Education and age of decline in verbal memory across countries: Evidence from the HRS,MHAS,and CHARLS

Abstract

Background:

Despite the well-established association between education and cognition in old age, age-related changes in cognition by education across different contexts remain underexplored.

Objective:

This study investigates cross-national differences in decline in verbal memory among adults aged 50 and older in the United States, Mexico, and China.

Methods:

Data come from the Health and Retirement Study (HRS, 2012–2020), Mexican Health and Aging Study (MHAS, 2012–2021), and China Health and Retirement Longitudinal Study (CHARLS, 2011–2020). Multilevel models estimated changes in verbal memory across six age groups and six educational levels, adjusting for physical health, residence, and gender.

Results:

HRS participants had the highest educational attainment, while CHARLS participants had the lowest. Higher education was associated with better verbal memory, with the most pronounced education gradient in the CHARLS. Declines in verbal memory were steeper for MHAS and CHARLS participants over 60 compared to HRS participants. MHAS participants experienced earlier declines in verbal memory across all education levels compared to HRS and CHARLS participants.

Conclusions:

Middle-aged and older adults in Mexico experienced earlier declines in verbal memory after age 50 compared to those in China and the United States. Future research should seek to identify other potentially modifiable factors that may contribute to cross-national differences in decline in verbal memory.

Keywords

age Alzheimer's disease cross-national comparison decline in verbal memory education

Introduction

There is an urgent need to identify potentially modifiable risk factors for cognitive decline and dementia, especially in low- and middle-income countries (LMICs), where dementia is expected to place a disproportionately heavy burden on societies, families, and individuals.^1,2 While disease-modifying pharmaceutical therapies are becoming available in high-income countries, no cure currently exists for the progression of cognitive decline and dementia. Therefore, identifying potentially modifiable risk factors remains a priority.^3,4 Up to 40% of dementia cases worldwide may be attributable to common potentially modifiable risk factors.⁴ Among these factors, low educational attainment accounts for the largest proportion of dementia cases globally,⁵ particularly in LMICs.⁶

Higher educational attainment has consistently been associated with better late-life cognitive functioning and lower dementia prevalence in both high-income countries^7–9 and LMICs,^10–12 but the relationship between educational attainment and cognitive decline is less clear. While many studies report similar rates of cognitive decline according to educational attainment,^8,13 others have found that older adults with higher educational attainment experience less cognitive decline.¹⁴ Additionally, some research suggests that higher educational attainment could delay the onset of accelerated cognitive decline before dementia is diagnosed.^15,16 However, evidence also suggests that higher-educated older adults with dementia experience more rapid cognitive decline compared to older adults with lower levels of education.^17,18

Several mechanisms have been proposed to explain the contributions of educational attainment to older adults’ cognitive functioning. Perhaps the most widely studied is the contribution of educational attainment to cognitive reserve. Cognitive reserve refers to a person's ability to maintain high cognitive functioning despite age-related or pathological changes to the brain.¹⁹ Education is important to the development of cognitive reserve because it is typically acquired early in life and serves as a key predictor of later-life factors that contribute to cognitive reserve, such as occupation, physical activity, and social engagement.¹⁹

While many studies have examined the association between educational attainment and the age of onset of cognitive decline,^15,16,20,21 few have compared this association across different countries. Most cross-country comparisons have focused on the United States and Europe, or between European countries. Educational attainment is consistently linked to cognitive outcomes, but the magnitude of these associations varies by country.^9,22,23 Research comparing the United States and high-income European countries with LMICs, such as Mexico and China, has shown that older adults in high-income countries generally have higher cognitive functioning than older adults in LMICs.⁷ Additionally, an analysis of adults aged 65 and older in the United States, England, Mexico, China, and India found that variations in educational attainment accounted for 50% to 90% of the differences in cognitive functioning observed between these countries.²⁴

Despite extensive research on the association between educational attainment and late-life cognitive decline, and the increasing number of cross-country comparisons of this relationship, little is known about how cognitive trajectories compare across countries or whether the relationship between educational attainment and the age at which cognitive decline begins is consistent across countries. To address these gaps, we used data from the US Health and Retirement Study (HRS), the Mexican Health and Aging Study (MHAS), and the China Health and Retirement Longitudinal Study (CHARLS) to assess the relationship between age, education, and cognitive decline. We selected these countries because of their notably different educational systems and the diverse education reforms they have experienced. In Mexico, high school education only became mandatory in 1993.²⁵ In 1970, 75% of men and nearly 80% of women had not completed their primary education.²⁶ By 2000, these proportions had decreased to 21% and 23%, respectively, marking the first time the majority of the population had completed any form of education.²⁷ In China, the proportion of individuals with no education declined from 27.6% in 1980 to 3.5% in 2010, while those with middle education or higher more than doubled, from 7.4% to 15.5%.²⁸ These substantial changes in educational attainment are particularly relevant, as individuals with no education are more likely to experience cognitive decline in both Mexico^10,29,30 and China.^31,32

