Intelligence or years of education: which is better correlated with memory function in normal elderly Japanese subjects?

Authors

  • Norio MURAYAMA,

    Corresponding author
    1. PET/CT Dementia Research Center, Juntendo Tokyo Koto Geriatric Medical Center, Juntendo University School of Medicine, Tokyo
    2. School of Allied Health Sciences, Kitasato University, Kanagawa
    Search for more papers by this author
  • Eizo ISEKI,

    1. PET/CT Dementia Research Center, Juntendo Tokyo Koto Geriatric Medical Center, Juntendo University School of Medicine, Tokyo
    2. Department of Psychiatry, Juntendo University School of Medicine, Tokyo, Japan
    Search for more papers by this author
  • Hirokuni TAGAYA,

    1. School of Allied Health Sciences, Kitasato University, Kanagawa
    Search for more papers by this author
  • Kazumi OTA,

    1. PET/CT Dementia Research Center, Juntendo Tokyo Koto Geriatric Medical Center, Juntendo University School of Medicine, Tokyo
    Search for more papers by this author
  • Koji KASANUKI,

    1. PET/CT Dementia Research Center, Juntendo Tokyo Koto Geriatric Medical Center, Juntendo University School of Medicine, Tokyo
    2. Department of Psychiatry, Juntendo University School of Medicine, Tokyo, Japan
    Search for more papers by this author
  • Hiroshige FUJISHIRO,

    1. PET/CT Dementia Research Center, Juntendo Tokyo Koto Geriatric Medical Center, Juntendo University School of Medicine, Tokyo
    2. Department of Psychiatry, Juntendo University School of Medicine, Tokyo, Japan
    Search for more papers by this author
  • Heii ARAI,

    1. Department of Psychiatry, Juntendo University School of Medicine, Tokyo, Japan
    Search for more papers by this author
  • Kiyoshi SATO

    1. PET/CT Dementia Research Center, Juntendo Tokyo Koto Geriatric Medical Center, Juntendo University School of Medicine, Tokyo
    Search for more papers by this author

Dr Norio Murayama, PhD, 1-15-1, Kitasato, Minami-ku, Sagamihara, Kanagawa 252-0373, Japan. Email: n-mura@kitasato-u.ac.jp

Abstract

Background:  We compared differences in intelligence and memory function between normal elderly Japanese subjects with more years of education and those with fewer years of education. We also investigated clinical and neuropsychological factors that are strongly correlated with memory function.

Methods:  There were 118 normal elderly subjects who underwent the Mini-Mental State Examination, Wechsler Adult Intelligence Scale, 3rd edition (WAIS-III), and Wechsler Memory Scale Revised. Subjects with at least 13 years of education were categorized as the H group, and those with 12 years of education or less were categorized as the L group.

Results:  Age and Mini-Mental State Examination scores were not significantly different between the two groups. On the WAIS-III, there were significant differences between the two groups in Verbal IQ and Full Scale IQ. On the Wechsler Memory Scale Revised, there were significant differences between the two groups in Visual Memory, General Memory, and Delayed Recall. Correlation coefficients between memory function and the other factors demonstrated significant but weak correlations between years of education and General Memory (R = 0.22) and between years of education and Delayed Recall (R = 0.20). Strong correlations were found between Verbal IQ and Verbal Memory (R = 0.45), between Verbal IQ and General Memory (R = 0.49), between Full Scale IQ and General Memory (R = 0.50) and between Full Scale IQ and Delayed Recall (R = 0.48).

Conclusions:  In normal elderly Japanese subjects, years of education weakly correlated with memory function while Verbal IQ, Full Scale IQ and Verbal Comprehension on WAIS-III had stronger correlations with memory function. Verbal IQ and Verbal Comprehension on WAIS-III were found to be insusceptible to the cognitive decline characteristic of Alzheimer's disease or amnestic mild cognitive impairment. Therefore, verbal intelligence, as measured by Verbal IQ and Verbal Comprehension, may be the most useful factor for inferring premorbid memory function in Alzheimer's disease or amnestic mild cognitive impairment patients.

