Association between education and rate of cognitive decline among individuals with Alzheimer's disease: A multi-national European observational study

Abstract

Background

The cognitive reserve (CR) hypothesis presumes higher tolerance of Alzheimer's disease (AD)-related pathology without functional decline for those with high education and more rapid decline after AD onset. Evidence supporting the second part of the hypothesis has been largely confined to U.S.-based studies.

Objective

To assess the relationship between education and cognitive decline in a multi-national European cohort of older adults living with AD.

Methods

We analyzed data from participants recruited into the GERAS-EU cohort study from AD clinics in the United Kingdom, Germany, and France. Linear mixed models were employed to assess the relationship between education (dichotomized using a 12-year cutoff) and cognitive decline measured by Mini-Mental State Examination (MMSE) scores during 1.5 to 3 years of follow-up, adjusting for age, sex, time from formal diagnosis, country, comorbidities, and AD treatment.

Results

A total of 1313 participants were analyzed, with mean age of 77.3 years (SD = 7.6), 715 (54.5%) females, and 378 (28.8%) with high education (≥12 years). Participants with high education experienced a 0.19-point greater decline (versus low education group) in MMSE scores every 6 months during follow-up (95% Confidence Interval: 0.03–0.35, p = 0.02). The secondary analyses (stratified by disease severity, sex, or country) showed a consistent direction of the association, although only significant in the severe AD group (p = 0.01).

Conclusions

Our findings provide partial support for the CR hypothesis. Delayed AD diagnosis in individuals with high education may contribute to faster decline after diagnosis, highlighting the importance of sensitive screening for early signs of cognitive impairment.

Keywords

Alzheimer's disease cognitive decline cognitive reserve cognitive test education

Introduction

Cognitive reserve (CR) attributes the reduced manifestation of brain dysfunction, despite underlying pathology, to the adaptability of cognitive processes.^1,2 According to the CR hypothesis, individuals with higher CR can better tolerate Alzheimer's disease (AD)-related pathology and exhibit delayed symptom onset, but tend to experience faster cognitive decline once symptoms manifest or a clinical diagnosis is established.^1,3

Education is one of the most common socio-behavioral proxies for CR.^4,5 Some observational studies have found that higher educational attainment is associated with a more rapid cognitive decline in participants with prodromal AD^6,7 or clinically diagnosed AD or dementia.^8–12 However, most studies exploring heterogeneous trajectories of cognitive decline after symptom onset have taken place in the United States (U.S.).^8–12 International studies have often been conducted in individual countries with small samples and inconsistent findings.^6,13–17 A large-sample analysis of the Health and Retirement Study and its global companion studies found that higher educational attainment (i.e., more years of education) was associated with a delayed onset of prodromal AD-related accelerated cognitive decline across multiple countries.⁷ However, the association between higher educational attainment and faster cognitive decline after the onset of accelerated decline was observed only among participants from the U.S. and Sweden.⁷ Studies involving approximately 200 outpatients with late-onset AD, recruited from a single hospital in Brazil and with an average of fewer than five years of schooling, found no significant association between education and cognitive decline over a one-year follow-up period.^15,16 Two other Brazil-based studies using similar study samples reported protective effects where more years of education were associated with a longer duration between dementia onset and reaching a low Mini-Mental State Examination (MMSE) score of 20 in females and APOE ε4 carriers of both sexes.^13,14 These studies make an important contribution to the literature by examining a population with lower overall educational attainment; however, the generalizability of the findings to populations with higher educational attainment warrants further investigation.^13–16

Gaps in these previous studies underscore the need for more robust international evidence to better understand the role of education and CR in cognitive decline following the clinical onset of AD. Furthermore, characterizing this relationship in diverse populations may support the development of more sensitive screening tools for the early detection of AD-related symptoms and pathology in individuals with varying levels of education. In this study, we aimed to assess the relationship between education and cognitive decline trajectories in participants with clinically diagnosed AD by leveraging a large, multinational sample with longitudinal cognitive follow-up. We assessed this relationship in the full sample (primary analysis) and across subgroups stratified by disease severity, sex, and country respectively (secondary analyses). We hypothesized that higher educational attainment would be associated with more rapid cognitive decline in participants with AD.

Methods

Study design

This study is a prospective secondary data analysis examining the relationship between educational attainment and cognitive decline in older adults living with AD. We analyzed data from the GERAS observational study conducted across three European Union countries (GERAS-EU).¹⁸ This data contained longitudinal follow-up assessments of cognitive function in participants living with AD and information about their education, allowing us to assess our hypothesis. We obtained the de-identified GERAS-EU dataset through the Alzheimer's Disease Data Initiative's AD Workbench, which aims to facilitate open data access and collaboration globally. Because this study analyzed de-identified data, it was exempt from the Institutional Review Board review.

Setting and sample

The GERAS-EU study¹⁸ is an 18-month prospective, multi-center cohort study conducted across three European countries: Germany, France, and the UK Participants were recruited by physicians, primarily from specialist secondary care facilities such as memory clinics. An extension of the study to a 3-year follow-up period was implemented in France and Germany. Community-dwelling individuals meeting criteria for probable AD based on the National Institute of Neurologic and Communicative Diseases and Stroke and the Alzheimer's Disease and Related Disorders Association AD criteria¹⁹ were recruited for participation. The inclusion criteria for participation in the GERAS-EU study included an MMSE score of 26 or less, age at recruitment of 55 or older, written consent of participant and caregiver, and caregiver's willingness to comply with study design requirements. Exclusion criteria included history, clinical signs or CT scan of stroke or transient ischemic attack, a history (e.g., treatment) of Parkinson's disease prior to or at the onset of AD, or probable Lewy-body disease, and simultaneous participation in studies involving treatment intervention or investigational drug treatments at the baseline observation. AD treatment could be prescribed as part of clinical practice throughout the study, and all treatment decisions were made solely at the discretion of the physician and patient. Patients who enrolled in a study with an interventional treatment or investigational drug during the follow-up period (n = 13) were allowed to remain in this observational study.