This study aims to (1) illustrate the age trajectories of verbal memory from middle to old age in the United States, Mexico, and China, and (2) examine the education gradient in these trajectories. We hypothesize that because the educational and demographic transitions in these three countries have differed, the association between education and the age of onset of decline in verbal memory will vary by country.

Methods

Data and sample

We used longitudinal data from the HRS, MHAS, and CHARLS databases. The HRS is a nationally representative cohort study of adults aged 50 and older in the United States, with biennial interviews conducted since 1992. The MHAS is a nationally representative cohort study of adults aged 50 and older in Mexico, which began in 2001. Follow-up interviews were completed in 2003, 2012, 2015, 2018, and 2021, with refresher cohorts added in 2012 and 2018 for participants born between 1952–1962 and 1963–1968, respectively. The CHARLS is a large, nationally representative panel study of adults aged 45 and older in China that started in 2011, with follow-up waves completed in 2013, 2015, 2018, and 2020. The study and survey designs of the MHAS and CHARLS are highly comparable to the HRS. We used four interview waves from each study: 2012, 2016, 2018, and 2020 from the HRS; 2012, 2015, 2018, and 2021 from the MHAS; and 2011, 2015, 2018, and 2020 from the CHARLS.

For each database, participants were included in the final analytical sample based on three criteria (Supplemental Figure 1). First, we selected participants who completed the survey independently, without assistance from a proxy. Second, we selected participants who were aged 50 or older at baseline (the 2012 wave for the HRS and MHAS, and the 2011 wave for the CHARLS). Third, we selected participants with complete information on demographic characteristics, socioeconomic status (education and income), health behaviors, health status, and cognitive function. The final sample sizes were 17,822 (HRS), 11,484 (MHAS), and 12,137 (CHARLS).

Supplemental Table 3 presents the characteristics of participants in the final sample compared to those with missing data, who were excluded. Missing cases in the HRS had comparatively lower standardized verbal memory scores (z = −0.007) and lower income. Missing cases in the MHAS primarily had lower standardized verbal memory scores (z = −0.74), no education or an incomplete elementary school education, and more IADL symptoms at baseline; they were also comparatively older (on average, five years older than those included in the sample, with missing cases mainly concentrated in the 76 + group). Missing cases in the CHARLS were predominantly individuals with a high school or college education, those older than 70, and urban residents.

Dependent variable

While the HRS, MHAS, and CHARLS were designed for cross-national comparisons, only the cognitive tests for verbal learning and verbal memory are comparable across all three studies. In each study, participants were randomly assigned one of two word lists. Verbal learning was measured by having participants immediately recall the word list, and verbal memory was measured by having them recall the word list after a delay. The word list contains ten words in both the HRS and CHARLS, and eight words in the MHAS. We used one round of word recall in each study. To enhance comparability across the studies, we standardized the final scores of verbal memory in each database and created a baseline-centered z-score. Specifically, we calculated participants’ z-scores using the sample mean and standard deviation at baseline, and then estimated their z-scores during the follow-up waves based on these baseline metrics.

Independent variables

We categorized age at baseline into six groups: 50–55, 56–60, 61–65, 66–70, 71–75, and 76 or older. Because this analysis investigated age-related changes in verbal memory, we used participants’ age at each follow-up observation as the measure of time. Thus, the baseline age group represents age cohort differences in verbal memory, whereas the continuous measure of age at each follow-up observation represents the change in verbal memory over time.

Education was measured by the highest degree attained, with six ordinal categories: no education, some elementary school, elementary school, middle school, high school, and college or professional degree education. Given the unbalanced distribution of education in each database, we set having a high school education as the reference group.

Covariates

Covariates included income, residence (rural or urban), health status, health behaviors, and demographic characteristics. Health status and health behaviors were obtained from participants’ baseline interviews. Income and residence were treated as time-variant covariates. To calculate income, we combined current wages, or income or earnings from self-employment, and pension earnings, and then categorized the final income into tertiles: low (1), medium (2), and high income (3). Residence was coded as urban (0) and rural (1) by dwelling places in the HRS and MHAS and household registration (hukou) status in the CHARLS.

