Differential Impact of Education on Cognitive Performance in Neurological Patients with Progressive Cognitive Decline

Abstract

Background:

Education has a protective effect toward cognitive decline in advanced age and is an important factor contributing to cognitive reserve.

Objective:

To elucidate the interaction effect of education and global mental status on cognitive performance of older patients with progressive cognitive decline.

Methods:

This retrospective study included 1,392 patients. We performed moderation regressions to examine the interaction between education and global mental status (Mini-Mental State Examination (MMSE) score) on performance in episodic memory, executive functions (EF), language, and constructional praxis tests. Significant interaction effects were further explored through separate linear regressions by MMSE level (inferior: ≤24; intermediate: 25–27; superior: 28–30).

Results:

There was an interaction between MMSE and education for some but not all variables. At intermediate and superior MMSE levels, high-educated people had a clear advantage relative to low-educated people in verbal memory and EF tests. This advantage was not significant at an inferior MMSE level. In object naming, constructional praxis recall, and constructional praxis, high-educated people performed better than low-educated people, independently of MMSE level.

Conclusion:

Education has a differential effect on cognitive performance in patients with cognitive decline. While high education is not helpful for episodic memory and EF at low cognitive levels, it is still beneficial for retrieving words or other semantic knowledge. These findings suggest an interaction between global mental status and education on different cognitive domains and have strong clinical implications. Diagnostic judgments should be based on the knowledge of such interaction. This study highlights the beneficial but selective effects of high education.

Keywords

Aging cognition cognitive dysfunction cognitive reserve diagnosis education neuropsychology

INTRODUCTION

Formal education is one measure of early and mid-life environmental and experiential enrichment that enhances the chances for higher cognitive stimula-tion over the years [1] and has an impact on cognitive fitness in later life [2]. Different concepts have been proposed that may maintain cognitive functioning in older age and may enhance resilience against nor-mal and pathological cognitive decline. These concepts include cognitive reserve (CR), brain reserve (BR), and brain maintenance (BM). Stern et al. [3] have attempted to arrive at a consensus on the def-inition of these concepts. CR stands for the adaptability of cognitive functions that helps to account for the differential susceptibility of cognition or day-to-day functioning to the effects of aging or neurodegeneration. There is no direct measure of CR. Various socio-behavioral indices have been proposed to contribute to the development of CR, including education, IQ, occupational complexity, leisure and physical activity, and other protective factors such as bilingualism [4]. There are also new tools, such as the Cognitive Reserve Index questionnaire (CRIq) [5], that have been developed to summarize different aspects contributing to CR during the lifespan. BR refers to the neurobiological capital (numbers of neurons, synapses, etc.) and structural brain variations that explain why some people cope better than others with brain aging and pathology before clinical or cognitive changes emerge [3]. BM refers to the fact that the development of age-related brain changes and pathology may be reduced [3]. The different concepts are intrinsically linked [6, 7], given that the brain is modifiable by experience and training [3].

In this study, we analyzed the effect of education and not of CR in general. Education improves neural resources during childhood and young adulthood, possibly by enhancing synaptic density, and then mitigates the effects of neural decline caused by aging or age-related diseases [6]. As reported for occupation, premorbid IQ, and mental activities, education has a protective effect toward cognitive decline and brain pathology in advanced age, and is often taken as a proxy for CR [3, 8]. Although education forms an integral part of CR, we acknowledge that CR is a broader and more comprehensive concept which may have additional effects.

It has been shown that healthy older people with high education perform better than those with low education across different cognitive measures [9, 10]. A high educational level has been associated with a slower cognitive decline with aging [11] and a lower risk of developing dementia [12 –14]. In healthy individuals, education seems to have a differential effect on cognitive abilities in later life. In a recent study with healthy older adults [15], education has been found to positively influence the retrieval of common names and performance in episodic memory tasks but not the retrieval of famous proper names [16 –18]. A study on later-life abstract math abilities [19] has shown that years of education (and not CR in general as defined by CRIq) [5] accounts for math performance in everyday tasks (NADL battery) [20]. Education has also been found to moderate the re-lationship between health numeracy and cognitive functions such as mental calculation and executive functions [21]. All in all, these studies suggest that the influence the educational level has on aging and cognitive decline is a very complex one and may depend on the specific cognitive domain investigated. However, how this influence varies is still very poorly understood.

