SciELO - Scientific Electronic Library Online

vol.62 issue2AUrban prevalence of epilepsy: populational study in São José do Rio Preto, a medium-sized city in BrazilCognitive dysfunction in children with sleep disorders author indexsubject indexarticles search
Home Pagealphabetic serial listing  

Services on Demand




Related links


Arquivos de Neuro-Psiquiatria

Print version ISSN 0004-282XOn-line version ISSN 1678-4227

Arq. Neuro-Psiquiatr. vol.62 no.2a São Paulo June 2004 

The role of education in mini-mental state examination: a study in Northeast Brazil


O papel da educação no mini exame do estado mental: um estudo no Nordeste do Brasil



Paulo Roberto de Brito-MarquesI; José Eulálio Cabral-FilhoII

IBehavioral Neurology Unit, Department of Neurology, Faculty of Medical Sciences University of Pernambuco, Recife PE, Brazil
IIInstituto Materno Infantil de Pernambuco (IMIP), Recife PE, Brazil




BACKGROUND: There is evidence that schooling can influence performance in cognitive assessement tests. In developing countries, formal education is limited for most people. The use of tests such as Mini-Mental State Examination (MMSE), could have an adverse effect on the evaluation of illiterate and low education individuals.
OBJECTIVE: To propose a new version of MMSE as a screening test to assess Illiterate and low education people.
METHOD: A study was carried out enrolling 232 individuals, aged 60 or more of low and middle socio-economic classes. Three groups were studied: Illiterate;1-4 schooling years; 5-8 schooling years. The new version (MMSEmo) consisted of modifications in copy and calculation items of the adapted MMSE (MMSEad) to Portuguese language. The maximum possible score was the same in the two versions: total, 30; copy, 1 and calculation, 5.
RESULTS: In the total test score ANOVA detected main effects for education and test, as well as an interaction between these factors: higher schooling individuals performed better than lower schooling ones in both test versions; scores in MMSE-mo were higher than in MMSE-ad in every schooling group.
CONCLUSION: Higher schooling levels improve the perfomance in both test versions, the copy and calculation items contributing to this improvement. This might depend on cultural factors. The use of MMSE-mo in illiterate and low school individuals could prevent false positive and false negative cognitive evaluations.

Key words: mini-mental state examination, cognition, education, cognitive assessement.


CONTEXTO: Existem evidências de que a escolaridade pode influenciar o desempenho em testes de avaliação cognitiva. Já que em paises subdesenvolvidos o nível educacional da maioria da população é baixo, isso poderia prejudicar resultados de avaliação por meio de testes. Assim é oportuno adequar o mini exame do estado mental (MEEM) a populações de baixa escolaridade.
OBJETIVO: Propor nova versão do MEEM como um teste geral para avaliar indivíduos analfabetos e com baixa escolaridade.
MÉTODO: Foram estudadas 232 pessoas de ambos os gêneros com 60 ou mais anos de idade, de classes socioeconômicas média e baixa. Foram considerados 3 grupos: analfabetos; 1-4 anos e 5-8 anos de escolaridade. A nova versão (MEEM-mo) consistiu de modificações nos itens cópia e cálculo do MEEM adaptado para a língua portuguesa (MEEM-ad). O escore máximo possível foi o mesmo nas duas versões: total 30 pontos; cópia, 1; cálculo, 5 pontos.
RESULTADOS: No escore total, o teste de ANOVA detectou efeitos principais para teste e escolaridade, assim como interação entre estes fatores: indivíduos com escolaridade mais alta realizaram melhor ambos os testes do que aqueles com mais baixa escolaridade. Os escores do MEEM-mo foram mais elevados do que os do MEEMad, em cada grupo de escolaridade.
CONCLUSÃO: Indivíduos com maior escolaridade apresentam melhor performance em ambas as versões dos testes; os itens cópia e cálculo foram responsáveis por este resultado. Isto pode depender de fatores culturais. O uso do MEEM-mo em indivíduos analfabetos e com baixa escolaridade pode prevenir resultados tanto falsopositivos como falsonegativos nas avaliações cognitivas.

Palavras-chave: mini exame do estado mental, cognição, educação, avaliação cognitiva.



