INTRODUCTION
Background
Alzheimer’s disease (AD) is the most common cause of dementia. People affected by the disease experience a progressive decline in memory, language, and cognition; eventually losing the ability to take care of themselves and often requiring institutionalization. By 2023, almost seven million people are living with AD in the United States,1) and the yearly cost of health care, long-term care, and hospice services was estimated to be close to $450 billion.2) Current pharmacologic treatments for AD have shown limited effects on quality of life and cognitive function. Women are particularly affected by the disease, representing approximately two thirds of all AD patients.3) Evaluation of preventive measures to delay the onset of symptoms or halt the progression of the disease in women is a key focus of current research in the public health arena.4)
Being a woman is one of the three main risk factors for the development of AD. Every woman goes through menopause at some point in her life, although not all women develop AD in older age. Menopause can be either spontaneous or induced. Spontaneous menopause is usually the consequence of aging, when the number of viable oocytes becomes insufficient to produce adequate levels of estradiol and progesterone to sustain regular menstrual cycles. Women can also experience spontaneous premature menopause before age 40 or early menopause between ages 40 and 45. Younger age at the onset of menopause has been associated with increased all-cause mortality.5)
17β-estradiol directly influences brain metabolism and regulation of neuroinflammation.6) Women who undergo premature surgical menopause have an earlier onset of cognitive decline often accompanied by the presence of neural amyloid plaques. In contrast, women prescribed hormone replacement therapy (HRT) soon after induced menopause, show less decline in global cognition and memory.7) 17β-estradiol crosses the blood-brain barrier and binds to receptors in different areas of the brain including the hippocampus, amygdala, and hypothalamus. The connection between lower estradiol production and susceptibility to neuroinflammation has been identified as a unifying mechanism that links menopause with elevated risk of AD in women.8) In addition, the biological reproductive lifespan, from menarche to menopause, determines the duration of brain exposure to estrogens.9) This exposure to estrogens is also affected by administration of hormone therapy during and after the reproductive lifespan.
The primary objective of this study was to evaluate the association between the duration of the reproductive lifespan and cognitive function as a surrogate for the development of AD. The secondary objective was to assess the relationship between the use of hormone therapy and cognitive function in older age. In this study, cognitive performance scores are used as proxies for cognitive function and potential risk of AD, consistent with their validation in prior epidemiological and clinical research. We utilized data from the National Health and Nutrition Examination Survey (NHANES) collected between 2011 and 2014 to examine this associations —reproductive lifespan, hormone therapy use, and cognitive function in a nationally representative sample of women in the United States.
MATERIALS AND METHODS
Study Design and Sample
Our study used publicly available data from the NHANES survey: 2011–2012 and 2013–2014 cycles (NHANES). These data cycles were selected due to the availability of comprehensive cognitive evaluation that began in 2011–2012 cycle. NHANES used a multistage probability sampling designed to produce a weighted, representative sample of the population residing in the United States. The National Center for Health Statistics Research Ethics Review Board approved all NHANES protocols, and informed consent was obtained from all participants.
We obtained data from women older than 60 who completed the cognitive assessment and gave information regarding their age at menarche, age at menopause and use of exogenous hormones. The total population (n=19,931) was screened for older than 60 years (n=3,632), female sex (n=1,872). We excluded participants with incomplete or missing information for the final study population (n=1,374). Fig. 1 illustrates the flow chart for selecting the study sample.
Sociodemographic Characteristics
Sociodemographic variables included age, race, and educational level. Race included Mexican American, other Hispanic, non-Hispanic White, non-Hispanic Black, and multiracial. The educational level included less than high school, high school, college, and beyond as presented in the original dataset.
Definition of Exposure
This study focused on two main variables. First, exposure to exogenous hormones was defined based on the NHANES survey question: Have you ever used female hormones for menopause? This includes any reported use of estrogen, progesterone, or combined hormone therapy, regardless of formulation, route, dose, or duration. Second, reproductive span was calculated as the difference between age at menopause and age at menarche, representing the natural duration of endogenous estrogen exposure.
