	ref	ref	ref	ref	ref	ref
>45 years	−0.072 (0.082)	−0.036 (0.082)	−0.022 (0.076)	−0.064 (0.069)	−0.024 (0.065)	0.015 (0.061)	0.103 (0.053)	0.073 (0.050)	0.049 (0.043)
Years with menses	−0.010 (0.008)	−0.010 (0.008)	−0.007 (0.006)	−0.005 (0.005)	−0.006 (0.005)	−0.002 (0.005)	0.007 (0.005)	0.007 (0.004)	0.005 (0.003)

	Δ B-SEVLT Sum			Δ B-SEVLT Recall			Δ WF			Δ DSS				Δ Global cognition
	M1	M2	M3	M1	M2	M3	M1	M2	M3	M1	M2	M3		Model 1	Model 2	Model 3
	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)		B (SE)	B (SE)	B (SE)
Female hormone use													Female hormone use
No	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref	No	ref	ref	ref
Yes	0.144 (0.110)	0.035 (0.099)	0.080 (0.100)	0.217 (0.166)	0.134 (0.157)	0.175 (0.161)	0.075 (0.140)	0.019 (0.138)	0.017 (0.138)	0.270* (0.112)	0.188 (0.105)	0.163 (0.105)	Yes	0.204 (0.158)	0.110 (0.144)	0.167 (0.141)
Contraceptive use													Contraceptive use
No	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref	No	ref	ref	ref
Yes	0.167* (0.072)	0.084 (0.069)	0.107 (0.065)	0.054 (0.062)	−0.011 (0.062)	−0.009 (0.064)	0.091 (0.068)	0.057 (0.068)	0.053 (0.064)	0.090 (0.060)	0.025 (0.060)	0.049 (0.060)	Yes	0.073 (0.064)	−0.003 (0.062)	0.027 (0.061)
Age at menarche	0.021 (0.021)	0.039 (0.020)	0.040* (0.019)	0.021 (0.019)	0.034 (0.019)	0.032 (0.019)	−0.012 (0.016)	−0.000 (0.016)	0.006 (0.017)	−0.001 (0.015)	0.012 (0.015)	0.009 (0.015)	Age at menarche	0.012 (0.018)	0.029 (0.017)	0.036* (0.017)
Number of live births	−0.008 (0.017)	0.029 (0.017)	0.033 (0.017)	−0.011 (0.020)	0.012 (0.020)	0.008 (0.021)	−0.027 (0.017)	0.005 (0.018)	0.022 (0.017)	−0.042* (0.016)	−0.022 (0.017)	−0.021 (0.018)	Number of live births	−0.028 (0.020)	0.004 (0.019)	0.014 (0.019)
Number of pregnancies	0.009 (0.015)	0.028 (0.015)	0.026 (0.014)	0.003 (0.017)	0.016 (0.017)	0.013 (0.016)	−0.013 (0.012)	0.004 (0.012)	0.012 (0.012)	−0.017 (0.014)	−0.005 (0.014)	−0.003 (0.014)	Number of pregnancies	−0.002 (0.016)	0.016 (0.015)	0.017 (0.015)
Age at menopause	0.000 (0.006)	0.007 (0.006)	0.004 (0.006)	0.005 (0.006)	0.011 (0.006)	0.008 (0.006)	0.009 (0.006)	0.009 (0.006)	0.007 (0.006)	0.003 (0.006)	0.009 (0.005)	0.010 (0.005)	Age at menopause	0.004 (0.006)	0.009 (0.005)	0.007 (0.005)
Menopause age (>45 years)													Menopause age (>45 years)
≤45 years	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref	≤45 years	ref	ref	ref
>45 years	0.117 (0.080)	0.133 (0.081)	0.109 (0.077)	0.095 (0.073)	0.117 (0.075)	0.100 (0.069)	0.050 (0.073)	0.015 (0.074)	−0.015 (0.072)	0.069 (0.064)	0.091 (0.063)	0.098 (0.064)	>45 years	0.098 (0.071)	0.109 (0.073)	0.082 (0.069)
Reproductive span	−0.001 (0.006)	0.003 (0.006)	0.001 (0.005)	0.002 (0.006)	0.007 (0.006)	0.004 (0.005)	0.009 (0.005)	0.007 (0.006)	0.004 (0.006)	0.002 (0.005)	0.007 (0.005)	0.008 (0.005)	Reproductive span	0.002 (0.005)	0.005 (0.005)	0.002 (0.005)

Notes. Sample size is unweighted; all other reported values are weighted to represent the target population.

