Effect of Age and Education on Performance on the Mini‐mental State Examination in a Healthy Older Population and During the Course of Alzheimer's Disease

To the Editor: The Mini-Mental State Examination (MMSE)1 is used widely to detect cognitive impairment in older people. However, the MMSE, as well as many other psychometric instruments, has been found to be heavily influenced by factors such as age and education, and its specificity and sensitivity in detecting cognitive impairment is ameliorated by correcting scores with normative data from the general population.2 The MMSE score is related positively to education level and negatively to age. This effect has been demonstrated in healthy populations, but the way in which these two factors interact with cognitive performance during the course of dementia is still unknown. Even if the clinical severity of the dementing illness is the main determinant of MMSE score variance, it is likely that a small part of the variance is also explained by age and education. Therefore, the study was designed to evaluate the magnitude of the association of age and education with the MMSE score in a “normal” elderly population and in a group of patients affected by Alzheimer's disease (AD) in order to examine the extent to which demographic variables retain their influence upon cognitive performance during the course of dementia. Data on normal elderly subjects were derived by two randomized epidemiological surveys conducted on subjects over the age of 65 in Ospitaletto (subjects = 498) and Tirano (subjects = 251) in northern Italy. This sample showed a mean age of 74.5 ± 6.6, a mean educational level of 5.4 ± 2.9 years of schooling, and a mean MMSE score of 26.7 ± 2.9. AD patients (n = 100) were recruited consecutively in the Alzheimer's Disease Unit of S. Cuore Fatebenefratelli Hospital in Brescia, northern Italy. Patients were diagnosed as probable AD according to current clinical criteria3 and were assessed with MMSE and CDR4 during their stay in the hospital. The AD group showed a mean age of 74.6 ± 8.7, a mean educational level of 6.3 ± 3.5 years, and a mean MMSE score of 13.2 ± 8.4. Fourteen patients presented with questionable clinical impairment (CDR = 0.5), 34 with mild (CDR = 1), 22 with moderate (CDR = 2), and 30 with severe (CDR = 3). Age was not statistically different between normal older controls and AD patients. The MMSE score was significantly different in the two groups (t test, P < .0001), as was education (t test, P = .021). The β-coefficients of multiple linear regressions of age and education with the MMSE score, adjusting for potential confounders, were adopted as an index of the association of these variables in normal subjects and AD patients (see Table 1). Because clinical severity is likely to be the major determinant of MMSE score, a model adjusted also for this variable was used for AD patients. Results of statistical analysis are presented in Table 1. As expected, in normal older subjects both age and education were significantly related to the MMSE score, presenting, respectively, a negative and a positive correlation. In AD patients, the model unadjusted for CDR showed a positive relationship of similar magnitude for education, whereas age was not a significant correlate of MMSE score. After adjusting for CDR, however, age was positively and significantly correlated with MMSE score. These results suggest that the influence of age and education on cognitive performance is different in AD patients compared with normal subjects, not only in terms of quantity but also quality. In the multiple regression model unadjusted for clinical severity, age is not correlated to MMSE score, indicating that older AD patients are not more cognitively impaired than younger. Furthermore, adjusting for clinical severity, we found a positive relation of age with MMSE score, suggesting a more profound cognitive disruption in younger AD patients. This finding is in agreement with previous reports of increased aggressiveness of AD in young subjects5,6 and a more severe language impairment in early-onset patients.7 Education was not significantly correlated to MMSE in the model adjusted for CDR and age, probably because the education effect is overcome by that of CDR. The study indicates that the interaction of cognitive performance with age and education observed in normal populations is not valid in the course of dementia. Furthermore, data seem to support the view that age is an important source of cognitive heterogeneity in Alzheimer's disease.