The NHANES questionnaire does not capture detailed information on the duration, timing, formulation, or dosage of hormone therapy. Thus, our analysis is limited to a binary exposure variable ("ever used" vs. "never used").
Outcome Variable: Cognition Assessment
Trained interviewers assisted participants in a series of cognitive tests. The cognitive battery consisted of the word learning and recall modules from the Consortium to Establish a Registry for Alzheimer’s Disease (CERAD), the Animal Fluency Test (AFT), and the Digit Symbol Substitution Test (DSST). Non-response to cognitive testing for any reason was treated as missing data and not included in the analyses.
CERAD assesses immediate learning and delayed recall of new verbal information.10) The W-L administers three learning trials and a delayed recall. In the three learning trials, participants are instructed to read aloud 10 words and immediately after, they are asked to recall as many words as possible. The delay recall occurred approximately 10 minutes after the start of the word learning trials. The maximum score in each trial is 10 and the maximum score for the word recall is 40 (sum of the three trials plus the delayed recall).11) The AFT assesses the categorical verbal fluency by asking participants to name as many different animals as possible within 1 minute. The DSST is a subtest within the Wechsler Adult Intelligence Scale 3rd edition (WAIS-III) that assesses processing speed, sustained attention, and working memory. It is administered on paper and includes a key with nine numbers and symbols. NHANES participants had 2 minutes to draw the unique symbol that corresponded with the number into 133 paired, blank boxes. The score reflects the total number of correct matches.
Covariates
Demographic characteristics and other covariates used in this analysis included age, race, and educational level. Race was categorized as non-Hispanic white, non-Hispanic black, Mexican-American, Hispanic and other/multiracial. Education was defined as having less than a high school education, having some college, or being a college graduate or higher.
Statistical Analysis
Normally distributed data are presented as means and standard deviation. Normality was evaluated with the Kolmogorov-Smirnov test. Comparisons between low and normal cognitive scores in terms of continuous covariates was performed with Student t-test. Number of observations and frequency were used to present categorical variables. Univariate analyses for these variables were performed by the chi-square test.
Participants were categorized into the low cognitive performance score and normal cognitive performance score if their overall score for word recall, animal fluency, and digit symbol was below or above the first quartile, respectively.12) We conducted univariate and multivariate logistic regression analyses to examine the association between HRT and cognitive performance as well as the association of reproductive span and cognitive performance. The analysis was performed using sample weights in line with the NHANES guidelines. One-way ANOVA and Tukey analyses were used to evaluate differences for educational level and race. We also employed restricted cubic spline curves to assess the relationship between age and cognitive performance using the information from the logistic regression models. In addition, we performed subgroup analyses for age groups to further evaluate the association between hormone therapy and cognitive performance. Finally, we performed sensitivity analysis before and after deletion of missing values. All analyses were conducted using R software (version 4.4.0; R Foundation for Statistical Computing, Vienna, Austria). p<0.05 was considered as statistically significant.
RESULTS
Baseline Characteristics
The baseline characteristics of the participants are shown in Table 1. We found significant differences between low and normal cognitive performance according to the CERAD word recall test in age, race, educational level, and hormone replacement. The variable description and their meaning from the NHANES codebook are shown in Table 2.
Association between Age, Reproductive Span and Cognitive Performance
Age for all groups and educational level exhibited a positive association with all cognitive scores (Fig. 2). Furthermore, the one-way ANOVA and Tukey analyses revealed that above 65 years of age, there is no statistically significant decline in cognitive word recall scores (p=0.07), digit symbol (p=0.64), and animal fluency (p=0.25).