Each reproductive history factor was entered separately into model with each cognitive outcome.

Model 1 was crude, Model 2 was adjusted for age and education, Model 3 was adjusted for Model 2 + language, Hispanic/Latino heritage, field center, income, marital status, insurance status, nativity, and body mass index.

In cognitive change models, time in years between cognitive function assessments were included in the calculation of the cognitive function outcome.

Abbreviations: Δ=change, B = beta; B-SEVLT = Brief-Spanish English Verbal Learning Test; DSS=Digit Symbol Substitution; M# = model; SE = standard error; WF=Word Fluency

^*p <0.05,

^**p <0.01,

^***p <0.001

Table 3.

Associations between each reproductive history factor with MCI (n=2,092) in the Study of Latinos-Investigation of Neurocognitive Aging

	Model 1	Model 2	Model 3
	OR [95% CI]	OR [95% CI]	OR [95% CI]
Female hormone use
No	ref	ref	ref
Yes	0.29* [0.11;0.76]	0.34* [0.13;0.91]	0.35* [0.13;0.94]
Contraceptive use
No	ref	ref	ref
Yes	0.77 [0.52;1.12]	0.85 [0.57;1.25]	0.81 [0.53;1.23]
Age at menarche	1.01 [0.91;1.11]	0.98 [0.89;1.08]	0.99 [0.90;1.09]
Number of live births	1.08 [0.99;1.19]	1.02 [0.92;1.12]	1.00 [0.91;1.11]
Number of pregnancies	1.04 [0.97;1.12]	1.01 [0.94;1.08]	0.99 [0.92;1.07]
Age at menopause	1.00 [0.96;1.03]	0.99 [0.96;1.03]	1.00 [0.97;1.03]
Menopause age (>45 years)
≤45 years	ref	ref	ref
>45 years	1.19 [0.81;1.75]	1.24 [0.85;1.83]	1.29 [0.87;1.91]
Reproductive span	0.99 [0.96;1.03]	0.99 [0.96;1.03]	1.00 [0.96;1.03]

Notes. Sample size is unweighted; all other reported values are weighted to represent the target population.

Each reproductive history factor was entered separately into model. Model 1 was crude, Model 2 was adjusted for age and education, Model 3 was adjusted for Model 2 + language, Hispanic/Latino heritage, field center, income, marital status, insurance status, nativity, and body mass index

Abbreviations: CI=confidence interval; M# = Model; MCI=mild cognitive impairment; OR=odds ratio;

^*p <0.05,

^**p <0.01,

^***p <0.001

Figure 1. Associations between reproductive history factors with cognitive performance (z-scored) at Visit 2 in the Study of Latinos-Investigation of Neurocognitive Aging.

Model 1 is crude; Model 2 is adjusted for age and education; Model 3 is additionally adjusted for language preference, Hispanic/Latino heritage, field center, income, marital status, insurance status, place of birth, and body mass index.

Abbreviations: B-SEVLT = Brief-Spanish English Verbal Learning Test; DSS = Digit Symbol Substitution

Figure 3. Association between female hormone use at Visit 1 with prevalent mild cognitive impairment at Visit 2.

Model 1 is crude; Model 2 is adjusted for age and education; Model 3 is additionally adjusted for language preference, Hispanic/Latino heritage, field center, income, marital status, insurance status, place of birth, and body mass index.