To derive the analytic sample for this study, we excluded participants who met any of the following criteria: (1) having fewer than two visits (n = 193); (2) missing baseline data for the MMSE total score and covariates included in the regression analysis (i.e., age, time from formal diagnosis, sex, the number of relevant comorbidities, AD treatment type, and study country) (n = 6); and (3) having a baseline MMSE score of zero (n = 17). We included the third condition because our study assessed cognitive decline based on changes in MMSE scores over time.

Data collection

The MMSE was administered to all participants at their baseline visits and in follow-up visits, approximately every 6 months. Demographic information and other measures were collected through participant's self-reports at the baseline evaluation.

Disease severity was characterized at baseline using MMSE scoring criteria consistent with UK clinical guidelines²⁰: Mild AD severity (MMSE 21–26 points), Moderate AD severity (MMSE 15–20 points), and Moderately severe/severe AD severity (MMSE <15 points). The 6-month repeated administration of MMSE was used to determine cognitive decline in the cohort over the 18- or 36-month follow-up period.

Education was measured by the total number of years of formal education. We developed a dichotomous education variable, with 0 (low education) representing less than 12 years of formal education and 1 (high education) representing 12 or more years of formal education. A cutoff of 12 years was used in line with the expected attainment across countries of upper secondary education. This cutoff was suggested for multinational analyses where question design prohibited more detailed evaluation of early education and education quality.²¹

Baseline comorbidities were measured as the number of the co-occurring disease diagnoses among the following: cancer, depression, diabetes mellitus, hypercholesterolemia, hypertension, ischemic cardiac events, obstructive pulmonary disease, and urinary tract disorder. AD treatment was categorized at the visit level as follows: (1) no medication taken, (2) acetylcholinesterase (AChEI) inhibitor only, (3) Memantine only, and (4) both AChEI and Memantine taken. Comorbidities and treatment type were recorded at each study visit using an electronic clinical report form.

Statistical analysis

All analyses were conducted using R version 4.3.1 via the AD Data Initiative's AD Workbench. We summarized the descriptive statistics for participants’ characteristics, including baseline AD severity, country, self-reported sex, years of education, age, time from formal AD diagnosis, comorbidities, treatment type, and baseline MMSE, and examined their distributions across the two education levels by using t-tests or chi-squared tests.

In our primary analysis, we evaluated the relationship between education and rate of cognitive decline by using three linear mixed regression models. Each model included a random intercept (to account for individual variability in baseline cognition), a random slope (to account for individual variability in trajectory of cognitive decline), education, visit number, and an interaction term (i.e., education × visit number). The interaction term was used to evaluate the impact of education on the speed or rate of cognitive decline. Model 1 was unadjusted. Models 2 and 3 were adjusted for fixed effects of baseline participant characteristics, including age, time from formal diagnosis, study country, and sex, as well as time-varying covariates such as the number of relevant comorbidities and AD treatment type. Model 2 was applied to the full sample; while Model 3 was restricted to participants who were followed for more than 18 months, which represented a subset of the Germany and France participants. This restriction was used to assess the impact of follow-up duration.

To further account for the distinct demographic composition of the high versus low education groups, a sensitivity analysis was conducted with the full sample using propensity score matching on age, time from formal diagnosis, sex, baseline MMSE score, and country. Three additional sensitivity analyses were conducted by (1) using the continuous education variable instead of the dichotomized (high/low) education variable in Models 1–3 used for the primary analysis; (2) including a non-linear time effect represented by a quadratic time variable visit number ^ 2 into Model 3 (adjusted, restricted to participants who were followed for more than 18 months); and (3) examining the interactions between education and country, as well as education and sex, separately by including a three-way interaction term (Education × Country × Visit number or Education × Sex × Visit number) into Model 2 (adjusted, full sample).

We conducted three secondary analyses by repeating all analyses on subgroups stratified by (1) baseline AD severity, as categorized by UK clinical guidelines,²⁰ (2) country, and (3) sex. We stratified by AD severity to evaluate how the relationship between education and rate of cognitive decline varies by disease severity. We stratified by country to understand the extent to which the independent educational systems and distributions of educational attainment influence the relationship between education and cognitive decline. We stratified by sex to understand differences given previously characterized differences by sex in educational attainment and cognitive decline.^14,22,23

The R nlme package was used to conduct the mixed effects modeling, and the MatchIt and lme4 packages were used for propensity score matching and modeling matched data, respectively.

Results

The GERAS-EU dataset included data from 1529 participants, of whom 1313 were eligible for this study. On average, each eligible participant had 4.6 visits (SD: 1.7, range: 2 to 7).

The participants’ characteristics (e.g., demographics, comorbidities, and cognitive performance) were summarized in Table 1, stratified by the level of education attainment. Mean years of education were 8.8 (SD: 1.7, range: 0 to 11) in the low education group and 14.3 (SD: 2.4, range: 12 to 28) in the high education group. More details on the distribution of educational attainment by country were provided in Supplemental Figure 1. The low education group, compared with the high education group, had more participants living with moderate or severe AD at baseline (p < 0.001), more females (57.5% versus 46.8%, p <0.001), were older (78.0 versus 75.5 years of age, p < 0.001), had more comorbidities (1.52 versus 1.26, p < 0.001), and had lower baseline MMSE scores (17.4 versus 18.9, p < 0.001). Baseline MMSE scores were not significantly different across the education groups within each baseline AD severity level. When stratified by study country, a significant difference in baseline MMSE score was present (France: p = 0.01, Germany: p = 0.02, United Kingdom: p = 0.04), with higher mean baseline MMSE in the high education group. The majority of participants used some treatment at baseline, with the most frequent treatment being AChEI (65.4%). Baseline treatment type did not vary significantly across educational attainment strata (Table 1) but showed significant difference across countries (p < 0.001; detailed in Supplemental Table 1).