There was no association between age at menopause for word recall (p=0.32), digit symbol (p=0.51), and animal fluency (p=0.92). No statistically significant associations between age at menarche and word recall, digit symbol, and animal fluency were found (p=0.32, p=0.5, and p=0.9, respectively). The reproductive span, defined as the difference between age at menopause and age at menarche, was associated with word recall performance (p=0.027) but not with digit symbol or animal fluency (p=0.11 and p=0.63, respectively).
Association between Hormone Therapy and Cognitive Performance
Having received hormone therapy exhibited significant association with word recall, digit symbol, and animal fluency cognitive performance (p=0.028, p=0.017, and p=0.009, respectively). Table 3 shows the odds ratio (OR) and 95% confidence interval (CI) for the association between exposure, covariates, and cognitive performance. Fig. 2 illustrates the associations between hormone therapy, reproductive span, and demographic variables with cognitive performance scores.
Effect of Hormone Therapy on Age-Related Cognitive Performance
Considering that age as a categorical variable exhibited significant associations with cognitive performance in the three evaluated dimensions (word recall, animal fluency, and digit symbol), we further explored the effect of hormone therapy on those associations by means of restricted cubic splines. Fig. 3 shows that the cognitive decline that occurs with age is slowed down by hormone therapy. Furthermore, the lowest score achieved with age is higher in those participants who received hormone therapy for all cognitive dimensions.
Stratified Analysis by Age
Table 4 shows the results of the subgroup analyses. Having received hormonal therapy is significantly associated with word recall performance in the age ranges 65–69 and >80 years, with animal fluency in ranges 70–74 and >80 years, and with digit score in ranges 65–69 and 70–74 years. These age ranges have some correspondence with the ages where the hormonal therapy has most effect on age-related cognitive performance (Fig. 3). Regarding the effect of reproductive span on cognitive performance by age group, statistical significance was found for animal fluency scores for the age groups of 60–64 and 65–69 years, and for digit recall score for the group of 60–64 years.
Sensitivity Analysis
Estimates from sensitivity analysis revealed that having received hormonal therapy is still associated with word recall (OR=0.703, 95% CI 0.55–0.89) and animal fluency score (OR=0.72, 95% CI 0.58–0.89). No association between hormone therapy and digit symbol was found (OR=0.79, 95% CI 0.59–1.06). Regarding reproductive span, no significant association with cognitive performance was found.
DISCUSSION
In this nationally representative sample of postmenopausal women older than 65 years, we observed an age-dependent cognitive decline assessed by the CERAD word recall, animal fluency, and digit symbol scores independent from ethnicity and educational level. The cognitive performance was associated with having received hormone therapy, and the protective effect of hormones is maximal for age groups with the most marked cognitive decline. In addition, the reproductive span (number of years between menarche and menopause) was not found to be associated with cognitive scores.
The decline in estrogen levels occurring during and after menopause has been proposed as an etiological factor involved in the higher prevalence of AD in women.13) However, some studies evaluating HRT as a strategy to prevent cognitive decline have shown inconsistent findings or even harm.14-18) In a controlled trial of 2,763 women randomized to receive conjugated estrogen (0.625 mg) plus medroxyprogesterone acetate (2.5 mg) or placebo for 4 years, cognitive function assessed with different tests including word list recall was not different between the groups.14) The Kronos Early Estrogen Prevention Study (KEEPS)-Cog trial and its continuation study (10-year follow-up) suggested no cognitive benefit or harm of HRT (estrogen plus progesterone) after 4 years of treatment when it was started within three years of the last menstrual period.15,16) A Cochrane database systematic review evaluated randomized controlled trials of estrogen and estrogen-progesterone hormone therapy to determine their effect on cognitive function after at least 2 weeks of treatment in postmenopausal women. Sixteen trials were analyzed to show that hormone therapy failed to prevent cognitive impairment after 5 years of treatment (OR= 1.05, 95% CI 0.72–1.54).17) In addition, the Women’s Health Initiative Memory Study concluded that estrogen plus progestin therapy increased the risk of dementia in postmenopausal women (hazard ratio=2.05, 95% CI 1.21–3.48).18) The studies mentioned above share heterogeneity in selected medications, treatment duration, and cognitive tests employed in the context of a relatively reduced observation window. Our study used the question, "ever used female hormones?" We feel that this question serves as a straightforward and practical measure for hormone therapy, in line with the simplicity and comprehensiveness of NHANES data collection methods. The question captures hormone use