During interviews, participants were asked whether a doctor had diagnosed them with heart disease, stroke, hypertension, or diabetes. We selected these four chronic conditions due to their relevance to cognitive function. These conditions were summed and recoded as dichotomous variables at baseline (no = 0, any = 1). We also included any limitations in instrumental activities of daily living (IADLs) at baseline (no = 0, any = 1). Health behaviors were measured based on engaging in energetic physical activities per week at baseline (no = 0, yes = 1), smoking at baseline (no = 0, yes = 1), and drinking alcohol at baseline (no = 0, yes = 1). Demographic characteristics included age, age-squared, and gender (male = 0, female = 1). As has been done previously,^33,34 we included an age-squared term to account for non-linear changes in verbal memory. Finally, properly specifying practice effects is important when studying cognitive trajectories, but they may not substantially influence the association between an exposure and rate of cognitive change.³⁵ Thus, we did not include a variable for practice effects in the regression models.

Statistical analysis

Descriptive results from the HRS, MHAS, and CHARLS are presented separately in Table 1. We used multilevel mixed models with random slopes to estimate the starting age of decline in verbal memory and the effects of education for each country. The time metric was chronological age at each wave. We tested the linearity of decline in verbal memory and age and found a non-linear change in verbal memory with age across all studies. The likelihood ratio test indicated that including the age-squared term improved the model fit. We also examined whether decline in verbal memory was dependent on baseline verbal memory but found no significant differences between models with and without baseline verbal memory as an independent variable. As a result, we excluded the baseline verbal memory scores from the analysis. All analyses were adjusted for income, rural residence, baseline IADLs, baseline chronic conditions, physical activity, age, age-squared, and gender. No variables in the final regression models had a variance inflation factor greater than 5. All data from the HRS, MHAS, and CHARLS are cleaned for extreme values before being made publicly available to outside investigators. Finally, the distribution of verbal memory test scores was right-skewed in the CHARLS sample (skewness = 0.66), but the residuals from the regression model were normally distributed, indicating that model assumptions were met.

Table 1.

Descriptive characteristics of participants aged 60 and older from the HRS, MHAS, and CHARLS.

Variables	United States (HRS)	Mexico (MHAS)	China (CHARLS)
Standardized verbal memory (range = −3 ∼ + 3)^a,b
Wave 1	0.004(0.997)	0.239(0.765)	−0.048(0.964)
Wave 2	0.004(0.998)	0.268(0.705)	−0.130(1.005)
Wave 3	0.092(0.973)	0.239(0.752)	−0.134(1.208)
Wave 4	0.108(1.037)	0.246(0.736)	0.117(1.174)
Education^c
High school(reference)	0.42	0.04	0.08
No education	0.01	0.16	0.32
Some elementary school	0.02	0.30	0.20
Elementary school	0.01	0.22	0.22
Middle school	0.05	0.18	0.17
College	0.49	0.09	0.01
Baseline age group^c
50–55(reference)	0.17	0.32	0.25
56–60	0.20	0.20	0.28
61–65	0.17	0.22	0.20
66–70	0.12	0.17	0.13
71–75	0.15	0.10	0.08
76+	0.19	0.10	0.06
Female (ref = male)^c	0.59	0.61	0.53
Any IADL at baseline (ref = 0)^c	0.16	0.08	0.18
Any chronic conditions at baseline (ref = 0)^c	0.66	0.53	0.34
Income tertiles^c
Low (reference)	0.47	0.32	0.46
Medium	0.24	0.37	0.26
High	0.30	0.31	0.28
Rural residents (ref = urban)^c	0.03	0.58	0.82
Current smoker (ref = no)^c	0.14	0.36	0.40
Current alcohol consumer (ref = no)^c	0.39	0.23	0.32
Any physical exercise/week (ref = 0)^c	0.88	0.34	0.58
Age at baseline^b	69.71(10.01)	67.35(9.13)	65.88(8.49)
Number of person-in-year	47,770	38,511	37, 952
Number of unique participants	17,822	11,484	12,137

The range for the HRS is (−2.65, 2.95), MHAS is (−2.16, 2.23) and CHARLS is (−1.58, 4); baseline centered; wave 1-HRS2012, MHAS2012, CHARLS2011; wave 2-HRS2016, MHAS2015, CHARLS2015; wave 3-HRS2018, MHAS2018, CHARLS2018; wave 4-HRS2020, MHAS2021, CHARLS2020.