Table 1.

Distribution of key demographic, health, and cognitive performance variables stratified by educational attainment in the GERAS-EU study.

	Overall (N = 1313)	<12 Years (N = 935)	≥12 Years (N = 378)	p
Baseline AD Severity, n (%)
Mild	517 (39.4)	327 (35.0)	190 (50.3)	<0.001*
Moderate	427 (32.5)	323 (34.5)	104 (27.5)
Severe	369 (28.1)	285 (30.5)	84 (22.2)
Country, n (%)
Germany	476 (36.3)	331 (35.4)	145 (38.4)	0.02*
France	390 (29.7)	298 (31.9)	92 (24.3)
United Kingdom	447 (34.0)	306 (32.7)	141 (37.3)
Sex, n (%)
Female	715 (54.5)	538 (57.5)	177 (46.8)	<0.001*
Male	598 (45.5)	397 (42.5)	201 (53.2)
Age (y), mean (std)	77.3 (7.62)	78.0 (7.31)	75.5 (8.08)	<0.001*
Time From Formal Diagnosis (y), mean (std)	2.23 (2.23)	2.18 (2.25)	2.37 (2.18)	0.17
Baseline AD Treatment Type, n (%)
No treatment	172 (13.1%)	132 (14.1%)	40 (10.6%)	0.09
AChEI	859 (65.4%)	611 (65.3%)	248 (65.6%)
Memantine	132 (10.1%)	96 (10.3%)	36 (9.5%)
AchEI + Memantine	150 (11.4%)	96 (10.3%)	54 (14.3%)
Number of comorbidities, mean (std)	1.45 (1.22)	1.52 (1.22)	1.26 (1.20)	<0.001*
Baseline MMSE Score, mean (std)
All	17.9 (5.91)	17.4 (5.83)	18.9 (5.99)	<0.001*
By severity level
Mild	23.3 (1.63)	23.2 (1.60)	23.5 (1.66)	0.07
Moderate	18.0 (1.65)	18.0 (1.66)	18.0 (1.62)	0.91
Severe	10.0 (3.77)	10.1 (3.60)	9.70 (4.28)	0.35
By country
France	17.4 (5.56)	17.0 (5.29)	18.7 (6.22)	0.01*
Germany	18.3 (6.15)	17.8 (6.22)	19.3 (5.87)	0.02*
United Kingdom	17.8 (5.93)	17.4 (5.87)	18.7 (5.99)	0.04*

*Indicates statistically significant (p < 0.05). p-values were calculated by using t-tests or chi-squared tests.

As shown in Table 2 (see full results in Supplemental Table 2), MMSE scores declined on average 1.11 points per each visit (p < 0.001) in analysis of the full sample (Model 2). The analysis of interaction between education and visit number indicated a significant relationship between education and the rate of decline, with the high education group having 0.19 points more decline per visit than the low education group (p = 0.02; Model 2). Restricting to participants followed for more than 18 months revealed a similar significant relationship, with the high education group having 0.22 points more decline per visit than the low education group (p = 0.02; Model 3). Unadjusted MMSE scores over time, stratified by country and level of educational attainment, were presented in Figure 1 for participants who completed all visits specified by the country-specific protocols (i.e., 4 visits over 18 months in the United Kingdom and 7 visits over 36 months in Germany and France). Unadjusted mean scores by country and educational attainment, overlaid with individual MMSE trajectories, were presented in Supplemental Figure 2.

Figure 1.

MMSE scores for participants who completed all visits, stratified by country. Unadjusted MMSE scores over time were presented for participants who completed all visits specified by the country-specific protocols (i.e., 4 visits over 18 months in the United Kingdom and 7 visits over 36 months in Germany and France). This restriction ensured that a consistent sample was used to estimate mean scores over time. Scores were stratified by country: (a) Germany, (b) France, and (c) the United Kingdom.

Table 2.

Association between education and cognitive decline.^a

	Model 1^b(N = 1313)		Model 2^c(N = 1313)		Model 3: participants with extended follow-up^c^,^d(N = 545)
Variable	Estimate(95% CI)	p	Estimate(95% CI)	p	Estimate(95% CI)	p
High versus Low Education	1.68(0.94, 2.41)	<0.001*	1.82(1.13, 2.52)	<0.001*	1.41(0.42, 2.40)	0.91
Visit Number	−1.12(−1.21, −1.03)	<0.001*	−1.11(−1.20, −1.02)	<0.001*	−0.95(−1.05, −0.85)	<0.001*
High (versus Low) Education × Visit Number	−0.20(−0.36, −0.03)	0.02*	−0.19(−0.35, −0.03)	0.02*	−0.22(−0.41, −0.03)	0.02*

Unadjusted.

Adjusted for age at baseline, self-reported sex, country, time from formal diagnosis, time-varying number of comorbidities, and time-varying AD treatment type.

Restricted to participants who were followed for more than 18 months, which represented a subset of the Germany and France participants.

*Indicates statistically significant (p < 0.05).

Results for the sensitivity analysis using propensity-score matched sample were compatible with these findings (Supplemental Table 3), although the association between the education and the rate of cognitive decline was only significant in the nearest neighbor matching analysis (p = 0.04). Results using the continuous education variable (Supplemental Table 4) were consistent with results using the dichotomized education variable. Specifically, each additional year of education was associated with 0.03-point greater decline in MMSE score per visit (p = 0.01). Extending model 2 from the primary analysis by accounting for the non-linear time effect through a quadratic time variable (Visit Number ^ 2) slightly improved model fit and yielded results consistent with model 2 (Supplemental Table 5).