At an equal degree of cognitive deterioration, pat-ients diagnosed with Alzheimer’s disease (AD) and with high education show greater parieto-temporal perfusion changes than those with low education, indicating that the disease stage is more advanced in the first group [22]. In line with these findings, a recent study with asymptomatic older participants has reported education as a moderator of the relationship between memory performance and AD-related pathophysiological changes [23]. In the first stages of AD, people with high education may compensate for the disease pathology by using their CR and BR. For them, first clinical signs of dementia would appear at a severe AD pathology stage, when these resources are exhausted, and the extended cerebral lesions cannot be compensated anymore. As a consequence, AD patients with a high educational level would show a more rapid cognitive decline and have a lower life expectancy after diagnosis [7, 24].

Few studies have traced the effect of education on single cognitive domains over the course of dementia. It therefore remains unclear to what extent educa-tion influences the specific pattern of cognitive deterioration after diagnosis. In a small group study (10 patients in each group), Le Carret et al. [25] analyzed whether low-educated patients and high-educated patients exhibit an identical or different pat-tern of cognitive deterioration at the same degree of dementia severity (matched by MMSE scores). All participants performed the same tests assessing language, memory, attention, visual perception, conceptualization, and abstract reasoning [25]. Patients with low education showed lower scores in interference inhibition and episodic memory, while patients with high education showed lower scores in abstract reasoning. As Le Carret et al. [25] suggested, the cognitive processes that make it possible to discover complex rules would be particularly vulnerable in AD. Overall, this study indicates that high-educated and low-educated patients seem to have a different pattern of cognitive decline at the same degree of global cognitive deterioration. The investigation shed light on the fact that single cognitive domains may show a differential pattern along cognitive decline. However, the study has limitations (only 10 patients per group; all patients had low MMSE scores; only one MMSE level) and leaves several open questions. Stern [7] suggested the use of challenging tests in the diagnostic setting for individuals with very high levels of cognitive functioning. Associative learning tasks relying on the hippocampus may have greater sensitivity than other tasks in high-educated individuals [7]; see also [26]. That means, learning tasks probing episodic memory should show a mar-ked decline in high- and low-educated individuals.

In this retrospective study, we aimed to elucidate the relationship between cognitive performance (me-mory, executive functions, language, and constructional praxis), global mental status (as defined by the MMSE score), and education (i.e., years of formal schooling successfully attended) in a large cohort of older patients admitted to our neurology department. We focused on older people with a diagnosis of either subjective cognitive deficits (in spite of an average objective cognitive performance), mild cognitive impairment, neurodegenerative disorder (e.g., Parkinsonism, cerebellar degeneration), or dementia (e.g., Alzheimer’s dementia, vascular dementia, non-otherwise specified dementia). Specifically, we were interested in the moderation effect of education on the association between global mental status and cognitive performance. To this aim, we analyzed whether low-educated patients and high-educated patients show an identical or different pattern of cognitive performance at different degrees of cognitive deterioration, after controlling for age and sex effects.

Following the previously quoted studies, we hyp-othesized that cognitive performance positively correlates with global mental status and education. Specifically, we expected that people with a high MMSE score perform better than those with a low MMSE score. We also hypothesized that people with a high education level perform better than those with a low education level. Furthermore, we expected that education moderates the association between global mental status and cognitive performance. This moderation effect should be particularly evident in tasks assessing episodic memory and executive functions, which are known to be particularly sensitive to cog-nitive decline with aging [7, 27]. Also, many neur-odegenerative conditions predominantly affect the hippocampus [28] and the frontal lobes. In case that these regions are affected, we may expect a decrease in episodic memory and executive functions when a certain point of degeneration is reached, and compensation is not possible anymore. Temporal lobe areas supporting and maintaining semantic knowledge are less often affected.

In this study, we expected that the advantage of the high-educated people relative to the low-educ-ated people would be weaker at a lower MMSE level. Furthermore, we hypothesized that education should equally influence all levels of global mental status when performance is relying on semantic—crystallized—knowledge like in an object naming task. High education should be a benefit at all cognitive levels as regards semantic knowledge. Indeed, semantic knowledge and crystallized intelligence have been shown to be stable in older age [27]. With regard to constructional praxis, effects of education are open for investigation. Since we assessed here abilities that are overlearned in school (including a 3-dimensional drawing of a cube), we hypothesized that constructional praxis (and constructional praxis recall) should rely on semantic knowledge as object naming does. The assessment of the effects of education on cognitive performance is not only important for the understanding of normal and pathological aging but also for refining neuropsychological diagnostic procedures in advanced age.

METHODS

Participants

Data used here were obtained from a larger data-base containing neuropsychological assessments performed between June 2009 and February 2020 at the Laboratory for Neuropsychology, Department of Neurology, Medical University of Innsbruck, Austria. Inclusion criteria were at least 49 years of age, at least 5 years of obligatory school, and at least one neuropsychological assessment through the CERAD-plus (Consortium to Establish a Registry of Alz-heimer’s Disease) battery [29, 30]. In case of repeated assessments, only one assessment (in general, the last complete one) was taken into consideration for analysis. Subjects were excluded if they completed less than 70% of the CERAD subtests.