Mean (SD).

Proportion.

To determine if changes in verbal memory varied by baseline age, we included an interaction term for age and age group in Model 2 (see Supplemental Table 1). Additionally, we included a three-order interaction term for age, age group, and education to determine if the association between baseline age and change in verbal memory was modified by educational attainment (see Supplemental Table 2). All analyses were completed using Stata 14.2.

Figure 1 displays decline in verbal memory across age groups based on the results presented in Supplemental Table 1. Figures 2 and 3 present decline in verbal memory among age groups based on the results presented in Supplemental Table 2. The effects of age on decline in verbal memory in each figure were estimated by using the following formula:

Y_{it} = b_{1 t (age group)} + b_{2 t (age)} + b_{3 (age - squared)} + b_{4 t (interaction effects)} + b_{5 t (covarites)} + b_{i 6 (random slope)} + b_{i 7 (random intercept)}

Figure 1.

Verbal memory decline by age groups. The downward curve means the coefficient of age-squared is negative in the model, the upward curve means the coefficient of age-squared is positive in the model.

Figure 2.

Verbal memory decline among respondents aged 50–65.

Figure 3.

Verbal memory decline among respondents older than 66.

The interaction effects refer to the interaction effects of age group and age in Figure 1, and the interaction effects of age group, age, and education in Figures 2 and 3.

Results

Descriptive characteristics

Table 1 presents the standardized verbal memory scores (z-scores), educational attainment, and demographic and health characteristics of the HRS, MHAS, and CHARLS participants in the final sample. The mean baseline z-score was 0.004 (SD = 0.997) for HRS participants, 0.239 (SD = 0.765) for MHAS participants, and −0.048 (SD = 0.964) for CHARLS participants. At baseline, HRS participants were, on average, the oldest at 69.7 years, followed by MHAS participants at 67.4 years, and CHARLS participants at 65.9 years. Nearly half (42%) of HRS participants had a high school education, compared to 4% of MHAS participants and 8% of CHARLS participants. About 16% of MHAS participants had no formal education and only 9% had a college education. CHARLS participants had the lowest levels of education, with 32% having no formal education and only 1% having a college education. The percentage of participants with limitations in one or more IADLs was similar in the HRS (16%) and CHARLS (18%), but lower in the MHAS (8%). The HRS had the highest percentage of participants with one or more chronic health conditions (66%), current alcohol consumption (39%), and regular weekly exercise (88%). Nearly 90% of CHARLS participants lived in rural areas, and 40% were current smokers. The MHAS had the highest percentage of female participants (61%), followed by the HRS (59%) and CHARLS (53%).

Change in verbal memory by baseline age group

We detected distinct age gradients in verbal memory scores (see Figure 1 and Supplemental Table 1). Baseline verbal memory scores decreased with increasing baseline age for participants across studies, except for HRS participants aged 66–70. Additionally, the quadratic term for age was statistically significant in the MHAS (p < 0.001) and CHARLS (p < 0.001) but not the HRS. In the HRS, verbal memory scores increased with age for participants aged 50–55, 56–60, and 61–65, and decreased for participants aged 66–70, 71–75, and 76 + . For CHARLS participants, the statistically significant quadratic term for age was positive (b = 0.002, p < 0.001). Consequently, verbal memory scores decreased and then increased with age to the point where scores were higher than at baseline for CHARLS participants aged 50–55, 56–60, and 61–65. Verbal memory scores decreased steeply for CHARLS participants aged 66–70, 71–71, and 76 + before increasing slightly. For MHAS participants, the statistically significant quadratic term for age was negative (b = −0.001, p < 0.001). Verbal memory scores with age were consistent for MHAS participants aged 50–55 and 56–60. Verbal memory scores decreased with age for MHAS participants aged 61–65. The trajectory of decline in verbal memory increased with baseline age and was steepest for MHAS participants aged 76 + .

Change in verbal memory by baseline age group and educational attainment

We present changes in verbal memory by baseline age group and educational attainment for participants aged 50 to 65 in Figure 2, and for participants aged 66–70, 71–75, and 76 + in Figure 3. The line color represents educational attainment, and the line type indicates the baseline age group. In all studies, verbal memory scores were higher with greater educational attainment and lower with older baseline age. In the HRS and MHAS, changes in verbal memory scores with age by baseline age group were consistent across educational attainment categories. In contrast, for CHARLS participants, verbal memory scores decreased with age for those with no formal education and some elementary education, but increased for those with an elementary, middle school, high school, or college education.

