Sex Difference in the Associations of Socioeconomic Status, Cognitive Function, and Brain Volume with Dementia in Old Adults: Findings from the OASIS Study

Abstract

Background: Sex differences in the association of cognitive function and imaging measures with dementia have not been fully investigated. Understanding sex differences in the dementia-related socioeconomic, cognitive, and imaging measurements is crucial for uncovering sex-related pathways to dementia and facilitating early diagnosis, family planning, and cost control. Methods: We selected data from the Open Access Series of Imaging Studies, with longitudinal measurements of brain volumes, on 150 individuals aged 60 to 96 years. Dementia status was determined using the Clinical Dementia Rating (CDR) scale, and Alzheimer’s disease was diagnosed as a CDR of ≥0.5. Generalized estimating equation models were used to estimate the associations of socioeconomic, cognitive, and imaging factors with dementia in men and women. Results: The study sample consisted of 88 women (58.7%) and 62 men (41.3%), and the average age of the subjects was 75.4 years at the initial visit. A lower socioeconomic status was associated with a reduced estimated total intracranial volume in men, but not in women. Ageing and lower MMSE scores were associated with a reduced nWBV in both men and women. Lower education affected dementia more in women than in men. Age, education, Mini-Mental State Examination (MMSE), and normalized whole-brain volume (nWBV) were associated with dementia in women, while only MMSE and nWBV were associated with dementia in men. Conclusions: The association between education and the prevalence of dementia differs in men and women. Women may have more risk factors for dementia than men.

1. Introduction

Dementia refers to a group of symptoms associated with a decline in mental ability, characterized by memory, reasoning, and other thinking skills [1]. Five million US adults of more than 65 years of age had dementia in 2014, and it is projected to be nearly 14 million by 2060 [2]. Worldwide, approximately 50 million people live with dementia and this number is expected to triple by 2050 with the aging of the population, presenting a substantial challenge to patients, families, and society [3].

Alzheimer’s disease (AD) is the most common cause of dementia, accounting for 60–80% of dementia cases [4]. The common risk factors for AD dementia include age, genetics, family history, race/ethnicity [5], low education, excessive alcohol consumption, smoking, obesity, hypertension, hearing impairment, depression, physical inactivity, diabetes, poor heart health, and traumatic brain injury [6,7]. Addressing preventable risk factors (e.g., low education and low socioeconomic status) may reduce the risk of cognitive decline and prevent or delay up to 40% of dementia cases [7].

A sex-based investigation of dementia has been discussed and implemented [8,9]. Although education and socioeconomic status (SES) have been associated with dementia, with lower education level and lower SES groups presenting a higher risk for dementia [10,11,12], little information is known about how the associations differ in men and women. Moreover, the diagnosis of dementia may involve screening medical tests (e.g., MMSE (Mini-Mental State Examination)) for measuring cognitive impairment and magnetic resonance imaging (MRI) for measuring brain volumes. However, sex differences in the association of these measures with dementia have not been fully investigated. Understanding sex differences in dementia-related socioeconomic, cognitive, and MRI-imaging measurements would be crucial for uncovering the sex-related pathway to dementia and facilitating early diagnosis, family planning, and cost control.

The aims of this study are to evaluate (1) sex differences in education, cognitive function, and brain volumes and (2) sex differences in the dementia-related risk associations with the goal of determining whether there is a difference in the associations of these factors with the prevalence of dementia in men and women.

2. Methods

2.1. Data Source

The dataset was obtained from the Open Access Series of Imaging Studies (OASIS II) project with longitudinal MRI data on 150 subjects aged 60 to 96 years [13]. Subjects in the OASIS II project were selected from a larger longitudinal pool of individuals who had participated in MRI studies at the Washington University Alzheimer Disease Research Center (ADRC). The selection criteria included the availability of at least two visits in which MRI and clinical data were collected, at least three acquired T1-weighted images per imaging session, and right-hand dominance [13,14]. The ADRC‘s cognitively normal and impaired subjects were recruited primarily through media appeals and word of mouth, with 80% of subjects initiating contact with the center and the remainder being referred by physicians. The OASIS II project included 64 subjects with AD at their initial visit, and 14 subjects were characterized as nondemented at the time of initial scan and then clinically determined to have AD at the time of a subsequent scan. Data were acquired on the same scanner using identical procedures. Written informed consent was obtained from each participant, and from a caregiver for patients with AD with significant cognitive impairment. All subjects participated in accordance with the guidelines of the Washington University Human Studies Committee. Approval for the public sharing of the anonymized data was also specifically obtained.