without requiring doses, routes of administration, or medications, which can often be inaccurate in retrospective studies. Moreover, this measure facilitates comparisons with large-scale studies, such as the Women’s Health Initiative Memory Study. We may argue that our study contributes to the understanding of the relationships between self-reported hormone therapy use and cognitive performance. Additionally, by not imposing constraints of time of treatment and observation as well as type of hormone therapy, our study adds value as a real-life setting where the effect of hormone therapy can be isolated from other covariates. We showed a protective effect of hormone therapy on cognitive performance using validated scores such as CERAD word recall.14) The lack of consistency between our results and those of studies such as the Women’s Health Initiative Memory Study, the Cochrane systematic review, and other reports highlights the complexity of hormonal effect on cognitive function.14-18) Unlike these trials, which evaluated specific regimens and dosages, our study provides a broader perspective by including all types of hormone therapy. The methodological difference may explain the discrepancy between the results, suggesting that population-level hormone use could have subtle but meaningful associations with cognitive function.
Our results point to a positive effect of hormone therapy on cognition. Recent evidence has shown that early institution of HRT at the beginning of post-menopause may have beneficial effects on cognitive control prefrontal mechanisms.19) In addition, Saleh et al.20) found that HRT improved delayed memory and larger entorhinal and amygdala volumes in APOE4 carriers. These findings demonstrate that the effect of HRT has a structural effector in different areas of the brain. An elegant study by Mosconi et al.21) used positron emission tomography to reveal higher estrogen receptor density during the menopause transition in cerebral estrogen-regulated networks. The higher density of estrogen receptors was associated with poorer memory performance during perimenopause. The literature has shown the role of estrogens as a modulator of dementia risk, with studies showing both beneficial and deleterious effects of HRT. These contradictory results underscore the complexity of the problem of menopause and cognitive performance, with variables as diverse as age of initiation, genetic predisposition, and specific formulations playing important roles that are difficult to isolate and whose effect on cognitive function difficult to disentangle.
Age-related cognitive decline is a human experience affected by a plethora of factors including genetic contributions, vascular health, and lifestyle habits among others.22) Cognitive aging seems to follow different trajectories between males and females. Specifically, at baseline, males outperform females on visuospatial ability, whereas females outperform males in other dimensions.23) The dimensions evaluated in this study have been used in other studies in the field and focus on learning ability for new verbal information,24) language skills,25) and attention and processing speed.26) We evidenced that in postmenopausal women, the decline in these cognitive dimensions was steeper between 65 and 70 years old, with a second drop after 75 years. This finding is consistent with the results of a population-based study in the Netherlands that showed after a linear decrease in cognitive function before 65 years, there is a steep drop between 65 and 70 years.27) To our knowledge, no prior report has revealed differential positive effects of hormone therapy on cognitive function in terms of specific age groups. Our findings clearly show that having received hormonal treatment serves as a buffer during the years of steeper deterioration of cognitive function. The mechanism for this finding is not understood yet. However, it may be related to age-related changes in brain volume, perfusion and/or estrogen receptor density. Naessen et al.28) showed that high estrogen levels in a group of 70-year-old patients was associated with unhealthy artery wall on ultrasound. The authors suggest that a high ratio of estrogen E2/testosterone represents an endogenous response to atherosclerosis. We may argue that as atherosclerotic disease becomes more prevalent with age, this hormonal compensatory response may explain the “bump” in cognitive function that we observed between 70 and 75 years. E2 estrogens have been ascribed neuroprotective properties.29) On the other hand, it has been demonstrated that the sharp decrease in estrogens after menopause is accompanied by upregulation of brain and estrogen receptors.21,30) It is also highly recognized that the medial temporal lobe and prefrontal cortex, necessary for normal cognitive function, are highly vulnerable to aging in humans.31) Furthermore, estrogen administration enhances hippocampal gray matter volume in postmenopausal women.32) Taken all the evidence together, we believe that the positive effect of hormones on cognitive function that we found in our study may be due to the regulation of brain estrogen receptor density and its effects on gray matter volume and vascular structure and function.