Verbal memory scores decreased with age for HRS and MHAS participants aged 66–70, 71–75, and 76+, regardless of educational attainment (see Figure 3). Additionally, the decline in verbal memory scores by baseline age group was similar across educational attainment categories. Conversely, trajectories of verbal memory scores by age group varied by educational attainment. For participants aged 66–70 and 71–75, trajectories of verbal memory scores decreased for those with no formal education and increased for those with a high school or college education. Verbal memory trajectories decreased with age for CHARLS participants aged 76+, regardless of educational attainment.

Robustness test

Given the different proportion of zero scores in each database, we re-estimated the model with non-zero verbal scores only to capture the possible impacts of the floor effects. The overall pattern of results remained consistent across datasets, with only slight differences in effect sizes. We also estimated the patterns of verbal memory decline including proxy using HRS and CHARS, as there is no measure of verbal memory in the proxy interview in MHAS. The patterns do not change. The effect size does not change either. The only difference is the predicted standardized results increased in CHARLS. Given the robust pattern, small proportion of proxy (<10% for each database), and the possible inaccuracy of proxy, we decided to exclude proxy data for each database.

Discussion

This analysis used data from three large, nationally representative cohort studies of aging in the United States, Mexico, and China to compare trajectories of verbal memory by age group and educational attainment. We detected clear gradients in the age at which verbal memory begins to decline across all three populations. For HRS and CHARLS participants, verbal memory began to decline among those aged 66–70 at baseline, whereas for MHAS participants, verbal memory began to decline at age 61–65, including for participants with high levels of education. Verbal memory declined for all CHARLS participants in 5-year age groups between 50 and 65 years of age with no formal education or only some elementary education. These findings suggest that educational attainment modifies the association between the baseline age group and the decline in verbal memory of participants in the CHARLS but not the HRS or MHAS.

The younger age of onset of decline in verbal memory in the MHAS compared to the HRS may be attributed to differences in early life experiences and educational opportunities between the two countries. Mexico has undergone several educational reforms since the 1930s.³⁶ These reforms, especially those implemented after the 1940s, increased federal spending on education and expanded infrastructure, including the number of elementary schools and teachers.³⁶ However, these gains were concentrated in urban areas of Mexico.³⁶ Many middle-aged and older Mexicans were born into poor socioeconomic conditions.³⁶ Consequently, children often had to forgo education to work and support their families.³⁷ These early-life disadvantages can accumulate over the life course and adversely affect cognitive functioning in later life.³⁸

Additionally, the coefficient of age group 56–60 in MHAS showed different patterns in interaction terms in Supplemental Tables 1 and 2, which suggested moderation effects of education for this age group. To be specific, those who age 56–60 in 2012 were born between 1952–1956. Mexico initialed a 11-year plan from 1959 which substantially expand the coverage of elementary school and junior middle school until 1971.³⁶ Moreover, compared to the age group 50–55 who showed higher proportion of middle school education and above, the age group 56–60 showed comparatively lower verbal memory (b = −0.633, see Supplemental Table 2). This special pattern is a result of age, cohort, and period effects.

Middle-aged and older adults in China have also experienced disadvantages in educational opportunities. Thus, it is unclear why MHAS participants showed a younger age of onset of decline in verbal memory than CHARLS participants. Middle-aged and older adults in China may have benefited from a more diverse labor market, beginning in the 1990s. Employment in cognitively demanding environments has been associated with higher cognitive functioning later in life.³⁹ China has also introduced pension programs for middle-aged and older adults. Prior research has linked pension benefits to higher cognitive functioning among older adults in China.⁴⁰

The association between education and verbal memory showed different patterns across the HRS, MHAS, and CHARLS. One possible explanation is that the effect size of education depends on its role in social status attainment, which varies across societies. Additionally, population health status may influence this relationship. For instance, given the higher prevalence of chronic conditions in HRS, the total effect of education on cognitive function may be underestimated due to health-related pathways. Moreover, health behaviors such as smoking, alcohol consumption, and physical activity contribute to the link between education and cognitive function. These factors may act as mediators rather than confounders, meaning that part of education's influence operates through them. Importantly, the extent of mediation may differ across countries due to variations in lifestyle behaviors, healthcare systems, and socioeconomic structures.

We found that the age of onset of decline in verbal memory among CHARLS participants varied by educational attainment. Specifically, participants aged 50–55 with no formal education exhibited a steep decline in verbal memory with age. Conversely, participants in 5-year age groups from 50 to 65 with elementary, high school, or college education showed increases in verbal memory with age. These findings are important given the high percentage of participants with no formal education.