2.2. Study Subjects

All subjects were screened for inclusion, and each subject underwent the ADRC‘s full clinical assessment as described below. Subjects with a primary cause of dementia other than AD (e.g., vascular dementia and primary progressive aphasia), active neurologic or psychiatric illness (e.g., major depression), serious head injury, history of clinically meaningful stroke, and use of psychoactive drugs were excluded, as were subjects with gross anatomical abnormalities evident in their MRI images (e.g., large lesions and tumors). However, subjects with age-typical brain changes (e.g., mild atrophy and leukoaraiosis) were included [13]. MRI acquisitions were obtained within the time window of one year before or after the clinical assessment for each subject (range from 0 to 352 days and mean around 111 days). Each subject was scanned for images on two or more occasions with an average of 2.53 visits and an average delay of 719 days between visits. The final dataset included 150 subjects and 373 imaging sessions—92 subjects with 2 imaging visits, 43 with 3 visits, 9 with 4 visits, and 6 with 5 visits. Our study was based on the above collected measurements and hence is a secondary analysis project with a longitudinal retrospective design.

2.3. Clinical Assessment

A Mini-Mental State Examination (MMSE) [15], including 11 questions, was conducted by trained physicians to assess the cognitive function in six areas of mental abilities, including orientation to time and place, attention and concentration, short-term memory (recall), language skills, visuospatial abilities, and ability to understand and follow instructions. The MMSE with a 30-point score was used to screen patients for cognitive impairment (a score of 24 or lower) and track changes in cognitive functioning over time in the study.

Dementia status was determined and staged using the Clinical Dementia Rating (CDR) scale [16]. The CDR scale is a staging instrument for dementia that rates subjects for impairment in six domains: memory, orientation, judgment and problem solving, function in community affairs, home and hobbies, and personal care. Based on the subject interviews and collateral sources, a global CDR score was derived from individual ratings in each domain. A global CDR of 0 indicates no dementia, and 0.5, 1, 2, and 3 represent very mild, mild, moderate, and severe dementia, respectively. The diagnosis of AD was established as a CDR greater than or equal to 0.5.

2.4. Image Acquisition and Processing

As detailed in the previous publications [13,17], each subject underwent an MRI scan on a 1.5-Tesla Vision scanner (Siemens, Erlangen, Germany), acquiring three or four T1-weighted MP-RAGE (Magnetization Prepared Rapid Gradient Echo) images in a single-session. The scanned files, originally in Siemens’ proprietary IMA format, were converted into 16-bit NiFTI1 format using a custom conversion program. To account for inter-scan head movement, the images were corrected and spatially aligned to an atlas space using a rigid transformation, which differed from the original piecewise scaling approach. This transformation ensured that all brains were placed within the same coordinate system and bounding box as the original atlas. Given the age range of the sample, an age-specific (old-only) atlas target was used. A 12-parameter affine transformation was then computed to minimize variance between the first MP-RAGE image and the atlas target. The remaining MP-RAGE images were aligned to the first, allowing for in-plane stretching, and resampled to generate a 1 mm isotropic image in the atlas space. This process resulted in a single high-contrast, averaged MP-RAGE image. Subsequent processing steps included skull removal using a loose-fitting atlas mask and intensity inhomogeneity correction to account for magnetic field non-uniformity. Intensity variations across contiguous regions were corrected using a quadratic inhomogeneity model fitted to phantom data.

2.5. Estimated Total Intracranial Volume and Normalized Whole-Brain Volume

The procedures used for measuring intracranial and whole-brain volumes have been described previously [13,17,18]. The estimated total intracranial volume (eTIV) was calculated by scaling the manually measured intracranial volume of the atlas using the determinant of the affine transformation that aligns each individual’s brain to the atlas. This method is minimally influenced by atrophy and maintains proportionality to manually measured total intracranial volume.

Normalized whole-brain volume (nWBV) was derived using the FMRIB Automated Segmentation Tool (FAST) within the FSL software suite version 6.0.8 (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki, accessed on 15 January 2025), designed for automated tissue segmentation. The image was first segmented to classify brain tissue into cerebrospinal fluid (CSF), gray matter, and white matter. The segmentation process iteratively assigned voxels to these tissue classes based on maximum likelihood estimates from a hidden Markov random field model, optimized using an Expectation–Maximization algorithm. Spatial proximity was incorporated to refine the probability of voxel classification based on intensity values. The nWBV was then computed as the proportion of brain mask voxels classified as tissue (either gray or white matter). The normalized volume is expressed as a percentage, representing the proportion of total gray and white matter voxels relative to the estimated total intracranial volume [18].