In secondary analyses stratified by baseline disease severity (mild versus moderate versus severe AD), we observed a positive association between higher educational attainment and faster cognitive decline for the severe AD group only (Table 3). Compared to the low education group, participants in the high education group experienced a 0.42-point greater decline per visit (p = 0.01). The associations between education and cognitive decline rates were not significant for the mild and moderate AD groups, although the direction of the relationship was consistent. In sex-stratified and country-stratified analyses, the association was in the expected direction but did not reach statistical significance for each subgroup. There was no significant interaction among education, cognitive decline, and country (Supplemental Table 6) or among education, cognitive decline, and sex (Supplemental Table 7).

Table 3.

Association between education and cognitive decline, stratified by baseline AD severity, sex and country.^a

	Mild AD(N = 517)		Moderate AD(N = 427)		Severe AD(N = 369)
	Estimate(95% CI)	p	Estimate(95% CI)	p	Estimate(95% CI)	p
High versus Low Education	0.37(−0.08, 0.82)	0.11	0.29(−0.33, 0.90)	0.36	0.80(−0.25, 1.85)	0.14
Visit Number	−1.05(−1.20, −0.90)	<0.001*	−1.21(−1.37, −1.06)	<0.001*	−1.04(−1.20, −0.89)	<0.001*
High (versus Low) Education × Visit Number	−0.18(−0.43, 0.07)	0.16	−0.13(−0.44, 0.17)	0.39	−0.42(−0.74, −0.09)	0.01*

	Male(N = 598)		Female(N = 715)
	Estimate(95% CI)	p	Estimate(95% CI)	p
High versus Low Education	2.23(1.25, 3.20)	<0.001*	1.50(0.50, 2.50)	<0.001*
Visit Number	−1.10(−1.25, −0.95)	<0.001*	−1.12(−1.23, −1.01)	<0.001*
High (versus Low) Education × Visit Number	−0.24(−0.48, 0.01)	0.06	−0.16(−0.38, 0.06)	0.17

	Germany(N = 476)		France(N = 390)		United Kingdom(N = 447)
	Estimate(95% CI)	p	Estimate(95% CI)	p	Estimate(95% CI)	p
High versus Low Education	1.85(0.68, 3.02)	<0.001*	1.77(0.47, 3.07)	0.01	1.81(0.64, 2.99)	<0.001*
Visit Number	−1.08(−1.23, −0.93)	<0.001*	−1.07(−1.20, −0.95)	<0.001*	−1.19(−1.39, −0.99)	<0.001*
High (versus Low) Education × Visit Number	−0.25(−0.50, 0.01)	0.06	−0.07(−0.33, 0.18)	0.57	−0.22(−0.56, 0.13)	0.22

The relationship between education and cognitive decline was assessed by linear mixed effects models. Cognitive decline was measured by the decline in the MMSE total score at each follow-up visit compared with the baseline MMSE score. The high education group refers to 12 or more years of formal education and the low education group refers to less than 12 years of formal education. All analyses were adjusted for age at baseline, time from formal diagnosis, time-varying number of comorbidities, and time-varying AD treatment type. Analyses stratified by baseline AD severity level were additionally adjusted for country and self-reported sex. Analyses stratified by sex were additionally adjusted for country. Analyses stratified by country were additionally adjusted for sex.

*Indicates statistically significant (p < 0.05).

Discussion

Using a large multinational European sample, we found that a high level of education was associated with faster cognitive decline in individuals with clinically diagnosed AD. Analyses restricted to participants followed for more than 18 months, as well as sensitivity analyses applying propensity score matching and using an alternative measure of educational attainment (i.e., years of education), yielded compatible results. The relationship between education and rate of cognitive decline was only significant in the severe AD group after stratification by baseline disease severity, although the direction of the relationship remained consistent across all disease stages, sexes, and countries in the GERAS-EU sample. Below, we discuss our findings and the implications of these findings in the context of CR and early detection of AD.