Abbreviations: MCI = mild cognitive impairment

We conducted three separate sets of supplemental analysis. First, we estimated an additional model that expands the primary analysis to include adjustments for cardiovascular risk as measured by Framingham Cardiovascular Risk score²⁴ and depressive symptoms measured using the Center for Epidemiologic Studies Depression Scale (CES-D-10).²⁵ Results from these models are included in Supplemental Tables 1–2. Second, we examined the associations between age at hysterectomy with our outcomes of interest for the subsample of women who underwent hysterectomy pre- or post-menopause (unweighted n=612; Supplemental Figure 2). We followed the same sequence of model adjustments as described above. The estimated coefficients and their standard errors for our continuous cognitive outcomes (Visit 2 performance and change) are presented in Supplemental Table 3, and odds ratios and their 95% confidence intervals for our categorical outcome (MCI) are presented in Supplemental Table 4. In post-hoc analyses, we calculated average marginal estimates and probabilities for significant associations and plotted these with their 95% confidence intervals to facilitate interpretation (Supplemental Figures 3–4). Third, given potential chronological age differences in the associations between our reproductive health exposures and cognitive outcomes, we tested for modification by age split by the mean (<60 vs. 60+ years) by adding an interaction effect between age and the reproductive health exposures, independently, to fully adjusted models in the primary analysis (model 3 as specified above). See supplemental Tables 5–6. Note, we did not have duration for female hormone use and, therefore, did not test age interactions with this exposure. When the interaction effects were consistently significant, we estimated stratified models whereby we refit model 3 within each age group (<60 vs. 60+ years). See Supplemental Table 7 and Supplemental Figures 5–6.

3. RESULTS

3.1. Descriptive statistics

The descriptive characteristics for the overall target female population, as described above, as well as by the type of menopause (induced vs natural) are shown in Table 1. The average age at Visit 1 was 59.5 years, more than 40% had an education level of high school or above, and 90% conducted their interviews in Spanish. Nearly 20% had income of less than $10,000, 45% were either married or cohabited, roughly 60% had health insurance coverage, and 82% were born outside the U.S. The average body mass index was 30.3. There was no significant difference between induced and natural menopause groups in the distributions of these characteristics.

For the reproductive health measures, 4% reported current female hormone use at Visit 1. In the full sample, 55% indicated a history of oral contraceptive use with 62% of those under 60 years reporting use versus 48% of those 60 years and older. The average age at menarche was 12.7 years, and the number of pregnancies and live births were 3.8 and 2.9, respectively. The average age at menopause was 47.9 years, and 69% underwent menopause after 45 years of age. The average reproductive span was 35.2 years. Individuals with induced menopause were more likely to use female hormone and had lower numbers of pregnancies and live births. The induced menopause group had a lower average age at menopause, higher proportion of women with early menopause (at or before 45-years old), and shorter reproductive span.

In the full sample, 10.8% met criteria for MCI. When split by age, 9.9% of those under 60 years and 11.8% of those 60 years and older meeting criteria.

3.2. Primary analysis

History of oral contraceptive use was associated with better cognitive performance across all outcomes in crude and age- and education-adjusted models. In fully adjusted models, oral contraceptive use maintained associations with global cognitive performance (β_Global=0.124 [SE=0.040], p<0.01), learning (β_B-SEVLT-Sum=0.175 [SE=0.059], p<0.01), and processing speed (β_DSS=0.145 [SE=0.049], p<0.01) at Visit 2. In fully-adjusted models, older age at menarche was associated with worse Visit 2 performance in processing speed (β_DSS =−0.042 [SE=0.012], p<0.001 and β_{Trails A} =0.068 [SE=0.020], p<0.001). Higher number of pregnancies and live births were each associated with slower processing speed on the DSS (pregnancies: β_DSS=−0.023 [SE=0.011], p<0.05 and live births: β_DSS=−0.044 [SE=0.014], p<0.01) in fully-adjusted models. Older age at menopause (continuous) was associated with better memory (β_{B-SEVLT-Recall}=0.013 [SE=0.006], p<0.05) whereas categorical age at menopause was not associated with cognitive outcomes at Visit 2. Reproductive span and female hormone use were not associated with Visit 2 performance on any cognitive outcomes.