Diagnoses were taken from medical records ava-ilable at the Department of Neurology, Medical University of Innsbruck. The diagnosis at the time of assessment was recorded. Patients were diagnosed according to standard criteria by expert neu-rologists. Out of a database of 2,054 patients with different neurological diagnoses, we selected the following groups: Neurodegeneration (n = 590), which included patients with degenerative dementias (e.g., Alzheimer’s disease, frontotemporal dementia), with basal ganglia related diseases (e.g., Parkinson’s dis-ease, atypical Parkinson’s disease, Huntington’s disease), or with cerebellar degeneration; vascular dementia (n = 103) and mixed dementia (both neurodegeneration and vascular lesions; n = 59); dementia not otherwise specified, which was retrospectively not attributable to either neurodegeneration or vascular causes (n = 200); mild cognitive impairment (n = 215); and subjective cognitive deficits despite average cognitive performance (n = 225). We excluded all patients with focal vascular lesions (e.g., stroke), with inflammatory diseases (e.g., multiple sclerosis, meningitis), with neurological sleep disorders (e.g., narcolepsy), or with other neurological conditions (e.g., epilepsy, tumor, traumatic brain injury). Patients with predominant psychiatric diagnosis or with major organic, non-neurological diseases were also excluded. In sum, out of 2,054 cases, we entered data from 1,392 patients into the analysis.

Ethic approval

This retrospective study was approved by the local ethic committee (approval number: 1206/2020).

Lifetime education

This is the number of years of formal education successfully attended over a person’s lifetime. This was obtained via self-reported information.

Global mental status

As an index of global mental status, we used the raw score in the Mini-Mental State Examination (MMSE; score range 0–30) [31], version taken from [29, 30].

Neuropsychological assessment

The CERAD-Plus battery [29, 30] is widely used in the evaluation of cognitive decline and/or dementia [32]. It includes the standardized neuropsychological subtests category fluency (animals/min; no. of words), object naming (Boston Naming Test, BNT, short version; score range 0–15), constructional pr-axis (score range 0–11), word list learning (score range 0–30), word list recall (score range 0–10), word list recognition (difference correct hits minus false positive; score range -10–10), constructional praxis recall (score range 0–11), psychomotor speed (Trail Making Test A, TMT-A; time in s), cognitive flexibility (TMT-B; time in s), and phonemic fluency (s-words/min; no. of words). Subtests are paper-and-pencil tasks. Administration takes ca. 20–30 min. Raw scores in each subtests were submitted to analysis.

Missing data

Missing data in the TMT-B were substituted with 300 s, which is the time limit to conclude this task.

Statistical analysis

We analyzed behavioral data through IBM SPSS Statistics, version 24.0 (Armonk, NY: IBM Corp.). In general, significance was set at α= 0.05.

Sex effects

A MANOVA with sex (male, female) was carried out for scores in CERAD subtests and demographic characteristics (age, education).

Correlation analysis

We ran a Pearson correlation analysis between scores in CERAD subtests and demographic characteristics. We were in particular interested in the relationship of cognitive performance with global mental status (MMSE score), age, and education.

Hierarchical regression analysis

In a first step, we ran a hierarchical linear regression analysis for each CERAD subtest. Age and sex were entered in the first block to control for potential confounding demographic factors. The MMSE score was entered in the second block to control for global mental status. Finally, education was entered in block 3 to determine whether education modulated performance after controlling for demographic factors and global mental status.

Moderation regression analysis

In a second step, we ran a moderation analysis to investigate whether education moderated the relationship between performance in CERAD subtests and global mental status, after entering age and sex as covariates. This analysis was performed for each CERAD subtest. The moderation analysis was carried out on raw data by means of the PROCESS tool, version 3.3 [33], which automatically transforms the predictors using grand mean centering.

Linear regression analysis by MMSE group

To further explore the interaction between education and global mental status, patients were divided into three subgroups based on their MMSE score: Group1 with a MMSE score≤24 (inferior level), Group2 with a MMSE score between 25 and 27 (intermediate level), and Group3 with a MMSE score≥28 (superior level). Multiple linear regressions were performed for each MMSE group separately with age, sex, and education entered in the model. This analysis was carried out only for those CERAD variables where a significant moderation effect of education emerged.

For all analyses, p < 0.05 was considered significant. Correction for multiple testing is reported when appropriate.