2.6. Other Characteristics

Socioeconomic characteristics of the subjects were collected, including age in years at the time of image acquisition, sex (men and women), years of education, and socioeconomic status (SES). Education was classified into two groups: high school or below and college or above in terms of years in school. SES was assessed using the Hollingshead Index of Social Position [19] and classified into categories ranging from 1 (highest status) to 5 (lowest status).

2.7. Statistical Analysis

Sample characteristics were summarized as means (standard errors) for continuous variables and counts (percentages) for categorical variables. Independent t tests were used to assess the significance of mean differences, and chi-square tests were used to evaluate the significance of percentage differences across sex and dementia groups. Generalized estimating equation (GEE) models were conducted to estimate adjusted odds ratios (ORs) with 95% confidence intervals (CIs) that measure the associations (main effects) of socioeconomic characteristics with eTIV and nWBV, and the associations of eTIV and nWBV with dementia. The association differences between men and women (interaction effects) were examined using GEE models by adding the first-order interaction terms of sex and other characteristics. Wald chi-square tests were used to assess the significance of the associations, including the main and interaction effects in the GEE models. Stratification analyses by sex were also conducted to identify the significant characteristics associated with eTIV, nWBV, and dementia in men and women, separately. In this study, an association estimate was considered statistically significant when the corresponding p-value was less than 0.05. All data analyses were performed using SAS software (version 9.4, Cary, NC, USA: SAS Institute Inc.).

3. Results

The present study sample consisted of 88 women (58.7%) and 62 men (41.3%) aged 60 to 96 years (

Table 1. Sample characteristics by sex and dementia status (n = 150).

	All	Sex		Dementia Status
		Women	Men	Yes	No
Age	75.4 ± 7.6	75.8 ± 7.9	74.9 ± 7.0	75.0 ± 6.8	75.8 ± 8.1
Sex
Men	41.3% (62)	--	--	55.4% (36)	30.6% (26) *
Women	58.7% (88)	--	--	44.6% (29)	69.4% (59) *
Education
High school or below	35.3% (53)	35.2% (31)	35.5% (22)	50.8% (33)	23.5% (20) *
College or above	64.7% (97)	64.8% (57)	64.5% (40)	49.2% (32)	76.5% (65) *
SES	2.5 ± 1.1	2.5 ± 1.1	2.4 ± 1.2	2.7 ± 1.2	2.3 ± 1.1 *
MMSE	27.6 ± 2.9	28.1 ± 2.6	26.8 ± 3.3 *	25.4 ± 3.3	29.2 ± 0.9 *
eTIV	1474.4 ± 143.8	1390.9 ± 121.8	1593.0 ± 170.3 *	1473.8 ± 173.1	1474.9 ± 176.9
nWBV	73.6 ± 3.5	74.3 ± 3.5	72.6 ± 3.6 *	72.5 ± 3.1	74.5 ± 3.8 *
Dementia status
Yes	43.3% (65)	32.9% (29)	58.1% (36) *	--	--
No	56.7% (85)	67.1% (59)	41.9% (26) *	--	--

Note: * indicates that the p-value is less than 0.05, and the difference in the corresponding characteristic across sex and/or dementia groups is significant at the 0.05 level. eTIV, estimated total intracranial volume; MMSE, Mini-Mental State Examination; nWBV, normalized whole-brain volume; and SES, socioeconomic status.

). At the time of their initial visit, the average age of the subjects was 75.4 years, and 65 (42.7%) had dementia with a CDR score greater than 0. Of all the subjects, 35.3% received a high school education or below. Compared to men, women had a lower eTIV (1390.9 vs. 1593.0, p < 0.0001), higher MMSE (28.1 vs. 26.8, p = 0.0073), higher nWBV (74.3 vs. 72.6, p = 0.0029), and lower dementia rate (32.9% vs. 58.1%, p = 0.0022). The subjects with dementia (vs. non-dementia) included fewer women (44.6% vs. 69.4%, p = 0.0022), were more likely to receive high school education or below (50.8% vs. 23.5%, p = 0.0005), and had a poorer SES (2.7 vs. 2.3, p = 0.034), lower MMSE (25.4 vs. 29.2, p < 0.0001), and lower nWBV (72.5 vs. 74.5, p = 0.0008).

The first-order interaction analysis of sex and other characteristics showed a significant sex–SES interaction effect on eTIV (p = 0.0064), indicating that the associations between SES and eTIV differ in men and women. No interaction effects of sex and other characteristics on nWBV were found. The stratification analysis by sex (

Table 2. GEE models for associations of socioeconomic characteristics with eTIV and nWBV by sex.