In fully-adjusted models, older age at menarche was protective against 7-year average adverse change in global cognition (β_Global=0.036 [SE=0.017], p<0.05) and learning (β_B-SEVLT-Sum=0.040 [SE=0.019], p<0.05). No other reproductive health factor was linked to change in cognitive outcomes. Only female hormone use was associated with lower odds of MCI prevalence (OR=0.35 [95% CI=0.13;0.94], p<0.05). See Tables 1–3.

3.3. Supplemental analysis

Additional adjustments for cardiovascular disease risk and depressive symptoms to the primary analysis did not have any quantitative or qualitative effects on the main results reported above (Supplemental Tables 1–2).

Older age at hysterectomy was associated with better (average) 7-year cognitive performance in executive functioning (β_{Trails B}=−0.011 [SE=0.005], p<0.05). In terms of cognitive change, older age at hysterectomy was linked to adverse change in learning (β_B-SEVLT-Sum=−0.011 [SE=0.005], p<0.05) and memory (β_{B-SEVLT-Recall}=−0.012 [SE=0.006], p<0.05), yet maintenance/improvement in verbal fluency (β_WF=0.015 [SE=0.006], p<0.01). Age at hysterectomy was not associated with MCI prevalence. See Supplemental Tables 3–4 and Supplemental Figures 3–4.

Our tests of age interactions indicated that the associations between binary age at menopause (≤45, >45 years) and reproductive span were significant with respect to global cognitive function, learning, and delayed recall. Specifically, women over the age of 60 years with later menopause (>45 years) had better learning and memory performance compared to their counterparts who underwent menopause earlier in life whereas age at menopause did not result in cognitive differences for women under 60 years of age. Similarly, later age of menopause was also protective against adverse change in learning and global cognition among older women but was unrelated to cognitive change in younger women. Additionally, longer reproductive span was associated with better memory only among women over 60 years of age. See Supplemental Tables 5–7 and Supplemental Figures 5–6.

4. DISCUSSION

In a population-based cohort of 2,126 Hispanic/Latina postmenopausal women, aspects of reproductive health (i.e., oral contraceptive use, number of pregnancies, number of live births, age at menarche, and age at menopause) were associated with cognitive performance and/or change in cognition over an average 7-year period. Additionally, hormone use at Visit 1 was linked to lower prevalence of MCI, suggesting it may be a protective factor against cognitive impairment. Our results inform our understanding of cognitive aging among Hispanic/Latina women, a population with a wide range of educational backgrounds, immigration histories, ancestral backgrounds, and lower rates of access to insurance.

Generally, existing literature suggests that higher levels of estrogen exposure are protective to cognition,²⁶ and we found mixed evidence to support this across reproductive health measures. Reproductive health factors had relatively distinct associations with cognitive outcomes. For example, consistent with existing studies, history of oral contraceptive use was associated with better cognitive outcomes (i.e., learning, processing speed/executive functioning (DSS), and global cognition).^27–30 Null findings have been reported elsewhere and may be due to small sample size, low prevalence of oral contraceptive use, and/or variability in the specific cognitive domains assessed.^6,31–33 Importantly, the prevalence of history of oral contraceptive in this Hispanic/Latina sample (57%) lags behind the U.S. average (82%),³⁴ suggesting many more Hispanic/Latina women may stand to benefit from oral contraception but thorough investigation is needed to confirm. Several factors impact oral contraception use in this population, including but not limited to medical mistrust (particularly in light of the U.S.’s forced sterilizations of Hispanic/Latina women),^35,36 cultural and religious beliefs surrounding conception,³⁷ and limited access to healthcare.^9,10