RESULTS

Demographics and global mental status

1,392 patients were included in the study. The mean age was 73.12 y (SD 8.65, range 49–97), the mean education was 11.09 y (SD 2.88, range 5–20). In total, there were 692 male (49.7%) and 700 female patients (50.3%). The mean MMSE score was 25.73 (SD 3.44, range 11–30). 397 patients (28.5%) obtained a MMSE score≤24 (Group1), 491 patients (35.3%) a MMSE score between 25 and 27 (Group2), and 504 patients (36.2%) a MMSE score≥28 (Group3).

Sex effects

Results of a MANOVA with Sex (male, female) as between-subjects factor indicated significant sex effects in cognitive performance (F [13, 1335>] = 24.26, p < 0.001). Women performed better than men in phonemic fluency, while men performed better than women in object naming, constructional praxis, constructional praxis recall, and cognitive flexibility (Table 1).

Table 1

Results of a Multivariate Analysis of Variance with sex as between-subjects factor

		Male (n = 692)				Female (n = 700)
Variables	Max. score	M	SD	Min	Max	M	SD	Min	Max	p
Age (y)		72.32	8.59	49.00	92.00	73.91	8.65	50.00	97.00	0.001
Education (y)		11.82	3.09	5.00	20.00	10.36	2.46	5.00	20.00	0.000
MMSE (total score)	30	25.77	3.33	11.00	30.00	25.70	3.55	12.00	30.00	0.359
Category fluency (no. of words)		15.84	6.77	2.00	40.00	15.22	6.66	1.00	38.00	0.050
Object naming (total score)	15	13.34	1.88	1.00	15.00	12.83	2.15	3.00	15.00	0.000
Constructional praxis (total score)	11	10.04	1.32	3.00	11.00	9.61	1.48	4.00	11.00	0.000
Word list learning (total score)	30	14.94	4.76	0.00	28.00	15.51	5.48	2.00	30.00	0.117
Word list recall (total score)	10	4.20	2.44	0.00	10.00	4.30	2.85	0.00	10.00	0.693
Word list recognition (correct hits minus false positive)	10	7.97	2.33	–1.00	10.00	7.78	2.46	–7.00	10.00	0.076
Constructional praxis recall (total score)	11	6.94	3.32	0.00	11.00	5.62	3.63	0.00	11.00	0.000
Psychomotor speed (s)		74.10	37.67	17.00	179.00	72.91	36.43	16.00	179.00	0.739
Cognitive flexibility (s)		204.10	90.76	38.00	300.00	212.66	91.49	36.00	300.00	0.043
Phonemic fluency (no. of words)		8.89	4.72	1.00	27.00	9.67	5.20	1.00	28.00	0.004

Correlation analysis

Performance in all CERAD subtests correlated highly significantly with the MMSE score, education, and age (Table 2). In general, better performance in CERAD subtests was associated with younger age, higher education, and higher global mental status.

Table 2

Results of a Pearson correlation analysis between demographical variables (age, education), global mental status (MMSE), and performance in different cognitive tasks. Uncorrected p-values are reported. When a more stringent criterion of significance is adopted (due to multiple testing, p < 0.005), correlations remain significant

	Category fluency	Object naming	Constructional praxis	Word list learning	Word list recall	Word list recognition	Constructional praxis recall	Psychomotor speed	Cognitive flexibility	Phonemic fluency
Age r	–0.303^**	–0.232^**	–0.151^**	–0.343^**	–0.417^**	–0.302^**	–0.368^**	0.299^**	0.354^**	–0.159^**
p	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000
n	1387	1392	1392	1392	1392	1392	1392	1392	1392	1349
Education r	0.307^**	0.256^**	0.305^**	0.241^**	0.251^**	0.178^**	0.271^**	–0.206^**	–0.316^**	0.324^**
p	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000
n	1387	1392	1392	1392	1392	1392	1392	1392	1392	1349
MMSE r	0.589^**	0.513^**	0.451^**	0.678^**	0.663^**	0.594^**	0.631^**	–0.557^**	–0.610^**	0.467^**
p	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000
n	1387	1392	1392	1392	1392	1392	1392	1392	1392	1349

MMSE, Mini-Mental State Examination.

Please note that age correlated negatively with education (r = –0.093, p = 0.001) and the MMSE score (r = –0.314, p < 0.001), whereas education correlated positively with the MMSE score (r = 0.253, p < 0.001). That is, older people had a lower education and lower MMSE scores than younger people; people with a higher education level obtained higher MMSE scores than those with a lower education level. When a more stringent criterion of significance is adopted (due to multiple testing, p < 0.005), these correlations remain significant.