The strong associations between educational attainment and older adults’ cognitive health mean the large differences in educational attainment between older adults in the United States, Mexico, and China have important implications for cognitive aging in each country. Nearly half of HRS participants aged 60 and older had a college level of education and less than ten percent of participants had less than a high school level of education. Conversely, sixteen percent of MHAS participants and thirty-two percent of CHARLS participants had no formal education and twenty-two percent of participants in both studies had an elementary school level of education. A person's level of education relative to the population will have a substantial role in that person's social standing or social status.⁴¹ While an elementary school level of education would be a considerable disadvantage for older adults in the United States, this same level of education could give an older adult in Mexico or China a relatively high social status, with access to better employment opportunities and other resources over the life course that are associated with higher cognitive functioning in late life.

Limitations

Our study addresses the critical need for research on changes in cognitive functioning among middle-aged and older adults who have experienced different social and cultural contexts throughout their lives. However, there are several limitations to consider when interpreting our results. First, the only measure of cognitive functioning we used in our analysis was verbal memory. Thus, our findings are not generalizable to other cognitive domains that may have different age-related trajectories or reflect differences in overall cognitive abilities. While the HRS, MHAS, and CHARLS were designed for cross-national comparative research, it is difficult to compare cognitive measures across these three studies because of differences in the cognitive items used in each survey. Previous research has used the Harmonized Cognitive Assessment Protocol (HCAP), which has been completed in the HRS, MHAS, CHARLS, and other studies, to conduct cross-sectional analyses of cognitive functioning. However, China has not completed follow up interviews of their HCAP cohort, which kept us from using HCAP data for this analysis.

Second, we had to exclude observations from participants who required a proxy to complete the interview. The MHAS does not include a proxy measure for a participant's verbal memory comparable to the verbal memory measure used in the direct interview. Participants who require a proxy to complete an interview are generally in poorer health than participants who can complete a direct interview.⁴² Thus, our sample will have included healthier participants who may have higher cognitive functioning.

Third, differential population aging rate, mortality and attrition between the HRS, MHAS, and CHARLS may have influenced our results, especially for the older age groups. Less healthy respondents are more likely to be lost-to-follow up or become deceased.^43,44 Additionally, epidemiological studies have found that cognitive decline accelerates before death.^45–47 Our results showed an increasing trend in verbal memory for some age and education groups, especially in MHAS and CHARLS, which may be due in part to healthier participants remaining in the sample.

Finally, differences in the timing of survey waves across the HRS, MHAS, and CHARLS may affect the precision of cognitive change estimates. Additionally, the year of education is not available in CHARLS and we harmonized education into categorical variables based on established classifications. While these steps support cross-national comparability, they may reduce measurement precision and potentially attenuate observed associations.

Conclusion

Using data from nationally representative studies of older adults in the US, Mexico, and China, we found that higher educational attainment was associated with higher performance in verbal memory but that age-related differences in verbal memory by educational attainment varied by country, with Mexican older adults exhibiting the steepest decline in verbal memory. These findings provide new evidence that educational attainment does not universally protect cognitive functioning across different social and cultural contexts. The age at which decline in verbal memory begins varies by educational attainment and other characteristics among older adults who have experienced diverse social and cultural contexts over the life course. Future research should also consider other important characteristics that shape older adults’ educational attainment and later life cognitive functioning, such as gender, race, and ethnicity. Continued cross-national research is needed to investigate the relationship between educational attainment and age-related changes in cognitive functioning as more highly educated generations reach older age.

Supplemental Material

sj-docx-1-alr-10.1177_25424823251351631 - Supplemental material for Education and age of decline in verbal memory across countries: Evidence from the HRS, MHAS, and CHARLS

Supplemental material, sj-docx-1-alr-10.1177_25424823251351631 for Education and age of decline in verbal memory across countries: Evidence from the HRS, MHAS, and CHARLS by Chengming Han, Octavio N Bramajo and Brian Downer in Journal of Alzheimer's Disease Reports

Footnotes

Acknowledgements

We appreciate the professional edition from Amber S. McIlwain, the editor of Sealy Center on Aging, UTMB.

ORCID iDs

Chengming Han

Octavio N. Bramajo

Brian Downer

Author contributions

Chengming Han (Conceptualization; Data curation; Formal analysis; Methodology; Writing – original draft; Writing – review & editing); Octavio N. Bramajo (Data curation; Writing – original draft; Writing – review & editing); Brian Downer (Methodology; Supervision; Writing – original draft; Writing – review & editing).