A short reproductive span has been explored as a surrogate of estrogen exposure in multiple studies and has been proposed as a risk factor for the development of cognitive dysfunction and dementia.33) The opposite has also been reported, with early menarche and/or late menopause being associated with higher dementia risk.34) We found no association between the reproductive span and cognitive function scores. It is possible that our study may have failed to capture the risk of cognitive decline in specific populations at higher risk of dementia such as carriers of the APOE4 genotype, other genetic predisposition, or exposure to certain lifestyle risk factors. In fact, Geerlings et al.35) found negative associations of reproductive lifespan and dementia in APOE4 carriers. Another imaging study showed larger MRI-derived gray matter volume in brain regions vulnerable to cognitive aging in patients with longer reproductive span and higher number of children.36) Future studies evaluating specific risk factors for cognitive deterioration and AD are necessary to determine the presence of the association between time of exposure to estrogens and cognitive decline.
Notably, the lack of data on hormone therapy duration in NHANES restricts our ability to assess whether longer treatment confers greater cognitive benefit. Evidence from clinical and observational studies suggests that early initiation of hormone therapy near menopause onset may be more protective against cognitive decline than later initiation.19,20) Conversely, prolonged use or initiation in late postmenopause has been associated with increased dementia risk in some trials.18) Our findings should therefore be interpreted as reflecting an aggregate association between any history of hormone use and cognitive performance, independent of treatment length or timing.
Our study has limitations. First, the use of self-reported hormone therapy without data on duration, timing, formulation, or dosage limits our ability to assess dose-response relationships or critical windows of intervention. Prior research suggests that the timing of hormone initiation, particularly soon after menopause, may influence its cognitive effects.16) Our binary measure precludes such nuanced analysis. Future studies should incorporate detailed treatment histories to better understand the role of duration and timing in cognitive outcomes. Second, the cross-sectional nature of NHANES data precludes causal inference. While we observed associations between hormone therapy and cognitive performance, temporal relationships and long-term effects cannot be established. Future longitudinal studies are needed to examine the trajectory of cognitive change in relation to hormone initiation, duration, and discontinuation. Third, our study population did not identify high-risk populations such as APOE4 carriers, limiting the generalizability of our results. Fourth, the cognitive tests used in NHANES, although validated, may not evaluate all dimensions of cognitive function relevant to age-related cognitive decline and AD. Future research should prioritize longitudinal studies to establish causality, incorporate genetic and imaging analyses to elucidate complex etiologic mechanisms, and investigate the role of specific hormone therapies and timing on cognitive trajectories in perimenopausal and postmenopausal women.
In conclusion, our study highlights the positive association between hormone therapy and cognitive performance in postmenopausal women, particularly in age groups with the steeper decline. While no association was found between reproductive span and cognitive function, these findings underscore the complexity of hormonal effects on cognitive function and the importance of considering multiple covariates in future research. Our study results support the potential benefit of hormone therapy to ameliorate the age-related cognitive decline affecting postmenopausal females. Our findings should be interpreted as associative rather than causal. Prospective cohort studies and randomized trials with extended follow-up are necessary to clarify whether hormone therapy mitigates cognitive decline and to identify optimal timing, formulation, and duration of treatment.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