Hierarchical regression analysis

These results are reported in detail in Supplementary Table 1. Here, we summarize results of the final model (Model 3) for each CERAD subtest. After controlling for demographic characteristics and global mental status, education emerged as significant predictor of performance in category fluency, object naming, constructional praxis, word list learning, word list recall, constructional praxis recall, psychomotor speed, cognitive flexibility, and phonemic fluency. Education did not emerge as significant predictor for word list recognition.

Moderation regression analysis

These results are reported in Table 3 and in Fig. 1. In summary, after controlling for the effects of age and sex, the MMSE score and education resulted significant predictors of performance in the majority of the CERAD subtests (category fluency, object naming, constructional praxis, word list learning, word list recall, constructional praxis recall, psychomotor speed, cognitive flexibility, and phonemic fluency), confirming results of the hierarchical regression analyses. In these subtests, people with high MMSE scores performed better than people with low MMSE scores. Also, people with high education performed better than people with low education. Results of the moderation analyses further indicated a significant interaction between MMSE and education for some but not all subtests.

Table 3

Results of moderation regression analyses

Dep. Var.	Ind. Var.MMSE	Moderator Education	Interaction MMSE^*Education	Covariates Age, Sex	R²	Δ R²
	p	p	p	p
Category fluency	0.000	0.000	0.000	0.000, 0.560	0.400	0.012
Object naming	0.000	0.000	0.335	0.006, 0.0002	0.294	0.001
Constructional praxis	0.000	0.000	0.003	0.965, 0.000	0.258	0.005
Word list learning	0.000	0.0001	0.0001	0.000, 0.000	0.498	0.005
Word list recall	0.000	0.000	0.0001	0.000, 0.0003	0.505	0.005
Word list recognition	0.000	0.335	0.777	0.000, 0.455	0.368	0.0001
Constructional praxis recall	0.000	0.0009	0.600	0.000, 0.000	0.463	0.0001
Psychomotor speed	0.000	0.008	0.047	0.000, 0.020	0.337	0.003
Cognitive flexibility	0.000	0.000	0.000	0.000, 0.420	0.443	0.014
Phonemic fluency	0.000	0.000	0.000	0.000, 0.000	0.304	0.016

Fig. 1

Interactive effect of education on the relationship between global cognitive status (MMSE) and performance in different cognitive tasks. Dark solid line: low education (-1SD); Light solid line: high education (+ 1SD); Dotted line: mean education.

There was a significant interaction between MMSE and education for category fluency, word list learning, word list recall, psychomotor speed, cognitive flexibility, and phonemic fluency. In these subtests, the association between cognitive performance and global mental status was steeper for people with high education. For the constructional praxis subtest, a significant interaction was also found. In this case, however, the association between cognitive performance and global mental status was steeper for people with low education.

The interaction between MMSE and education was not significant for object naming and constructional praxis recall.

Linear regression analysis by MMSE group

These results are reported in detail in Supplementary Table 2. Here, we summarize results regarding education. For people with a MMSE score≤24 (Group1, inferior MMSE level), we found a significant effect of education on cognitive flexibility and constructional praxis but not on other cognitive variables. For people with a MMSE score between 25 and 27 (Group2, intermediate MMSE level), there was a significant effect of education on category fluency, phonemic fluency, cognitive flexibility, word list learning, word list recall, and constructi-onal praxis but not psychomotor speed. For people with a MMSE score≥28 (Group3, superior MMSE level), education emerged as a significant predictor of performance in all variables, i.e., in category fluency, phonemic fluency, psychomotor speed, cognitive flexibility, word list learning, word list recall, and constructional praxis.

Please note that due to the high number of missing values (N = 579) in the TMT-B test, which we subst-ituted with 300 s for analysis purposes, results of the regression analyses regarding cognitive flexibility have to be interpreted cautiously. A closer look at the distribution of missing values suggests an effect of education on cognitive flexibility at all MMSE levels (Table 4), in line with results of the linear regression analyses.

Table 4

Distribution of missing values in the TMT-B by MMSE group and education level

	MMSE
	Inferior (≤24 pt.)	Intermediate (25–27 pt.)	Superior (≥28 pt.)	Total
Education
Low (≤9 y)	161/183 (88.0%)	75/150 (50.0%)	16/99 (16.2%)	252/432 (58.3%)
Intermediate (10–12 y)	134/174 (77.0%)	101/247 (40.9%)	19/251 (7.6%)	254/672 (37.8%)
High (>12 y)	32/40 (80.0%)	38/94 (40.4%)	3/154 (1.9%)	73/288 (25.3%)
Total	327/397 (82.4%)	214/491 (43.6%)	38/504 (7.5%)	579/1392 (41.6%)

Legend. People in the low education subgroup attained a max. of 9 years of obligatory school. People in the intermediate education group attained between 10 and 12 years of school, including secondary or professional school. People in the high education subgroup attained more than 12 years of school, including pre-university or university education. This distribution is based on the Austrian educational system.