Assessment of cognitive function is essential for accurate diagnosis and management in general neurological practice. Detailed assessment by the neurologist and psychologist requires a high degree of specialist training and is time-consuming. It is desirable to have a standardized simple and quick test of cognitive function which could be routinely used by the admiting physician1. Among the cognitive tests there is the MiniMental State Examination (MMSE) which was proposed to differentiate non specified organic brain syndrome and depression from normal patients. It is useful in quantitatively estimating the severety of cognitive impairment and in documenting serially cognitive changes2. It can be also used in community-based epidemiologic studies to evaluate the current mental functioning of participants. Assessing MMSE in research protocols is particularly important for establishing the incidence or prevalence of cognitive impairment and for examining correlates of mental status3. Since the general practitioners are usually the prime mover in ensuring that medical and social needs of their patients must be recognized and met, it seems worth determining how well informed they are about the health and welfare of their older patients4. However, there is evidence that clinicians have often difficulty in recognizing disturbances of cognition among their patients5. Williamson et al.4, surveying an elderly population, found that only 13% of demented subjects were recognized as such by the physicians in charge of their care. Another study showed that 37% of cognitively impaired general medical ward patients were unidentified by physicians, while nurses failed to identify 55% and medical students, 46%6. Other authors, studying 33 patients with cognitive impairment on a neurologic inpatient service, found that 30% of them were not recognized by the attending neurologists, and 87% by general practitioners as well7. Therefore, is low for neurologists, the capacity to recognize and diagnose these disorders. Evidence that psychiatric disorders are common among neurological patients and that they are frequently unrecognized would support the need for further psychiatric training of neurologists. Potencially such failure to recognize cognitive impairments has considerable implications for patient care, as such changes are often the first indication of many underlying neuropathologic conditions7. Considering psychiatrist's standardized clinical diagnosis as a criterion, the MMSE is 87% sensitive and 82% specific in detecting dementia and delirium among hospital patients on a general medical ward5. The false positive ratio is 39% and the false negative ratio is 5%. All false positive people had less than 9 years of formal education. So, the MMSE has been shown to lack specificity in poorly educated persons, particularly those who did not attend high school5. Conversely, it has been observed a number of highly educated dement patients whose MMSE scored the conventional normal range8. It is possible that performance on specific MMSE items could be related to education. In fact, several studies have shown that low MMSE scores are correlated with low instructional level and poorly educated subjects score below 26 even without dementia. On the other hand equations have been developped at adjusting MMSE scores for educated people, because specific items of the MMSE are influenced by education10. So the use of tests which include problems beyond the comprehension capacity of an individual could mask the evaluation of this individual's insight for the same test. Items measuring recall of three words, pentagon copy, and orientation in time seem to be most sensitive to both normal aging and dementing illnesses11. According to Fratiglioni et al.12, the performance on the MMSE is variable among countries. This may be due to different degrees of education in the population, or to different curricula structures of the same schooling level in different countries. Thus, ";norms"; or ";adjustment techniques"; derived from one country can not be considered universally applicable. In order to prevent the possibility of education to mask the result of this test and, as a consequence, to represent a potential risk factor for the cognitive assessment, it is opportune to reassess the validity of the MMSE whenever it is going to be used in a new population. In developping countries, such as Brazil, most of people have a schooling level lower than eight years13.

The main aim of the present study was to propose a MMSE modified version as a screening test to assess low schooling old people in Brazil.



A randomized cross-sectional study was performed with 232 individuals of both genders, aged 60 or more years old (69.4 ± 6.8 years), belonging to low and middle socioeconomic classes, residents in Olinda city, Pernambuco, Brazil. Previously to applying the test people were inquired about their normal daily routine. People able to conduct theirselves and recognize the primary and secondary colors, the time in a clock, cash money, and capable to use a tin opener were admitted to the study. A clinical interview was proceeded in order to investigate neurological and psychiatric diseases. People with low visual or auditory acuity, motor or rheumatic disturbance, chronic alcoholism, cardiovascular disease, recent head trauma (last 12 months) or not motivated, were excluded. Three groups were formed according to the schooling level, as follows: group illiterate (n=28)-individuals without any formal schooling instruction; group 1-4 degree (n=119)-individuals from 1 to 4 years of formal instruction; and group 58 degree (n=85)-individuals from 5 to 8 years of formal instruction.