Funding

This research was supported by the National Institutes of Health/National Institute on Aging (grant numbers R01AG068988, P30AG024832, P30AG059301). The MHAS (Mexican Health and Aging Study) is partly sponsored by the National Institutes of Health/National Institute on Aging (grant number NIH R01AG018016) in the United States and the Instituto Nacional de Estadística y Geografía (INEGI) in Mexico.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data availability statement

The survey data from the HRS, MHAS, and CHARLS databases used in the present study are all available to the public.

Supplemental material

Supplemental material for this article is available online.

References

Sosa-Ortiz

Acosta-Castillo

Prince

. Epidemiology of dementias and Alzheimer's disease. Arch Med Res 2012; 43: 600–608.

Prince

Ali

Guerchet

, et al. Recent global trends in the prevalence and incidence of dementia, and survival with dementia. Alzheimers Res Ther 2016; 8: 23.

Jones

Ali

Kenny

, et al. Potentially modifiable risk factors for dementia and mild cognitive impairment: an umbrella review and meta-analysis. Dement Geriatr Cogn Disord 2024; 53: 91–106.

Livingston

Huntley

Sommerlad

, et al. Dementia prevention, intervention, and care: 2020 report of the Lancet Commission. Lancet 2020; 396: 413–446.

Barnes

Yaffe

. The projected effect of risk factor reduction on Alzheimer’s disease prevalence. Lancet Neurol 2011; 10: 819–828.

Stephan

BCM

Cochrane

Kafadar

, et al. Population attributable fractions of modifiable risk factors for dementia: a systematic review and meta-analysis. Lancet Healthy Longev 2024; 5: e406–e421.

Skirbekk

Loichinger

Weber

. Variation in cognitive functioning as a refined approach to comparing aging across countries. Proc Natl Acad Sci U S A 2012; 109: 770–774.

Lenehan

Summers

Saunders

, et al. Relationship between education and age-related cognitive decline: a review of recent research. Psychogeriatrics 2015; 15: 154–162.

Formanek

Kagstrom

Winkler

, et al. Differences in cognitive performance and cognitive decline across European regions: a population-based prospective cohort study. Eur Psychiatry 2019; 58: 80–86.

10.

Gonçalves

Avila

Bertola

, et al. Education and cognitive function among older adults in Brazil and Mexico. Alzheimers Dement (Amst) 2023; 15: e12470.

11.

Llibre Rodriguez

Ferri

Acosta

, et al. Prevalence of dementia in Latin America, India, and China: a population-based cross-sectional survey. Lancet 2008; 372: 464–474.

12.

Díaz-Venegas

Samper-Ternent

Michaels-Obregón

, et al. The effect of educational attainment on cognition of older adults: results from the Mexican Health and Aging Study 2001 and 2012. Aging Ment Health 2019; 23: 1586–1594.

13.

Lövdén

Fratiglioni

Glymour

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

14.

Zahodne

Stern

Manly

. Differing effects of education on cognitive decline in diverse elders with low versus high educational attainment. Neuropsychology 2015; 29: 649–657.

15.

Clouston

SAP

Glymour

Terrera

. Educational inequalities in aging-related declines in fluid cognition and the onset of cognitive pathology. Alzheimers Dement 2015; 1: 303–310.

16.

Hall

Derby

LeValley

, et al. Education delays accelerated decline on a memory test in persons who develop dementia. Neurology 2007; 69: 1657–1664.

17.

Mungas

Gavett

Fletcher

, et al. Education amplifies brain atrophy effect on cognitive decline: implications for cognitive reserve. Neurobiol Aging 2018; 68: 142–150.

18.

Wilson

Segawa

Boyle

, et al. The natural history of cognitive decline in Alzheimer’s disease. Psychol Aging 2012; 27: 1008–1017.

19.

Stern

Arenaza-Urquijo

Bartrés-Faz

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

20.

Wilson

Lamar

, , et al. Education and cognitive reserve in old age. Neurology 2019; 92: e1041–e1050.

21.

Jester

Palmer

Thomas

, et al. Impact of educational attainment on time to cognitive decline among marginalized older adults: cohort study of 20,311 adults. J Am Geriatr Soc 2023; 71: 2913–2923.

22.

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.

23.

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.

24.

Zhang

O’Shea

, et al. Educational attainment and later-life cognitive function in high- and middle-income countries: evidence from the Harmonized Cognitive Assessment Protocol. J Gerontol B Psychol Sci Soc Sci 2024; 79: gbae005.