DISCUSSION

This study was concerned with the relationship between education, cognitive decline, and specific cognitive domains. Higher performance in tests ass-essing episodic memory, executive functions, langu-age, and constructional praxis correlated with higher education and better global mental status as defined by the MMSE score. In contrast to previous invest-igations [25], the large data set of this study allowed us to assess in detail the moderation effect of education on the relationship between global mental status and cognitive performance. After controlling for age and sex, we found a significant interaction between MMSE and education for some but not all cognitive variables. Education moderated significantly the relationship between MMSE and performance in tests of verbal memory and executive functions. The moderation effect of education was not significant for object naming and constructional praxis recall. That means, in subtests probing lexical-semantic knowledge (object naming task) or asking for recalling previously drawn geometrical figures (constructional praxis recall task), the advantage of high-educated people relative to low-educated people was similar at all MMSE levels.

To better figure out the interaction effect between education and MMSE, we divided our sample into three MMSE subgroups (inferior, intermediate, and superior MMSE level) and performed separate linear regression analyses. At intermediate and superior MMSE levels (i.e., MMSE score≥25), the effect of education was significant in several measures of verbal memory and executive functions, with the high-educated people showing a clear advantage relative to the low-educated people. This effect was not significant at an inferior MMSE level (MMSE score≤24), as both high- and low-educated people performed very poorly in verbal memory and executive function measures. Regarding constructional praxis, the effect of education was significant at all MMSE levels, with a slightly stronger effect at an inferior MMSE level. In sum, education emerged as a significant moderator of the relationship between MMSE and cognitive performance in tests of verbal memory and executive functions. In tests of object naming, constructional praxis recall, and to some extent constructional praxis, high-educated people performed better than low-educated people, independently of the MMSE level.

Results of this study suggest that, when the global mental status is low (e.g., because the disease has progressed), education has a differential effect on cognitive abilities. Indeed, both high- and low-edu-cated patients perform low in episodic memory and executive function tasks. This finding converges with previously reported results showing that advanced age is associated with a decline specifically in those cognitive domains (for a review, see [27]). Importantly, high-educated people maintain an advantage in tasks relying on semantic knowledge (e.g., object naming). Even when the global mental status is low, high-educated people relative to low-educated people profit from the previously acquired knowledge. This finding is in line with previous studies showing maintenance of semantic knowledge and crystalized intelligence in advanced age [27]. Based on these results, we may also assume that, in case of neurodegenerative diseases, episodic memory and executive functions decline first in high-educated patients, while naming is still preserved. As suggested by some authors [34], while episodic memory is localized to mesial temporal structures, semantic memory seems to involve a more widely distributed brain network, which makes it more resistant to changes of normal and pathological aging.

Importantly, this study provides evidence regarding the effect of education on single cognitive dom-ains at various levels of global mental status. The effect of high/low education has been amply documented in diagnostic batteries such as the CERAD [29, 30]. Such batteries allow the comparison of single patients’ performance with healthy age- and education-matched controls. However, they give no insight into the development of cognitive decline and the differential effect that education may have at a particular level. Our research thus adds important evidence regarding the effects of education in case of cognitive decline.

We should acknowledge at least two limitations of this retrospective study. First, although all patients have been diagnosed through magnetic resonance imaging, positron emission tomography, or other bio-logical markers, these data are not available for analysis purposes. Therefore, we do not have a measure about the degree of the neuropathological changes that the patients presented at the time of the neuropsychological assessment. This information could be combined with the MMSE score to measure disease severity. Second, we do not have information about the time when cognitive complaints have first appeared and about progression of cognitive decline. Following previous studies [7, 24], one could expect an interaction between these measures and educational level, with the high-educated patients showing a more rapid cognitive decline relative to the low-educated patients.

In conclusion, our results imply that the effect of high/low education is not uniform across different levels of cognitive decline and different cognitive domains. While high education is not helpful for episodic memory and executive function tasks at low cognitive levels, it is still beneficial for retrieving words or other semantic knowledge (e.g., drawing a cube). Semantic knowledge seems to profit from high education at different levels of cognitive decline. These findings should find application in diagnostic procedures. For example, an individual with high education may score in the average range in a task assessing semantic knowledge (such as vocabulary), while scoring in the impaired range in tasks probing executive functions. Our results underline the necessity of careful examination across different cognitive domains when cognitive decline is suspected. Ideally, such an assessment would also take into account the effects of high/low education on the different cognitive variables. As shown by our study, a single summary score does not reflect the true picture.