Higher number of pregnancies and live births were each associated with slower processing speed. Fox and colleagues³⁸ found that number of first trimesters rather than third trimesters in pregnancy was protective against Alzheimer’s disease. This might suggest that the improved immunoregulation that occurs early in pregnancy is driving the protective effect more so than the increase in estrogen levels in later stages of pregnancy which then decreases drastically postpartum with the delivery of the placenta.^39–41 Complicating this further, estrogen levels tend to stay lower postpartum for individuals who breastfeed⁴² nulliparous women may have higher estrogen levels than parous individuals,⁴³ and those with four or more births may have particularly low estrogen levels post-menopause.⁴⁴ Alternatively, births may capture the influence of parenting on cognition.^45,46 Birthing has been associated with declines in frontal and temporal volumes for up to two years among new mothers but not among fathers,⁴⁷ yet at older ages mothers and fathers seem to show reduced brain aging relative to non-parents,⁴⁶ suggesting a long-term benefit of parenting. Although, we did not detect associations between number of pregnancies or births with cognitive change or MCI, number of full-term births has been connected to risk for cognitive impairment/Alzheimer’s disease.⁷ However, the specific number of births that confer risk varies widely with some investigations finding that zero births increases risk,⁴⁸ others stating that zero births decreases risk,³¹ and still others finding that several births increases risk.^29,30,49 This variation may be indicative of cohort differences in the psychosocial changes that accompany pregnancy and raising children.

Older age at menarche was associated with cognitive advantages and disadvantages. Similar to a study of French women,³² we found that older age at menarche was associated with slower processing speed on both the DSS and TMT A, yet it was associated with better 7-year maintenance of learning and global cognition. Differences in outcomes may reflect age at menarche serving as an indicator of various early-life exposures that are independently associated with cognitive outcomes. Certain early-life exposures may lower menarche age (e.g., high childhood body mass⁵⁰ and cardiovascular disease risk,⁵¹ childhood sexual abuse⁵²) whereas others may raise menarche age (e.g., early-life nutritional deprivation, financial hardships). Adverse early life exposures may be more prevalent among the Hispanic/Latino community relative to the non-Latino White population.⁵³

The associations between age at menarche with change in learning suggest that younger age at menarche could be a risk factor for Alzheimer’s disease. Later age at menarche was associated with lower levels of Alzheimer’s disease biomarkers (amyloid beta and tau) among post-menopausal Swedish women.⁵⁴ However, we did detect associations between age at menarche with MCI possibly due to our small proportion of participants with MCI. Existing literature primarily focuses on dementia status, and results range from null findings^6,49,55,56 to older age at menarche being associated with increased risk for dementia.^8,57 Of note, Prince and colleagues⁴⁹ population-based study which included over 7,000 women in Latin America did not find differential risk for dementia by age at menarche whereas Gilsanz and colleagues⁸ conducted a smaller study including 274 U.S. Hispanic/Latina women (<5% of the sample) that found that older age at menarche (16–17 years) was associated with increased risk for dementia. Our largely foreign-born cohort may be more similar to individuals still living in Latin America, or discrepancies with Gilsanz et al.⁸ may be related to differences in study design (i.e., medical record review of individuals enrolled at Kaiser Permanente in Northern California versus multi-state population-based cohort) which may sample groups that differ in socioeconomic background on whole.

Use of female hormones at Visit 1 was the only reproductive health factor associated with MCI, specifically lower prevalence. Notably, only 4% of the overall sample indicated female hormone use at Visit 1. There may be a critical period to capture the benefits of hormone therapy on cognition (e.g., closer to menopause or before 65 years), but findings are not definitive.^26,58–60 While we did not find associations between female hormone use with cognitive performance or cognitive change, female hormone use may nonetheless impact subjective cognitive decline—another component of our MCI measure.²³

Longer reproductive span was associated with better memory performance only among women over 60 years old, suggesting a potential delayed protection in older adulthood. Nonetheless, we did not find evidence of reproductive span-related protection in 7-year cognitive change or MCI for older women. In a recent study, Najar and colleagues⁵⁶ observed that longer reproductive period was associated with increased incidence of Alzheimer’s disease, especially among individuals 75 years and older, but not dementia with cerebrovascular disease. Given this distinction between dementia etiologies, relationships between reproductive span and cognitive status may be more difficult to detect in populations with higher risk for cerebrovascular disease pathology, such as Hispanic/Latino adults.⁶¹ Importantly, in the present study, results remained significant when controlling for a variety of sociodemographic factors, depressive symptoms, and cardiovascular disease risk, suggesting robust influences of reproductive health factors on cognitive aging among Hispanic/Latina women.