INTRODUCTION

Amnestic mild cognitive impairment (aMCI), a subtype of MCI, presents as a decline in memory function that frequently progresses to Alzheimer's disease (AD).1,2 According to the current clinical diagnostic criteria for MCI,1,3 aMCI patients are characterized as having memory complaints (preferably corroborated by an observer), impaired memory function (especially given their age and education), preserved general cognitive function, intact activities of daily living and no dementia. aMCI is characterized by a Clinical Dementia Rating of 0.5 or a rating on the Global Deterioration Scale of 3.4,5 Therefore, an objective assessment of impaired memory function is essential for a diagnosis of aMCI.

Neuropsychological assessment has played an important role in detecting early AD or aMCI,1 and numerous neuropsychological studies have been conducted.6,7 However, most of these studies of aMCI are intended for patients in the late stage of aMCI, close to early AD. Since Petersen et al. reported that aMCI patients show an average decline of 1.5 SD in memory tasks, a number of studies have defined aMCI based on a memory decline of 1.0 or 1.5 SD from the average score of normal elderly subjects.2,8–10 Clinically, however, most aMCI patients who meet this requirement show slight impairment in their activities of daily living and are likely to correspond to early AD patients.

Petersen et al. also noted the importance of considering educational levels when assessing the memory function of normal elderly subjects.1,11 Subjects with more years of education tend to demonstrate a better memory function than those with fewer years of education. However, educational levels are diverse depending on country and ethnic background, and the relation of educational level to memory function may differ between Japan and other countries.

In order to detect an early stage of aMCI, between the normal elderly state and a late stage of aMCI, memory function of aMCI patients should be examined with a longitudinal assessment that compares their current memory function with their premorbid memory function. Yet, very few aMCI patients have had their memory function assessed before cognitive impairment became apparent. This is why a cross-sectional assessment comparing current memory function of aMCI patients with average memory function of normal elderly subjects in the same age group is commonly used. However, with cross-sectional assessment, subjects with higher premorbid memory function may not demonstrate significantly objective impaired memory function even if they are aMCI.12

Murayama et al. indicated that early stage aMCI patients without significantly objective impaired memory function can be detected by examining discrepancies between scores on the Wechsler Memory Scale Revised (WMS-R) and Wechsler Adult Intelligence Scale, 3rd edition (WAIS-III).12–16 However, it is necessary to examine which WAIS-III and WMS-R scores correlate to one other in order to evaluate memory function of aMCI appropriately.

In the present study, we examined the differences in intelligence and memory function between normal elderly Japanese subjects with more years of education and those with fewer years of education. In addition, we investigated the clinical and neuropsychological factors that strongly correlate with memory function.

METHODS

There were 118 normal elderly Japanese subjects (84 women and 34 men) were identified in the present study. The subjects participated in the normative database construction of 18F-fluorodeoxyglucose positron emission tomography or consulted our memory clinic at Juntendo Tokyo Koto Geriatric Medical Center (Tokyo, Japan) between 2006 and 2010.17

There were several criteria for subject selection. The age of subjects ranged from 60 to 74 years because WMS-R applies to subjects 74 years old and younger.14 None of subjects fulfilled the diagnostic criteria of MCI or dementia,1,3,18 or had neurological or psychiatric conditions that could potentially affect cognitive function, such as stroke, epilepsy, parkinsonism, schizophrenia or mood disorder. Evaluations were made by psychiatrists specializing in dementia. None of the subjects met the criteria for alcohol and/or drug abuse.

All subjects underwent brain magnetic resonance imaging and either 18F-fluorodeoxyglucose positron emission tomography or 123I-single photon emission computed tomography examinations. The magnetic resonance imaging examinations revealed no significant brain atrophy or vascular change inconsistent with their ages. The 18F-fluorodeoxyglucose positron emission tomography or 123I-single photon emission computed tomography examinations using 3D-stereotactic surface projection analysis displayed no significant hypometabolism or hypoperfusion in any brain region including the posterior cingulate gyrus and parietotemporal area.19,20

Subjects with 13 years or more of education were categorized as the high-education group (H group), and subjects with 12 years or less of education were categorized as the low-education group (L group).