It is certainly important to acknowledge the interaction between global mental status and level of education on different cognitive domains and to base diagnostic judgments on the knowledge of such interaction. Overall, this research delineates that education may contribute to CR and has beneficial effects on cognitive performance. However, it also shows the limits in case of neurodegeneration.

Footnotes

ACKNOWLEDGMENTS

We thank M. Zelger-Rier for her help in preparing the database. We also thank the reviewers for their valuable suggestions to a first version of the manuscript.

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Meng

, D’Arcy

(2012) Education and dementia in the context of the cognitive reserve hypothesis: A systematic review with meta-analyses and qualitative analyses. PloS One 7, e38268.

Kramer

, Bherer

, Colcombe

, Dong

, Greenough

(2004) Environmental influences on cognitive and brain plasticity during aging. J Gerontol A Biol Sci Med Sci 59, M940–957.

Stern

, Arenaza-Urquijo

, Bartrés

, Belleville

, Cantilon

, Chetelat

, Ewers

, Franzmeier

, Kempermann

, Kremen

, Okonkwo

, Scarmeas

, Soldan

, Udeh-Momoh

, Valenzuela

, Vemuri

, Vuoksimaa

, Arenaza Urquiljo

, Bartrés

, Belleville

, Cantillon

, Chetelat

, Clouston

SAP

, Estanga

, Ewers

, Franzmeier

, Gold

, Habeck

, Jones

, Kempermann

, Kochhann

, Kremen

, Lim

, Martínez-Lage

, Morbelli

, Okonkwo

, Ossenkoppele

, Pettigrew

, Rosen

, Scarmeas

, Soldan

, Song

, Udeh-Momoh

, Stern

, Valenzuela

, Van Loenhoud

, Vemuri

, Vuoksimaa

(2020) Whitepaper: Defining and investigating cognitive reserve, brain reserve, and brain maintenance. Alzheimers Dement 16, 1305–1311.

Craik

FIM

, Bialystok

, Freedman

(2010) Delaying the onset of Alzheimer disease: Bilingualism as a form of cognitive reserve. Neurology 75, 1726–1729.

Nucci

, Mapelli

, Mondini

(2012) Cognitive Reserve Index questionnaire (CRIq): A new instrument for measuring cognitive reserve. Aging Clin Exp Res 24, 218–226.

Cabeza

, Albert

, Belleville

, Craik

FIM

, Duarte

, Grady

, Lindenberger

, Nyberg

, Park

, Reuter-Lorenz

, Rugg

, Steffener

, Rajah

(2018) Maintenance, reserve and compensation: The cognitive neuroscience of healthy ageing. Nat Rev Neurosci 19, 701–710.

Stern

(2012) Cognitive reserve in ageing and Alzheimer’s disease. Lancet Neurol 11, 1006–1012.

Stern

(2013) Cognitive reserve: Implications for assessment and intervention. Folia Phoniatr Logop 65, 49–54.

Delazer

, Kemmler

, Benke

(2013) Health numeracy and cognitive decline in advanced age. Neuropsychol Dev Cogn B Aging Neuropsychol Cogn 20, 639–659.

10.

Guerra-Carrillo

, Katovich

, Bunge

(2017) Does higher education hone cognitive functioning and learning efficacy? Findings from a large and diverse sample. PloS One 12, e0182276.

11.

Albert

, Jones

, Savage

, Berkman

, Seeman

, Blazer

, Rowe

(1995) Predictors of cognitive change in older persons: MacArthur studies of successful aging. Psychol Aging 10, 578–589.

12.

Snowdon

, Kemper

, Mortimer

, Greiner

, Wekstein

, Markesbery

(1996) Linguistic ability in early life and cognitive function and Alzheimer’s disease in late life. Findings from the Nun Study. JAMA 275, 528–532.

13.

Stern

, Gurland

, Tatemichi

, Tang

, Wilder

, Mayeux

(1994) Influence of education and occupation on the incidence of Alzheimer’s disease. JAMA 271, 1004–1010.

14.

Valenzuela

, Sachdev

(2006) Brain reserve and dementia: A systematic review. Psychol Med 36, 441–454.

15.

Delazer

, Lenhart

, Zamarian

, Nagele

, Gizewski

, Benke

, Scherfler

(2020) Cognitive reserve does not support the retrieval of well-known proper names in older people. Neuropsychology 34, 667–674.

16.