Among Hispanic/Latina women who underwent hysterectomy, age at hysterectomy had cognitive domain-specific associations. The majority of existing literature links younger age of hysterectomy with poorer cognitive outcomes,⁶² yet we observed better maintenance of learning and memory performance over time. Despite this, younger age at hysterectomy appeared to be disadvantageous to executive functioning, specifically worse cross-sectional performance on TMTB and greater negative change in phonemic fluency over an average 7-year period. Younger age at hysterectomy has been associated with increased presence of neuritic plaques⁶³ and increased risk for dementia,⁶² but we did not detect associations with MCI. Women who undergo hysterectomies often have medical comorbidities that contribute to the decision to undergo the procedure.⁶⁴ Disentangling the role of these contributing factors on brain and cognitive health from that of the hysterectomy alone should be further examined among Hispanic/Latina women, particularly because of the difficulties accessing healthcare that are unique to this population.⁹

Our findings should be interpreted within the context of limitations. First, we used retrospective, self-report data to ascertain reproductive health information though this has been shown to be reliable.⁶⁵ Second, regarding oral contraceptives, we did not examine other forms of contraception (e.g., injectable contraceptives), specific hormonal components (exogenous estrogen and progestin may provide the most neuroprotection),^66,67 or duration of use. Some have found that use for ≤5 years is associated with lower risk of cognitive impairment³⁰ whereas others have found better cognitive performance with longer use.^27,28 Third, we lacked information on onset of hormone use and previous use. Initiating hormone use more than five years after menopause may be associated with greater cognitive declines and higher tau deposition.^68,69 Fourth, we had a small number of women who reached menopause prior to 40 years (n = 81), indicating premature menopause. Future studies should examine differences in cognitive decline between premature menopause, early menopause (ages 40 to 45 years), and menopause after 45 years. Fifth, we do not have information on trans-/non-binary experiences. Individuals who are trans-/non-binary often have complicated reproductive health histories and report more subjective cognitive decline than cis-gender individuals.^70,71 Finally, other reproductive factors (e.g., age at first child),⁷² potential confounds (e.g., parenthood),⁴⁵ and sociocultural factors⁷³ may be critical to consider in the context of reproductive health and cognitive aging.

5. CONCLUSION

Hispanic/Latina women’s reproductive health throughout the life course was linked to various aspects of later life cognition, 7-year average change in cognition, and MCI. Our findings underscore the need to investigate sex/gender-specific factors in relation to cognitive aging among underserved populations.

Figure 2. Associations between age at menarche with 7-year cognitive change in the Study of Latinos-Investigation of Neurocognitive Aging.

Model 1 is crude; Model 2 is adjusted for age and education; Model 3 is additionally adjusted for language preference, Hispanic/Latino heritage, field center, income, marital status, insurance status, place of birth, and body mass index.

Abbreviations: Δ = change; B-SEVLT = Brief-Spanish English Verbal Learning Test

Research in context:

Systematic review:

We reviewed previous peer-reviewed publications using traditional online search engines (e.g., PubMed). The review was largely focused on publications related to female reproductive health (e.g., estrogen, number of births, menarche, menopause) and cognitive functioning in middle aged and older adults. Relevant articles are cited.

Interpretation:

In a sample of over 2,000 middle-aged and older Hispanic/Latina pos-menopausal women in the U.S., several reproductive health factors (e.g., age at menarche, oral contraceptive use, pregnancies, live births, and age at menopause) were associated with cognition and to a lesser extent 7-year cognitive change and mild cognitive impairment prevalence.

Future directions:

Our findings warrant investigation into whether interventions (medical or public health) to promote specific reproductive health factors might be 1) ethically and culturally appropriate and 2) beneficial to maintaining cognitive functioning in middle-age and older adulthood, among Hispanic/Latina women.