The CR theory attributes greater tolerance of AD-related pathology, without evident functional decline, to protective socio-behavioral factors (which are proxies of CR) accrued throughout the life course.¹ Education is among the most widely used socio-behavioral proxies for CR^4,5 and has been associated with a delayed onset of AD-related symptoms.^24–26 The CR theory attributes this relationship to the adaptability of cognitive processes rather than to preserved neurobiological capital.^1,2 As such, the CR hypothesis suggests that cognitive decline occurs more rapidly in those with greater levels of CR after the symptoms present.^6,7 The association between higher levels of education and an increased rate of cognitive decline in individuals with probable AD has been demonstrated in previous U.S.-based observational studies^8–12; however, findings from international studies remain inconclusive.^6,13–17 A previous study¹⁷ followed 53 Spanish older adults with dementia (35 diagnosed with AD) over three years, assessing cognitive function with the 37-item Spanish version of the MMSE (MMSE-37) at baseline and follow-up. The study found that MMSE-37 scores declined by 3.34 ± 4.98 points among low-educated participants (illiterate or basic literacy) compared to 7.90 ± 4.88 points among high-educated participants (primary school certificate or higher) (p = 0.02).¹⁷ A France-based study⁶ followed 215 participants with prodromal AD over 9 years, assessing cognitive performance at five time points using four neuropsychological (NP) measures, including a French version of the MMSE. The study found that participants with higher educational levels experienced faster declines across all NP measures during the 3–4 years preceding AD diagnosis.⁶ On the other hand, studies conducted in Sweden²⁷ and Brazil^13–16 did not find significant associations between higher level of education and faster cognitive decline following AD diagnosis. Furthermore, most international studies relied on relatively small sample size (50–220 participants)^6,13–17 or were not exclusively focused on AD.^17,27 Our study contributes to this literature by leveraging a large, multinational, community-based AD cohort. We have findings compatible with the U.S.-based studies^8–12 and two studies conducted in the European countries.^6,17 Notably, both European studies defined high versus low educational attainment based on whether participants had obtained a primary school diploma.^6,17 In contrast, 75% of participants in our study had seven or more years of education, and we defined high versus low education based on completing 12 years of schooling. It is worth noting that two studies involving approximately 200 Brazilian patients with late-onset AD and low average education (<5 years) found protective effects of education.^13,14 Specifically, more years of education were associated with a longer time from dementia onset to reaching a low MMSE score of 20, although significant effects were observed only among females and APOE ε4 carriers in stratified analyses.^13,14 Our study did not find an interaction effect among education, cognitive decline, and sex (Supplemental Table 7), although sex-stratified analyses suggested a stronger association (e.g., a larger magnitude of beta coefficient and a smaller p value) between high education and faster cognitive decline in males (Table 3). One strength of the Brazil-based studies was including APOE ε4 status in their analyses; while this information was not available to our study. This difference, along with factors such as variations in study design (e.g., exclusion criteria such as whether individuals with Lewy-body disease were excluded), data analysis methods, and overall mean education level, may contribute to the differing findings. Future studies should aim to gather more evidence from non-U.S. and non-European countries to enhance global understanding of the relationship between CR and cognitive decline in individuals with AD. In addition, although this study is multi-national, the participants had relatively high levels of educational attainment. Further research with larger samples encompassing a broader range of educational backgrounds is needed to better contextualize the relationship between education and accelerated cognitive decline following AD onset.

As previously noted, differences in the average educational levels of study participants may partially account for the variation in findings across international studies. Moreover, even among studies involving similarly educated populations, establishing consistent measures of education across countries remains challenging. Because the education systems and the influence of education on other social determinants of health (e.g., occupational quality and income) can be different across countries, we conducted country-stratified analyses to evaluate the consistency of findings given the distinct national characteristics. We found that the associations between the education level and the cognitive decline rate varied in strength across countries. This finding may be partially driven by the differences in educational systems. Previous work has demonstrated that the positive impact of education on cognitive performance varies by country.²⁸ A U.S. study has also shown that lower system-level (state-level) educational quality was associated with higher risk of dementia.²⁹ In addition, a deeper understanding of the relationship between education and AD development and cognitive decline requires granular information such as early childhood education, degree attainment, and ongoing informal education.³⁰ Both early-life education and later-life educational attainment have demonstrated relationships with cognitive function.^31,32 These expanded measures, along with advanced methods for harmonizing education variables across countries,³³ would be useful in analyses spanning multiple countries in the future. Additionally, educational attainment is partially a product of parental education and income, and contributes to occupational complexity, income, health-related knowledge, service use, and social support access.³⁴ Further understanding how these factors may moderate or mediate the relationship between education and the rate of cognitive decline is important for having a comprehensive view of how education impacts AD development and cognitive decline.

In the severity-stratified analyses, we observed a significant association between education and cognitive decline rate in only the severe AD group. Meanwhile, we observe that the estimated beta coefficient for this relationship in the mild AD group (i.e., −0.18) was similar as that in the primary analyses (i.e., −0.19 in Model 2), but with greater variability as demonstrated by a wider 95% confidence interval (i.e., −0.43 to 0.07). The MMSE test has both floor and ceiling effects which may lead to reduced ability to identify cognitive decline at both the early and late stages of disease.³⁵ The MMSE's floor effect may have contributed to the large variation in measuring cognitive decline, leading to the wider 95% confidence interval of the estimated association between education and cognitive decline (i.e., −0.74 to −0.09) observed in the severe AD group. The ceiling effect of the MMSE may limit detection of the early-stage cognitive decline (e.g., transition from normal MMSE to mild impairment MMSE) among individuals with high education.³⁶ In other words, the MMSE may only be able to detect the onset of cognitive impairment in more highly educated individuals at a point when the individual being tested has been further along the disease continuum. This ceiling effect is closely related to CR and also has been seen in other NP tests.³⁷ The limitation of MMSE and other NP tests in early AD detection and diagnosis is compatible with our observation that highly educated individuals appeared to experience faster cognitive decline than individuals with lower level of education during the mild AD stage, although this relationship was not statistically significant. If individuals with high education are in fact at a later disease stage (e.g., having greater AD pathological burden), compared to those with low education when they are first clinically identified as having mild AD, they would be expected to have a steeper cognitive decline immediately after clinical diagnosis. On the other hand, AD treatments tend to be more effective in the early stages of AD.³⁸ In this study, the majority of participants (86.9%) were receiving some form of treatment at baseline, and the proportion of individuals not using any treatment was lower in the high education group compared to the low education group (10.6% versus 14.1%). This differential treatment usage may have weakened the association between education and cognitive decline after AD diagnosis, potentially contributing to the nonsignificant results observed in the mild and moderate AD groups.

Despite the limitations of MMSE in measuring cognitive impairment, we used it in this study due to data availability constraints. MMSE has been used in prior studies to assess the relationship between education and cognitive decline^{6,12,13,15,17} or track progression of cognitive impairment.^39,40 According to the existing literature, a change of 1.4 or more points in MMSE is considered clinically significant.^41,42 The main effect of time observed in this study (2.2 per year) matched this threshold of significance. The interaction effect between time and education we identified was small (0.38–0.40 per year for the full sample; 0.84 per year for severe AD group) and did not reach clinical significance, which is expected since interaction effects are typically smaller than main effects. Additionally, most participants received AD treatment during the study, which may have attenuated the influence of other factors, including education. Future studies utilizing more sensitive cognitive assessments may offer greater insight into the clinical significance of differential cognitive decline across education levels.