Neuropsychological assessments, including Mini-Mental State Examination (MMSE),21 WAIS-III,15,16 and WMS-R,13,14 were conducted by clinical psychologists. In WAIS-III, intelligence was assessed by three IQ types including Verbal IQ, Performance IQ, and Full Scale IQ, and four indexes including Verbal Comprehension, Working Memory, Perceptual Organization, and Processing Speed, as well as 14 subtests. In WMS-R, memory function was assessed by four quotients including Verbal Memory, Visual Memory, General Memory, and Delayed Recall.

The t-test was used to examine differences between the H group and the L group. The Pearson product–moment correlation coefficient was used to detect the degree of correlation among these clinical and neuropsychological factors. For all statistical analyses, the null hypothesis was rejected at a significance level of P < 0.01 or P < 0.05.

The present study is part of a research project approved by the research ethics committee of Juntendo Tokyo Koto Geriatric Medical Center, and all subjects gave written informed consent.

RESULTS

The clinical and neuropsychological factors (age, years of education, MMSE, factors of WAIS-III and WMS-R) of all subjects are presented in Table 1. Three IQ types, four WAIS-III indexes and four WMS-R quotients were scored relative to a benchmark of 100, and 14 subtests of WAIS-III were scored relative to a benchmark of 10; 100 or 10 are the respective average scores for the subjects' age group.13,15

Table 1. Clinical and neuropsychological factors for all subjects
 All subjectsEducation t-test
H groupL group
Mean ± SDMean ± SDMean ± SD
  1. The null hypothesis was rejected at a significance level of **P < 0.01 or *P < 0.05.

  2. H group, high-education group (i.e. 13 or more years of education); L group, low-education group (i.e. 12 years or less of education); NS, not significant.

Age (years)69.0 ± 4.369.1 ± 4.368.8 ± 4.3NS
Education (years)12.8 ± 2.715.2 ± 1.210.6 ± 1.7 **
Mini-Mental State Examination score27.9 ± 2.127.8 ± 2.128.1 ± 2.2NS
Wechsler Adult Intelligence Scale, 3rd edition scores    
IQ    
 Verbal IQ108.5 ± 8.7111.8 ± 8.8105.5 ± 7.4 **
 Performance IQ103.5 ± 7.9104.8 ± 8.3102.3 ± 7.4NS
 Full scale IQ107.0 ± 7.7109.6 ± 8.3104.7 ± 6.4 **
Index    
 Verbal Comprehension104.4 ± 9.1108.1 ± 8.8101.0 ± 8.0 **
 Working Memory106.0 ± 10.2108.3 ± 10.5104.0 ± 9.6 *
 Perceptual Organization102.1 ± 7.8102.7 ± 8.1101.6 ± 7.5NS
 Processing Speed102.5 ± 8.4104.8 ± 9.4100.4 ± 6.8 **
Subtests    
 Picture Completion10.6 ± 1.810.8 ± 1.710.5 ± 1.8NS
 Vocabulary10.8 ± 1.911.4 ± 2.110.3 ± 1.6 **
 Digit Symbol-Coding10.7 ± 2.011.1 ± 2.310.4 ± 1.7NS
 Similarities11.3 ± 1.911.8 ± 1.710.8 ± 1.9 **
 Block Design10.3 ± 2.010.3 ± 2.110.4 ± 1.9NS
 Arithmetic10.1 ± 2.410.4 ± 2.710.0 ± 2.2NS
 Matrix Reasoning10.4 ± 1.610.6 ± 1.810.2 ± 1.5NS
 Digit Span13.1 ± 2.313.7 ± 2.312.6 ± 2.3 **
 Information10.3 ± 2.611.3 ± 2.59.3 ± 2.4 **
 Picture Arrangement11.0 ± 1.911.2 ± 2.010.7 ± 1.8NS
 Comprehension12.4 ± 2.112.6 ± 2.312.2 ± 1.9NS
 Symbol Search10.3 ± 1.710.8 ± 1.89.9 ± 1.5 **
 Letter-Number Sequencing10.1 ± 2.010.5 ± 2.19.8 ± 1.9 *
 Object Assembly10.6 ± 2.010.6 ± 2.010.5 ± 2.0NS
Wechsler Memory Scale Revised    
 Verbal Memory106.4 ± 8.4107.5 ± 8.3105.4 ± 8.3NS
 Visual Memory107.5 ± 10.3110.1 ± 10.1105.1 ± 10.0 **
 General Memory107.5 ± 8.8109.5 ± 8.7105.7 ± 8.7 *
 Delayed Recall104.0 ± 8.8106.0 ± 9.0102.2 ± 8.3 *