The original MMSE2, adapted to the Portuguese language14 (thereafter called MMSE-ad) was compared to a modified version of it (thereafter called MMSE-mo) (see appendix). The modification consisted of: first, the copy of intersecting apendice pentagons (MMSE-ad) was changed to the copy of an intersecting equilateral triangles (MMSE-mo); second, the subtraction 7 from 100, and then 7 from the result, repeated 5 times (MMSE-ad) was changed to the serial 1 subtracted from 25 backward, and then 1 from the result, repeated 25 times (MMSE-mo). Each subject of the sample was submitted consecutively to the two versions, first MMSE-ad, and second MMSE-mo. Concerning to the copy and calculation items the MMSE-mo version was applied immediately after the same items of the MMSE-ad version. Both test versions had the maximum possible score equivalent to 30 points, copy and calculation items having maximum score equivalent to1 and 5, respectively.

Aggreement between MMSE-ad and MMSE-mo tests concerning the scores obtained in copy item was determined by Cohen's Kappa aggreement coeficient. Aggreement between the tests concerning the scores obtained in calculation item was determined by Kappa weighted aggreement coeficient15. The chi square test was used to compare frequencies. Comparisons between means of the total scores achieved in each test were made by a twoway ANOVA for repeated measures. The Spjotvoll-Stoline test was used for posthoc multiple comparisons of means. Comparisons of mean values not fulfilling parametric requirements were made by KruskalWallis analysis of variance followed by Dunn test. The null hypothesis was rejected when p<0.05. The statistic software Sigma Stat 2.0 was used for calculation.



In the copy item (Table 1) a poor agreement between MMSE-ad and MMSE-mo was observed in 1-4 years degree (K=0.14) and in 5-8 years degree (K= 0.18) groups. The agreement for the illiterate group was fair ( K= 0.35).



Like in copy item the agreements between the tests in calculation scores (Table 2) were poor in 1-4 and 5-8 schooling degree (K= 0.16 and K= 0.10, respectively) and fair for illiterates (K= 0.39).



From the results expressed in Table 1, we can observe also that global aggreement (g.a) for the copy ite - defined as the addition of WW plus CC answers of the same individual between MMSEad and MMSEmo tests, presented the following proportions: illiterate 21/28 (75%), 1-4 degree 59/119 (49%), and 5-8 degree 42/85 (49%) (Chi-square= 6.41, p=0.040). For the calculus item (Table 2) g.a showed the following proportions: illiterate 11/28 (39%), 1-4 degree 23/119 (19%), and 58 degree 19/85 (22%) (Chi-square= 7.09, p=0.029). The g.a for copy and calculation between 1-4 and 5-8 degree did not differ significantly. However, each one of these groups differ from illiterate (p<0.01) and (p<0.01), respectively.

Comparing the total scored values between the tests (Table 3) ANOVA indicated a significant main effect for test (p < 0.001) and for schooling degree (p < 0.001) as well as an interaction between these two factors (p = 0.007). The values were significantly higher for MMSE-mo than for MMSE-ad, in every schooling degree. Moreover, in both tests illiterates scored lower than literates ones (p<0.01).



In Table 4, we can observe that in every schooling degree MMSE-mo produced higher proportions of individuals answering correctly to copy than MMSE-ad: illiterate, (p<0.034); 1-4 years, (p<0.001); and 5-8 years, (p<0.001). In both test versions, a significant difference was also observed among schooling groups: MMSE-ad (Chi-square= 6.095; p=0.047) and MMSE-mo (Chi-square= 12.178; p= 0.002). In both test versions the illiterate groups presented lower proportions of correct answers than the literate ones.



Mean scores for calculation (Table 5) indicate that, in MMSE-mo, individuals performed better than in MMSE-ad, for all schooling degrees: Illiterate (p=0.011), 1-4 years (p<0.001) and, 5-8 years (p<0.001). For each test in particular, significant difference among schooling degrees was detected: (MMSE-ad, p< 0.018); (MMSE-mo, p< 0.001). By comparing schooling groups we can verify that, in MMSE-ad, the first group (illiterate) is similar to the second one, and differ of the third schooling group (p<0.05). However, the second group does not differ from the third group, the higher score having been verified in this last group. In the MMSE-mo, we observe that the illiterate scored lower than 1-4 and 5-8 degrees (p<0.05), but these two last groups did not differ between them.