Our findings also highlight the importance of sensitive screening for early, subtle signs of cognitive impairment. As aforementioned, one of the driving factors for the steeper cognitive decline observed in the high education group may be the delayed recognition of symptoms by the most common cognitive assessments such as the MMSE. The MMSE is a frequently used tool for screening participants for diagnosis and inclusion in AD clinical trials, but it is susceptible to missing early signs of cognitive impairment due to ceiling effects in those with higher-level education, which is often the most represented education group in many trials.^35,43 Attempts to address the shortcomings in sensitivity of the MMSE in highly educated individuals include adjusting the cutoff score of the assessment, e.g., increasing the cut-off score for individuals with high education.⁴⁴ However, MMSE testing with such adjustments may lead to more false positive identification in individuals with high education.⁴⁵ Standardized NP tests that assess multiple cognitive functions, such as memory, attention, processing speed, reasoning, and language comprehensively also provide a greater level of sensitivity.⁴⁶ However, this approach is hard to scale for a wide population for AD surveillance for several reasons. First, many cognitive assessment methods were developed within the U.S. and Europe and are subject to educational, language, and/or cultural bias⁴⁷ and therefore are not readily appropriate for other countries. Second, the administration of these NP tests requires trained examiners, which may not be available and/or is costly for continual monitoring of cognitive decline in people with high risk of AD. Third, the NP tests can be time-consuming, reducing people's interest in undergoing routine assessment. Application of self-administered cognitive tests using mobile devices may provide a greater opportunity for disease detection at early disease stages.⁴⁸ The recent growth of the digital health and wearable technology market has introduced tools capable of collecting large volumes of data across various cognitive and functional domains, thereby increasing opportunities to detect changes at currently asymptomatic stages.⁴⁹

Several limitations should be noted in our analysis. First, same as other observational studies, the relationship between education and cognitive decline we found in this study is not causal and may be affected by unmeasured or residual confounding. We adjusted for unbalanced baseline characteristics and conducted a sensitivity analysis with a propensity score matched sample and found similar results. Our analysis is limited, however, in the inability to account for a broad range of covariates, which were not collected as part of the GERAS-EU study, and which have been shown to influence CR effects. The lack of information about race and ethnicity for this sample is a key limitation, as structural factors contribute to differences in both access to education and rates of decline.⁵⁰ GERAS-EU also did not capture information about APOE-ε4 carrier status which may have modified the effect of education on decline rates. Second, we used MMSE total scores as our outcome measure. The MMSE composite score does not allow assessment of specific cognitive domains (e.g., memory, executive function, etc.) that may be driving early cognitive decline. In addition, it is less sensitive in detecting cognitive decline compared to more comprehensive NP tests. The categorization of AD severity and the clinical implications of our findings are also restricted by our reliance on the MMSE. To better understand how the relationship between education and cognitive decline after AD onset varies by disease stage, future studies will need to incorporate separate markers of disease severity and cognitive decline. Third, a more in-depth description of education can provide detailed information that helps to interpret the results from the country-stratified analyses. We used a threshold of 12 years of education to distinguish between high versus low education levels, an approach in line with prior studies.²¹ Granular information (e.g., early childhood and life-long education attainment) is needed to improve data harmonization across countries and compare impacts of education on cognitive decline across countries in the future. In addition, the relatively high average education level of our study participants (8.8 years even within the low-education group) may limit the generalizability of our findings to populations with lower overall levels of education. Lastly, we share results in the context of CR. Our results extend the work of previous analyses evaluating how rates of cognitive decline vary after the point of symptom onset based on education as a proxy for CR. We are unable, however, to assess the impact of education on disease onset in this cohort given that participants were enrolled after disease diagnosis. Further, we lack any measure of pathological burden. Despite these caveats, the strengths of this analysis in characterizing longitudinal decline in a multi-national, large European sample represent a contribution to the literature and our findings further underscore the importance of developing and utilizing innovative monitoring methods to support early detection of AD at a time when interventions can be most effective.

Conclusions

Research on the relationship between education and cognitive decline after AD diagnosis remains limited. Our study utilized a large, multi-national sample to demonstrate that high education is associated with accelerated cognitive decline after diagnosis, highlighting the importance of sensitive screening of early signs of cognitive impairment. Country-specific results showed a consistent direction in the association, although they were not statistically significant. Taken together, these findings partially support the CR theory. Future research should investigate potential influencing factors (e.g., sociocultural differences, healthcare access, and educational systems) and employ more sensitive measures of cognitive decline to better understand the variability of CR effect across diverse populations.

Supplemental Material

sj-docx-1-alz-10.1177_13872877251352216 - Supplemental material for Association between education and rate of cognitive decline among individuals with Alzheimer's disease: A multi-national European observational study

Supplemental material, sj-docx-1-alz-10.1177_13872877251352216 for Association between education and rate of cognitive decline among individuals with Alzheimer's disease: A multi-national European observational study by Zachary T Popp, Ting Fang Alvin Ang, Phillip H Hwang, Rhoda Au and Jinying Chen in Journal of Alzheimer's Disease

Footnotes

Acknowledgements

Data discovery services and computational resources contributing to this work were provided in-kind by the AD Data Initiative []. The authors would like to acknowledge participants in the GERAS-EU study for contributing the time and data that allows for this analysis to be conducted. The authors also would like to acknowledge the sponsor of the GERAS-EU study for making this data available upon request through AD Data Initiative's AD Workbench for analysis.

ORCID iDs

Zachary T Popp

Ting Fang Alvin Ang

Phillip H Hwang

Rhoda Au

Jinying Chen

Ethical considerations

This study was conducted as a secondary data analysis of deidentified data made available through the AD Data Initiative's AD workbench and was exempt from IRB review.