There were several differences in these factors between the H and L groups. The mean age of the H and L groups were 69.1 ± 4.3 years and 68.8 ± 4.3 years, respectively, and the t-test demonstrated no significant difference between the two groups. The mean MMSE scores of the H and L groups were 27.8 ± 2.1 and 28.1 ± 2.2, respectively, and there was no significant difference between the two groups.

In WAIS-III, there were significant differences between the mean Verbal IQ of 111.8 ± 8.8 in the H group and 105.5 ± 7.4 in the L group and between the mean Full Scale IQ of 109.6 ± 8.3 in the H group and 104.7 ± 6.4 in the L group (P < 0.01). There was no significant difference in mean Performance IQ between the two groups. Among the indexes, there were significant differences between the mean Verbal Comprehension of 108.1 ± 8.8 in the H group and 101.0 ± 8.0 in the L group, between the mean Processing Speed of 104.8 ± 9.4 in the H group and 100.4 ± 6.8 in the L group (P < 0.01), and between the mean Working Memory of 108.3 ± 10.5 in the H group and 104.0 ± 9.6 in the L group (P < 0.05). There was no significant difference in mean Perceptual Organization between the two groups. As for subtests, there were significant differences in six of fourteen subtests between the two groups, most of which were included in Verbal IQ.

In WMS-R, there were significant differences between the mean Visual Memory of 110.1 ± 10.1 in the H group and 105.1 ± 10.0 in the L group (P < 0.01), between the mean General Memory of 109.5 ± 8.7 in the H group and 105.7 ± 8.7 in the L group, and between the mean Delayed Recall of 106.0 ± 9.0 in the H group and 102.2 ± 8.3 in the L group (P < 0.05). No significant difference in mean Verbal Memory was demonstrated between the two groups.

Pearson product–moment correlation coefficients between memory function and the other factors are shown in Table 2. There were significant correlations between years of education and both General Memory (R = 0.22) and Delayed Recall (R = 0.20), although the correlations were weak. There was no significant correlation between MMSE score and memory function. There were significant correlations between the three WAIS-III IQ types and memory function (P < 0.01). As the top fifth of the strongest correlation coefficient were 0.45 and higher, this was defined as strong correlation. Strong correlations were found between Verbal IQ and Verbal Memory (R = 0.45), Verbal IQ and General Memory (R = 0.49), Full Scale IQ and General Memory (R = 0.50), and Full Scale IQ and Delayed Recall (R = 0.48). There were also significant correlations between the four WAIS-III indexes and memory function (P < 0.01 or P < 0.05), and a strong correlation was found between Verbal Comprehension and General Memory (R = 0.46). In addition, there were significant correlations between the 14 WAIS-III subtests and memory function (P < 0.01 or P < 0.05), although none demonstrated a strong correlation. Meanwhile, correlations between WAIS-III (Verbal IQ, Performance IQ, and Full Scale IQ) and years of education were 0.47, 0.16, and 0.40, respectively.