	Women						Men
	eTIV			nWBV			eTIV			nWBV
	Estimates	Standard Errors	p Values	Estimates	Standard Errors	p Values	Estimates	Standard Errors	p Values	Estimates	Standard Errors	p Values
Age	0.90	0.90	0.32	−0.33	0.02	<0.0001	2.93	1.67	0.08	−0.33	0.04	<0.0001
Education (high school or below vs. college or above)	7.94	31.81	0.80	−1.61	0.85	0.06	−18.38	65.54	0.78	0.07	1.38	0.96
SES	−15.60	13.62	0.25	0.46	0.38	0.22	−59.24	25.17	0.02	0.28	0.54	0.61
MMSE	−2.32	1.17	0.05	0.16	0.06	0.006	0.29	1.65	0.86	0.15	0.02	<0.0001

Notes: eTIV, estimated total intracranial volume; MMSE, Mini-Mental State Exam; nWBV, normalized whole-brain volume; and SES, socioeconomic status.

) showed that a lower SES was associated with a reduced eTIV in men (−59.24 ± 25.17, p = 0.02), but not in women. Older age was associated with a reduced nWBV (−0.33 ± 0.04, p < 0.0001 for men; −0.33 ± 0.02, p < 0.0001 for women), and higher MMSE scores were associated with an increased nWBV (0.15 ± 0.02, p < 0.0001 for men; 0.16 ± 0.06, p = 0.006 for women) in both men and women.

The changes in eTIV and nWBV with age are presented in Figure 1. The figure shows that nWBV decreased with age while eTIV did not change with age, with parallel trend lines (approximately identical slopes) in men and women indicating that there were no sex differences in the associations of age with eTIV and nWBV.

In the general study sample including all subjects, age, sex, MMSE score, eTIV, and nWBV were significant predictors of dementia (Figure 2). GEE models involving interactions showed that the interaction effect of sex and education on dementia was significant, indicating a difference in the association of education with dementia in men and women. The stratification analysis by sex revealed that age, education, MMSE score, and nWBV were associated with dementia in women; however, only MMSE score and nWBV were associated with dementia in men (

Table 3. The GEE models for associations of socioeconomic and brain imaging characteristics with dementia by sex in the longitudinal study sample.

	Women			Men
	OR	95% CI	p Values	OR	95% CI	p Values
Age	0.84	0.75–0.93	0.0008	0.96	0.89–1.03	0.25
Education (high school or below vs. college or above)	7.06	1.36–36.56	0.02	0.81	0.15–4.23	0.80
SES	0.66	0.33–1.33	0.25	0.86	0.43–1.73	0.68
MMSE	0.26	0.13–0.52	0.0002	0.74	0.65–0.83	<0.0001
eTIV	1.00	0.99–1.01	0.13	1.00	0.99–1.00	0.034
nWBV	0.71	0.59–0.86	0.0004	0.81	0.69–0.95	0.010

Notes: CI, confidence interval; eTIV, estimated total intracranial volume; MMSE, Mini-Mental State Exam; nWBV, normalized whole-brain volume; OR, odds ratio; and SES, socioeconomic status.

). A lower education level was associated with a higher risk of dementia in women but not in men.

4. Discussion

Based on the frequency count of all individuals with AD, more women than men are living with a diagnosis of AD [4,8,20]. Women also have a greater lifetime risk of developing AD [4]. An important contributor to this sex difference in both the frequency and the lifetime risk is that women live longer than men. However, when assessing the prevalence/rate (total number with disease divided by the total number of individuals in the population) of AD by sex, the results are conflicting. Several studies in Europe have reported a higher prevalence of AD among women [21]. However, the studies in the United States [22], and around the world [23], have not reported a significant sex difference. In the present study sample, aged 60–96 years, we found that the dementia rate was higher in women than in men, which is consistent with the previous reports [21,24,25]. The conflicting results appear to be somewhat dependent on the time period and geographic regions where the studies were conducted.

nWBV has been reported to decline with age, while eTIV does not vary much with age [17,26]. Our results confirm those of previous reports and further show that the trend in nWBV and eTIV over age did not differ in men and women, although the baseline nWBV was higher and the baseline eTIV was lower in women than in men (Figure 1 and Table 1). A previous study demonstrated the presence of significant differences in neuroanatomical structure volumes between men and women in a population sample of participants above 66 years of age [27]. It revealed that the adjusted gray and white matter volume was larger in women. Since the total brain volume is calculated as the sum of the gray and white matter volumes, this finding is consistent with our analysis that the average nWBV was higher in women than in men. Our GEE analysis showed that nWBV was negatively linked to the prevalence of dementia, with a larger nWBV indicating a lower risk of dementia. The results together could partly explain why the dementia rate was lower in women in the present study sample, while the strength of the association of nWBV with dementia did not differ in men and women.