Mondini

, Semenza

(2016) Cognitive reserve and ageing. What does cognitive reserve protect in ageing? Front Psychol. 10.3389/conf.fpsyg.2016.68.00041.

17.

Montemurro

, Mondini

, Crovace

, Jarema

(2019) Cognitive reserve and its effect in older adults on retrieval of proper names, logo names and common nouns. Front Commun 4, 14.

18.

Montemurro

, Mondini

, Nucci

, Semenza

(2018) Proper name retrieval in cognitive decline: The role of cognitive reserve. Ment Lex 13, 215–229.

19.

Arcara

, Mondini

, Bisso

, Palmer

, Meneghello

, Semenza

(2017) The relationship between cognitive reserve and math abilities. Front Aging Neurosci 9, 429.

20.

Semenza

, Meneghello

, Arcara

, Burgio

, Gnoato

, Facchini

, Benavides-Varela

, Clementi

, Butterworth

(2014) A new clinical tool for assessing numerical abilities in neurological diseases: Numerical activities of daily living. Front Aging Neurosci 6, 112.

21.

Zamarian

, Lenhart

, Nagele

, Steiger

, Gizewski

, Benke

, Scherfler

, Delazer

(2020) Effects of cognitive functioning and education on later-life health numeracy. Gerontology 66, 582–592.

22.

Stern

, Alexander

, Prohovnik

, Mayeux

(1992) Inverse relationship between education and parietotemporal perfusion deficit in Alzheimer’s disease. Ann Neurol 32, 371–375.

23.

Joannette

, Bocti

, Dupont

, Lavallée

, Nikelski

, Vallet

, Chertkow

, Joubert

(2020) Education as a moderator of the relationship between episodic memory and amyloid load in normal aging. J Gerontol A Biol Sci Med Sci 75, 1820–1826.

24.

Contador

, Stern

, Bermejo-Pareja

, Sanchez-Ferro

, Benito-Leon

(2017) Is educational attainment associated with increased risk of mortality in people with dementia? A population-based study. Curr Alzheimer Res 14, 571–576.

25.

Le Carret

, Auriacombe

, Letenneur

, Bergua

, Dartigues

J-F

, Fabrigoule

(2005) Influence of education on the pattern of cognitive deterioration in AD patients: The cognitive reserve hypothesis. Brain Cogn 57, 120–126.

26.

Papp

, Amariglio

, Dekhtyar

, Roy

, Wigman

, Bamfo

, Sherman

, Sperling

, Rentz

(2014) Development of a psychometrically equivalent short form of the Face-Name Associative Memory Exam for use along the early Alzheimer’s disease trajectory. Clin Neuropsychol 28, 771–785.

27.

Spreng

, Turner

(2019) The shifting architecture of cognition and brain function in older adulthood. Perspect Psychol Sci J Assoc Psychol Sci 14, 523–542.

28.

Sabuncu

, Desikan

, Sepulcre

, Yeo

BTT

, Liu

, Schmansky

, Reuter

, Weiner

, Buckner

, Sperling

, Fischl

, Alzheimer’s Disease Neuroimaging Initiative (2011) The dynamics of cortical and hippocampal atrophy in Alzheimer disease. . Arch Neurol 68, 1040–1048.

29.

Berres

, Monsch

, Bernasconi

, Thalmann

, Stähelin

(2000) Normal ranges of neuropsychological tests for the diagnosis of Alzheimer’s disease. Stud Health Technol Inform 77, 195–199.

30.

Schmid

, Ehrensperger

, Berres

, Beck

, Monsch

(2014) The extension of the German CERAD Neuropsychological Assessment Battery with tests assessing subcortical, executive and frontal functions improves accuracy in dementia diagnosis. Dement Geriatr Cogn Disord Extra 4, 322–334.

31.

Folstein

, Folstein

, McHugh

(1975) “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 12, 189–198.

32.

Fillenbaum

, van Belle

, Morris

, Mohs

, Mirra

, Davis

, Tariot

, Silverman

, Clark

, Welsh-Bohmer

, Heyman

(2008) Consortium to Establish a Registry for Alzheimer’s Disease (CERAD): The first twenty years. Alzheimers Dement 4, 96–109.

33.

Hayes

(2012) PROCESS: A versatile computational tool for observed variable mediation, moderation, and conditional process modeling [White paper]. Retrieved from http://www.afhayes.com/public/process2012.pdf

34.

Darby

, Brickhouse

, Wolk

, Dickerson

, Alzheimer’s Disease Neuroimaging Initiative (2017) Effects of cognitive reserve depend on executive and semantic demands of the task. J Neurol Neurosurg Psychiatry 88, 794–802.