Table 2. Pearson product–moment correlation coefficients between memory and other factors
 Wechsler Memory Scale Revised scores
Verbal MemoryVisual MemoryGeneral MemoryDelayed Recall
  1. The null hypothesis was rejected at a significance level of **P < 0.01 or *P < 0.05. Strong positive correlations of 0.45 and higher are underlined.

Years of education0.170.160.22*0.20*
Mini-Mental State Examination score0.120.120.140.17
Wechsler Adult Intelligence Scale, 3rd edition scores    
IQ    
 Verbal IQ 0.45 ** 0.34** 0.49 ** 0.41**
 Performance IQ0.25**0.31**0.33**0.43**
 Full scale IQ0.44**0.38** 0.50 ** 0.48 **
Index    
 Verbal Comprehension0.39**0.35** 0.46 ** 0.36**
 Working Memory0.32**0.26**0.35**0.36**
 Perceptual Organization0.22*0.24**0.26**0.29**
 Processing Speed0.27**0.24*0.33**0.38**
Subtests    
 Picture Completion0.28**0.21*0.28**0.38**
 Vocabulary0.37**0.31**0.43**0.31**
 Digit Symbol-Coding0.130.24**0.23*0.32**
 Similarities0.24**0.33**0.33**0.26**
 Block Design0.010.050.050.04
 Arithmetic0.26**0.20*0.27**0.19*
 Matrix Reasoning0.150.26**0.22*0.20*
 Digit Span0.20*0.180.23*0.25**
 Information0.33**0.21*0.36**0.30**
 Picture Arrangement0.20*0.21*0.24**0.36**
 Comprehension0.27**0.050.21*0.21*
 Symbol Search0.35**0.150.34**0.32**
 Letter-Number Sequencing0.21*0.23*0.27**0.37**
 Object Assembly0.160.18*0.21*0.28**

For neuropsychological factors that demonstrated a strong correlation in Pearson product–moment correlation coefficients (Table 2), partial correlation analysis with years of education as covariance was conducted. The results indicated that a partial correlation of 0.20 was shown between Verbal IQ and Verbal Memory, 0.22 between Verbal IQ and General Memory, 0.26 between Full Scale IQ and General Memory, 0.27 between Full Scale IQ and Delayed Recall, and 0.23 between Verbal Comprehension and General Memory. All partial correlations were significant (P < 0.01).

Scatter plots illustrating the relationships between General Memory and four factors (years of education, Verbal IQ, Full Scale IQ, and Verbal Comprehension) are shown in Figure 1. General Memory had a stronger correlation with Verbal IQ, Full Scale IQ, and Verbal comprehension than with years of education.

Figure 1.

Scatter plots between General Memory and four factors (years of education, Full Scale IQ, Verbal IQ, and Verbal Comprehension). Vertical axes show General Memory in all charts. Correlation coefficient (R) for years of education, Full Scale IQ, Verbal IQ, and Verbal Comprehension are 0.22, 050, 0.49 and 0.46, respectively.

DISCUSSION

In the present study, we first examined whether intelligence or memory function differs according to years of education in normal elderly Japanese subjects. Concerning intelligence, the results of WAIS-III, a representative intelligence test, showed that both Full Scale IQ and Verbal IQ were significantly different according to years of education, while Performance IQ was not. All indexes, including verbal tasks such as Vocabulary and Information in subtests, also showed significant differences according to years of education, with the sole exception of Perceptual Organization. Some studies of normal elderly subjects in other countries also used intelligence tests to demonstrate correlations between years of education and Vocabulary, Information and Comprehension,22 as well as the correlation between years of education and Verbal IQ,23 which are consistent with the present results. Subjects with higher educational levels tend to show higher scores in these intelligence tests. Williams suggested that IQ could be estimated by subtracting 7 from a benchmark of 100 for every uncompleted year of education through high school (i.e. out of a possible total of 12 years of education) and by adding 7 to a benchmark of 100 for every year of education completed above high school.24