Our analysis showed that, after adjusting for socioeconomic factors, MMSE scores and nWBV were negatively associated with dementia in both men and women while there was not a sex difference in this association. The MMSE has been used to screen patients for cognitive impairment, track changes in cognitive functioning over time, and often to assess the effects of therapeutic agents on cognitive function. Its utility decreases when patients with mild cognitive decline and psychiatric conditions are assessed [28,29] and should not be used alone as a tool for diagnosing dementia [30]. As a similar strength of its association with dementia was found in both sexes, the higher average MMSE scores in women may be a contributor to the sex difference in the prevalence of dementia. nWBV is the normalized total brain volume corrected for the intracranial volume. Consistent with this study, it has been shown to be larger in women than in men, although uncorrected brain volume measurements, such as gray matter and white matter, are smaller in women [27]. As the strength of its association with dementia is the same in both men and women, similar to MMSE, the higher average nWBV in women may be a contributor to the lower dementia rate in the female group. In addition, the MMSE score and nWBV both declined with age. The role of the MMSE score and nWBV in relation to dementia may follow the proceeding pathway: aging factors→decline in MMSE and nWBV→dementia

In this study, a lower SES was found to be associated with a smaller eTIV, and the association of SES with eTIV was weaker in women than in men. eTIV appeared to be negatively associated with dementia, and yet, compared to the MMSE score and nWBV, its effect size was smaller in terms of odds ratios (very close to 1, indicating no effect) that characterize the risk of the disease. The eTIV did not change with age (Figure 1), indicating that aging-related factors may not affect eTIV. Whether the eTIV can predict the diagnosis of dementia requires further investigation.

Socioeconomic characteristics, including educational level, income level, and occupation, have been investigated as potential risk factors for dementia [12,31,32]. A Danish population study with 10,191 individuals reported that higher household incomes (vs. lower incomes) were associated with less likelihood of dementia diagnosis after referral, and led to earlier dementia diagnosis and less severe disease at diagnosis [12]. Another study systematically reviewed 71 studies with 88 study populations and found that lower education is associated with a greater risk for dementia in many but not all studies [31]. An epidemiologic project studied 10,781 Health and Retirement Study participants interviewed from 1998 to 2012 (US) and showed that, compared to a low SES at all three time points (childhood, early adulthood, and older adulthood), a stable high SES predicted the best memory function and slowest decline, high school completion had the largest estimated effect on memory, and a high late-life income had the largest estimated benefit for slowing memory declines. In the present study, we reviewed 150 participants aged 60–96 years and found that education and SES were not associated with dementia (Figure 2). However, the interaction and stratification analysis with GEE models showed that sex moderated the association between education and dementia, and a lower education level was associated with a higher risk of dementia in women, but not in men. The level of education associated with the risk of dementia may vary with geographic regions and study populations, and more years of education may not uniformly attenuate the risk. Sex differences in such associations and the underlying pathways need to be further investigated in the future research.

There are a few limitations in this study. First, this is a secondary data analysis study using the dataset from the Open Access Series of Imaging Studies II (OASIS II) project. The inclusion criteria for selecting subjects in the OASIS II project may cause selection errors/bias that confound the interpretations of results. Second, the study sample is relatively small (only 150 subjects), especially for an observation study. However, each subject had multiple visits and hence multiple clinical measurements, so the number of outcome measurements is large enough. The repeated measure design of our study can potentially raise the analysis power and improve the precision and accuracy of the scientific findings. Third, as a review pointed out, considering that the time data were collected, the data might not accurately reflect the current situation and could limit the ability to address real-time research questions effectively.

5. Summary

In this study, we examined older adults, aged 60–96 years, and found that dementia prevalence was lower in women than in men. Women also had larger normalized whole-brain volume (nWBV), higher Mini-Mental State Examination (MMSE) scores, and smaller estimated total intracranial volume (eTIV). In both sexes, aging and lower MMSE scores were linked to decreased nWBV, with no significant sex differences in this relationship. Lower socioeconomic status (SES) correlated with smaller eTIV, though this effect was stronger in men. Significant predictors of dementia included age, education, MMSE score, and nWBV in women, while in men, only MMSE score and nWBV were significant predictors. Additionally, lower education levels were more strongly associated with dementia in women than in men. These findings suggest that the relationship between education and dementia prevalence differs by sex, and women may have more risk factors for dementia than men.