Consent to participate

Not applicable.

Consent for publication

Not applicable.

Author contributions

Zachary T Popp: Data curation; Formal analysis; Investigation; Methodology; Writing – original draft; Writing – review & editing.

Ting Fang Alvin Ang:(Funding acquisition; Investigation; Supervision; Writing – review & editing.

Phillip H Hwang: Investigation; Writing – review & editing.

Rhoda Au: Funding acquisition; Investigation; Writing – review & editing.

Jinying Chen: Formal analysis; Investigation; Methodology; Supervision; Validation; Writing – original draft; Writing – review & editing.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study is supported by AD Data Initiative. JC was also supported by a 2023 pilot award from the Framingham Heart Study Brain Aging Program, which is funded by the National Institute on Aging (U19-AG068753), which also provides support to RA, AA. The American Heart Association (20SFRN35360180) provided funding support to RA and ZP. PH is supported by the National Institute on Aging (5K01AG080119) and the Alzheimer's Association (23AARF-1019048). The funders had no decisional role in study design, data collection and analysis, interpretation of data, or preparation of the manuscript.

Declaration of conflicting interests

The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: RA is a scientific advisor to Signant Health and NovoNordisk. None of these potential competing interests overlap in any way with the content of the current study.

Data availability statement

GERAS-EU data can be requested through the AD Data Initiative's AD Workbench.

Supplemental material

Supplemental material for this article is available online.

References

Stern

. Cognitive reserve in ageing and Alzheimer’s disease. Lancet Neurol 2012; 11: 1006–1012.

Stern

Arenaza-Urquijo

Bartrés-Faz

, et al. Whitepaper: defining and investigating cognitive reserve, brain reserve, and brain maintenance. Alzheimers Dement 2020; 16: 1305–1311.

Van Loenhoud

Van Der Flier

Wink

, et al. Cognitive reserve and clinical progression in Alzheimer disease: a paradoxical relationship. Neurology 2019; 93: E334–E346.

Sohrabi

Weinborn

Shemehsavar

, et al.

Education or socioeconomic status: which proxy of cognitive reserve associates better with cognitive function and amyloid imaging results?

Alzheimers Dement 2023; 19: e079768.

Pinto

Peixoto

Dores

, et al. Measures of cognitive reserve: an umbrella review. Clinical Neuropsychol 2024; 38: 42–115.

Amieva

Jacqmin-Gadda

Orgogozo

, et al. The 9 year cognitive decline before dementia of the Alzheimer type: a prospective population-based study. Brain 2005; 128: 1093–1101.

Clouston

SAP

Smith

Mukherjee

, et al. Education and cognitive decline: an integrative analysis of global longitudinal studies of cognitive aging. J Gerontol B Psychol Sci Soc Sci 2020; 75: E151–E160.

Wilson

Aggarwal

, et al. Education and the course of cognitive decline in Alzheimer disease. Neurology 2004; 63: 1198–1202.

Rasmusson

Carson

Brookmeyer

, et al. Predicting rate of cognitive decline in probable Alzheimer’s disease. Brain Cogn 1996; 31: 133–147.

10.

Teri

McCurry

Edland

, et al. Cognitive decline in Alzheimer’s disease: a longitudinal investigation of risk factors for accelerated decline. J Gerontol A Biol Sci Med Sci 1995; 50A: M49–M55.

11.

Scarmeas

Albert

Manly

, et al. Education and rates of cognitive decline in incident Alzheimer’s disease. J Neurol Neurosurg Psychiatry 2006; 77: 308–316.

12.

Andel

Vigen

Mack

, et al. The effect of education and occupational complexity on rate of cognitive decline in Alzheimer’s patients. J Int Neuropsychol Soc 2006; 12: 147–152.

13.

De Oliveira

De Almeida

Chen

, et al. Lifetime risk factors for functional and cognitive outcomes in patients with Alzheimer’s disease. J Alzheimers Dis 2018; 65: 1283–1299.

14.

Oliveira

Chen

Smith

, et al. Predictors of cognitive and functional decline in patients with Alzheimer disease dementia from Brazil. Alzheimer Dis Assoc Disord 2016; 30: 243–250.

15.

de Oliveira

Pereira

Pivi

GAK

, et al. Effects of APOE haplotypes and measures of cardiovascular risk over gender-dependent cognitive and functional changes in one year in Alzheimer’s disease. Int J Neurosci 2018; 128: 472–476.

16.

De Oliveira

Pivi

GAK

Chen

, et al. Risk factors for cognitive and functional change in one year in patients with Alzheimer’s disease dementia from São Paulo, Brazil. J Neurol Sci 2015; 359: 127–132.

17.

Contador

Bermejo-Pareja

Pablos

, et al. High education accelerates cognitive decline in dementia: a brief report from the population-based NEDICES cohort. Dement Neuropsychol 2017; 11: 297–300.

18.

Wimo

Reed

Dodel

, et al. The GERAS study: a prospective observational study of costs and resource use in community dwellers with Alzheimer’s disease in three European countries – study design and baseline findings. J Alzheimers Dis 2013; 36: 385–399.

19.

McKhann

Drachman

Folstein

, et al. Clinical diagnosis of Alzheimer’s disease: report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer’s disease. Neurology 1984; 34: 939–944.

20.

National Institute for Health and Clinical Excellence. Donepezil, galantamine, rivastigmine and memantine for the treatment of Alzheimer’s disease (NICE Technology Appraisal 217). London: NICE, https://www.nice.org.uk/guidance/ta217 (2011).

21.

Schneider

Wysmułek

Kwak

, et al. The classification of education in surveys: a generalized framework for ex-post harmonization. Qual Quant 2022; 56: 1829–1866.