Concerning memory function, WMS-R, a representative memory test, showed that Visual Memory, General Memory, and Delayed Recall significantly differed according to years of education, while Verbal Memory did not. These results indicate that it is also important in Japan to consider years of education when memory function is assessed, as suggested by Petersen.1,11 Although few studies in Japan have examined the correlation between either years of education or intelligence and memory function, Harada et al. examined the correlations between years of education and neuropsychological tests in normal elderly Japanese subjects aged 60 years and older, and indicated that years of education had no correlation with story recall test, one of the verbal memory function tests, which is inconsistent with the present results.25

Secondly, in the present study, we examined which clinical and neuropsychological factors have strong correlation with memory function. Years of education had significant but weak correlations with both General Memory and Delayed Recall on the WMS-R, while it had no significant correlation with either Verbal Memory or Visual Memory. In contrast, the three WAIS-III IQ types had significant correlations with all four WMS-R quotients. In particular, strong correlations were observed between Verbal IQ and both Verbal Memory and General Memory, and between Full Scale IQ and both General Memory and Delayed Recall. The four WAIS-III indexes were also significantly correlated with all WMS-R quotients, and Verbal Comprehension and General Memory had strong correlations. These results indicate that memory function of normal elderly Japanese subjects has a stronger correlation with intelligence (especially Verbal IQ, Full Scale IQ, and Verbal Comprehension) than with years of education. However, partial correlation analysis with years of education as covariance showed weaker a correlation than Person product-moment correlation coefficients. This indicates that years of education has a significant influence on correlation coefficients between memory function and intelligence.

In other countries, several studies have reported that years of education strongly correlated with intelligence and weakly correlated with memory function.26,27 The correlation between years of education and intelligence may be explained as follows. Many subtests of conventional IQ batteries have contents that are explicitly taught in school. For example, one would expect that scores on the Vocabulary subtest of WAIS-Revised, the previous version of WAIS-III, correlate with years of education and other demographic variables that reflect school history.24 This is also supported by the results of the present study that indicates a significant correlation between IQ (especially Verbal IQ) and years of education. In contrast, possible causal models for correlation between years of education and memory function include the general effects of test-taking styles associated with higher educational levels,24 indicating that subjects with higher educational levels are usually more experienced in memory tests and their approach is apparently more skillful.24 This model may be applied to any kind of test, including intelligence and memory function.

The present study aims to establish an appropriate method to assess the decline of current memory function from premorbid memory function in early AD or aMCI patients by comparing WMS-R scores with WAIS-III scores. For that reason, we need to identify the factors that have strong correlations with premorbid memory function in early AD or aMCI patients and that are insusceptible to cognitive decline after patients progress to AD or aMCI. From the present study, we conclude that the correlation factors may be Verbal IQ, Full Scale IQ, and Verbal Comprehension of WAIS-III, which have strong correlations with memory function. For the insusceptibility factors, previous studies revealed that verbal tasks of intelligence tests, such as Vocabulary, Information, and Comprehension, are insusceptible to organic brain syndrome.28–30 These findings suggest that detection of discrepancies between WMS-R scores and scores of Verbal IQ and Verbal Comprehension of WAIS-III are helpful in assessing the decline of memory function in early AD or aMCI patients. Full Scale IQ of WAIS-III is not appropriate for this assessment, because it includes subtests that are susceptible to organic brain syndrome, such as Block Design and Digit-Symbol Coding.

In conclusion, years of education correlated with memory function at a certain level in normal elderly Japanese subjects. However, Verbal IQ, Full Scale IQ, and Verbal Comprehension of WAIS-III demonstrated stronger correlations with memory function than did years of education. Among these three factors of WAIS-III, Verbal IQ and Verbal Comprehension are insusceptible to the cognitive decline characteristic of AD or aMCI. Therefore, verbal intelligence, as measured by Verbal IQ and Verbal Comprehension, may be the most useful factor for inferring premorbid memory function in AD or aMCI patients, although further research that compares normal elderly subjects with aMCI patients is needed.

ACKNOWLEDGMENTS

The present study was supported in part by the Ogasawara Foundation for the Promotion of Science and Engineering (Tokyo, Japan).

Ancillary