25.

Zorrilla

. La educación secundaria en México: Al filo de su reforma. REICE Revista Iberoamericana Sobre Calidad, Eficacia y Cambio en Educación 2004. https://doi.org/10.15366/reice2004.2.1.004

26.

Instituto Nacional de Estadística, Geografía e Informática , et al. Estadísticas de educación. Aguascalientes. Ags 1994: 503.

27.

Esteve Palós

. Tendencias en homogamia educacional en México: 1970–2000. Estud Demogr Urbanos Col Mex 2005; 20: 341–362.

28.

Treiman

. Trends in educational attainment in China. Chin Sociol Rev 2013; 45: 3–25.

29.

Macías-Islas

MÁ

Aguayo-Arelis

Quintana

, et al. Deterioro cognitivo en pacientes con esclerosis múltiple del occidente de México. Rev Mex Neuroci 2016; 16: 3–12.

30.

Mejía-Arango

Miguel-Jaimes

Villa

, et al. Deterioro cognoscitivo y factores asociados en adultos mayores en México. Salud Pública Méx 2007; 49: S475–S481.

31.

. Cognitive function trajectories and factors among Chinese older adults with subjective memory decline: CHARLS Longitudinal Study Results (2011–2018). Int J Environ Res Public Health 2022; 19: 16707.

32.

Zhu

Qiu

Zeng

, et al. Leisure activities, education, and cognitive impairment in Chinese older adults: a population-based longitudinal study. Int Psychogeriatr 2017; 29: 727–739.

33.

Cho

Gross

, et al. Negative wealth shocks in later life and subsequent cognitive function in older adults in China, England, Mexico, and the USA, 2012–18: a population-based, cross-nationally harmonised, longitudinal study. Lancet Healthy Longev 2023; 4: e461–e469.

34.

Maharani

Dawes

Nazroo

, et al. Associations between self-reported sensory impairment and risk of cognitive decline and impairment in the health and retirement study cohort. J Gerontol B Psychol Sci Soc Sci 2020; 75: 1230–1242.

35.

Vivot

Power

Glymour

, et al. Jump, hop, or skip: modeling practice effects in studies of determinants of cognitive change in older adults. Am J Epidemiol 2016; 183: 302–314.

36.

Creighton

Park

. Closing the gender gap: six decades of reform in Mexican education. Comp Educ Rev 2010; 54: 513–537.

37.

Saenz

Milani

Mejía-Arango

. Gender, personality, and cognitive resilience against early-life disadvantage. J Gerontol B Psychol Sci Soc Sci 2023; 78: 913–924.

38.

Al Hazzouri

Haan

Galea

, et al. Life-course exposure to early socioeconomic environment, education in relation to late-life cognitive function among older Mexicans and Mexican Americans. J Aging Health 2011; 23: 1027–1049.

39.

Smart

Gow

Deary

. Occupational complexity and lifetime cognitive abilities. Neurology 2014; 83: 2285–2291.

40.

Peng

Burr

Han

. Cognitive function and cognitive decline among older rural Chinese adults: the roles of social support, pension benefits, and medical insurance. Aging Mental Health 2022; 27: 771–779.

41.

Hendi

Elo

Martikainen

. The implications of changing education distributions for life expectancy gradients. Soc Sci Med 2021; 272: 113712.

42.

Downer

Avila

Chen

, et al. Imputation procedures for cognitive variables in the Mexican health and aging study: evaluating the bias from excluding participants with missing data. Real Datos Espacio 2021; 12: 90.

43.

Salthouse

. Selectivity of attrition in longitudinal studies of cognitive functioning. J Gerontol B Psychol Sci Soc Sci 2014; 69: 567–574.

44.

Jacobsen

Ran

Liu

, et al. Predictors of attrition in a longitudinal population-based study of aging. Int Psychogeriatr 2021; 33: 767–778.

45.

Chatfield

Brayne

Matthews

. A systematic literature review of attrition between waves in longitudinal studies in the elderly shows a consistent pattern of dropout between differing studies. J Clin Epidemiol 2005; 58: 13–19.

46.

Wilson

Beckett

Bienias

, et al. Terminal decline in cognitive function. Neurology 2003; 60: 1782–1787.

47.

Wilson

Beck

Bienias

, et al. Terminal cognitive decline: accelerated loss of cognition in the last years of life. Psychosom Med 2007; 69: 131–137.

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

0.06 MB