22.

Levine

Gross

Briceño

, et al. Sex differences in cognitive decline among US adults. JAMA Netw Open 2021; 4: e210169.

23.

Zeki Al Hazzouri

Haan

Neuhaus

, et al. Cardiovascular risk score, cognitive decline, and dementia in older Mexican Americans: the role of sex and education. J Am Heart Assoc 2013; 2: e004978.

24.

Almeida

Schultz

Austin

, et al. Effect of cognitive reserve on age-related changes in cerebrospinal fluid biomarkers of Alzheimer disease. JAMA Neurol 2015; 72: 699–706.

25.

Langa

Larson

Crimmins

, et al. A comparison of the prevalence of dementia in the United States in 2000 and 2012. JAMA Intern Med 2017; 177: 51–58.

26.

Zhu

, et al. The protective impact of education on brain structure and function in Alzheimer’s disease. BMC Neurol 2021; 21: 423.

27.

Agüero-Torres

Fratiglioni

Guo

, et al. Prognostic factors in very old demented adults: a seven-year follow-up from a population-based survey in Stockholm. J Am Geriatr Soc 1998; 46: 444–452.

28.

Rodriguez

Hofbauer

Röhr

. The role of education and income for cognitive functioning in old age: a cross-country comparison. Int J Geriatr Psychiatry 2021; 36: 1908–1921.

29.

Soh

Whitmer

Mayeda

, et al. State-level indicators of childhood educational quality and incident dementia in older Black and White adults. JAMA Neurol 2023; 80: 352.

30.

Langella

Barksdale

Vasquez

, et al. Effect of apolipoprotein genotype and educational attainment on cognitive function in autosomal dominant Alzheimer’s disease. Nat Commun 2023; 14: 5120.

31.

Lövdén

Fratiglioni

Glymour

, et al. Education and cognitive functioning across the life span. Psychol Sci Public Interest 2020; 21: 6–41.

32.

Soh

Whitmer

Mayeda

, et al. Timing and level of educational attainment and late-life cognition in the KHANDLE study. Alzheimers Dement 2024; 20: 593–600.

33.

Schröder

Ganzeboom

HBG

. Measuring and modelling level of education in European societies. Eur Sociol Rev 2014; 30: 119–136.

34.

Khalatbari-Soltani

Maccora

Blyth

, et al. Measuring education in the context of health inequalities. Int J Epidemiol 2022; 51: 701–708.

35.

Arevalo-Rodriguez

Smailagic

Roqué-Figuls

, et al. Mini-mental state examination (MMSE) for the early detection of dementia in people with mild cognitive impairment (MCI). Cochrane Database Syst Rev 2021; 7: CD010783.

36.

Robitaille

van den Hout

Machado

RJM

, et al. Transitions across cognitive states and death among older adults in relation to education: a multi-state survival model using data from six longitudinal studies HHS public access. Alzheimers Dement 2018; 14: 462–472.

37.

Alenius

Ngandu

Koskinen

, et al. Education-based cutoffs for cognitive screening of Alzheimer’s disease. Dement Geriatr Cogn Disord 2022; 51: 42–55.

38.

Huang

Kuan

Lin

, et al. Clinical trials of new drugs for Alzheimer disease: a 2020-2023 update. J Biomed Sci 2023; 30: 83.

39.

Mattsson-Carlgren

Salvadó

Ashton

, et al. Prediction of longitudinal cognitive decline in preclinical Alzheimer disease using plasma biomarkers. JAMA Neurol 2023; 80: 360–369.

40.

Macdougall

Whitfield

Needham

, et al. Predicting progression to Alzheimer’s disease dementia using cognitive measures. Int J Geriatr Psychiatry 2024; 39: e6067.

41.

Borland

Edgar

Stomrud

, et al. Clinically relevant changes for cognitive outcomes in preclinical and prodromal cognitive stages: implications for clinical Alzheimer trials. Neurology 2022; 99: E1142–E1153.

42.

Watt

Veroniki

Tricco

, et al. Using a distribution-based approach and systematic review methods to derive minimum clinically important differences. BMC Med Res Methodol 2021; 21: 41.

43.

Chapman

Bing-Canar

Alosco

, et al. Mini mental state examination and logical memory scores for entry into Alzheimer’s disease trials. Alzheimers Res Ther 2016; 8: 9.

44.

Erdodi

Shahein

Fareez

, et al. Increasing the cutoff on the MMSE and DRS-2 improves clinical classification accuracy in highly educated older adults. Psychol Neurosci 2020; 13: 93–113.

45.

Franco-Marina

Juan García-González

Wagner-Echeagaray

, et al. The mini-mental state examination revisited: ceiling and floor effects after score adjustment for educational level in an aging Mexican population. Int Psychogeriatr 2010; 22: 72–81.

46.

Jak

Preis

Beiser

, et al. Neuropsychological criteria for mild cognitive impairment and dementia risk in the Framingham Heart Study. J Int Neuropsychol Soc 2016; 22: 937–943.

47.

Nielsen

. Cognitive assessment in culturally, linguistically, and educationally diverse older populations in Europe. Am J Alzheimers Dis Other Demen 2022; 37: 15333175221117006.

48.

Öhman

Hassenstab

Berron

, et al. Current advances in digital cognitive assessment for preclinical Alzheimer’s disease. Alzheimers Dement (Amst) 2021; 13: e12217.

49.

Popp

Low

Igwe

, et al. Shifting from active to passive monitoring of Alzheimer disease: the state of the research. J Am Heart Assoc 2024; 13: e031247.

50.

Avila

Rentería

Jones

, et al. Education differentially contributes to cognitive reserve across racial/ethnic groups. Alzheimers Dement 2021; 17: 70.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

1.09 MB