The very low (poor) or low (fair) agreements according to Kappa coefficients15 verified between the scores of MMSE-ad and MMSE-mo in the copy and calculation items, here reported, indicates that the test modification induced changes in the ability of the individuals to solve the proposed tasks. However, the higher agreement for illiterate group compared to schooling instructed (both in Kappa coefficient and in observed percent agreement) suggest that the influence of the test modification is lesser in illiterate individuals. In addition, the comparisons between mean values of the total score obtained in the two test versions reveal that these changes correspond to an improved performance when the individuals are submitted to MMSE-mo. Of course, the total MMSE-mo score refleted the alterations introduced in the copy and calculation items, because the other items were not modified. Indeed, the higher proportion of subjects correctly designing the triangles, as well their better performance in calculus in MMSE-mo supports this point. Since this improvement was observed in every schooling degree, including in illiterate individuals, it is likely the highest MMSE-mo copy achievement to be associated with basic knowledge of culture features in which they have lived.

In Northeastern Brazil some of these features, such as recognizing a triangle, is easier than recognizing a pentagon. In their daily activity the people routinely deal with triangle shapes (popular musical instruments, objects for playing behavior), but not with pentagon shapes16. This recognition might be even more difficult when the overlapping design of the figures in the test is required. So the cultural background acquired out the school can improve their capacity to solve the problem, because a familiar figure is included in MMSE-mo but no in MMSE-ad. These findings indicate that informal knwoledge about geometric shape is an important way to face the test challenge. On the other hand, the association of calculation and praxia with education suggest that the improved performance of the highest schooling groups in both test versions is accounted for by educationdependent skills.

There is mounting evidence that education has an effect on testing cognitive state in different culture groups5,8-12. So in our population-based sample, education and culture background could also influence responses to the calculation question when changing subtracting serial sevens to subtracting serial one. Further our data show that differences observed between MMSE-ad and MMSE-mo are related not only to the test structure or to the schooling degree working as independent factors (indicated by their main effects), but also to an active association of them (indicated by the interaction effect). Therefore we can admit these factors influencing each other by inducing a better performance in MMSE-mo. In conclusion, the present data suggest that social and psychological factors contribute substantially to cognitive test scores and emphasize the importance of detailed assessment procedures for aplication of MMSE-ad and MMSE-mo tests as clinical instruments. In this way an individual belonging to a non afluent population could be adequately assessed in its insight capacity without damaging its ability at solving the problems presented in the test.

Therefore the MMSE new version here developed supports the view that the criterious use of both test versions (applying them according to social and educational groups) could prevent falsepositive and false-negative evaluations.



1. Dick JPR, Guiloff RJ, Stewart A, et. al. Minimental state examination in neurological patients. J Neurol Neurosurg Psycchiatry 1984;47:496499.        [ Links ]

2. Folstein MF, Folstein SE, McHugh PR. "Minimental state": a practical method for grading the cognitive state of patients for the clinician. J Psychiat Res 1975;12:189-98.        [ Links ]

3. Magaziner J, Bassett SS, Hebel JR. Predicting performance on the minimental state examination. J Am Geriatr Soc 1987;35:996-1000.        [ Links ]

4. Williamson J, Stokoe IH, Gray S, et al. Old people at home: their unreported needs. Lancet 1964;1117-1120.        [ Links ]

5. Anthony JC, LeResche L, Niaz U, VonKorff MR, Fosltein MF. Limits of the "minimental state" as a screening test for dementia and delirium among hospital patients. Psychol Med 1982;12:397-408.        [ Links ]

6. Knights EB, Folstein MF. Unsuspected emotional and cognitive disturbance in medical patients. Ann Intern Med1977;87:723-724.        [ Links ]

7. De Paulo JR, Folstein MF. Psychiatric disturbances in neurological patients: detection, recognition, and hospital course. Ann Neurol 1978;4:225-228.        [ Links ]

8. Uhlmann RF, Larson EBL. Effect of education on the minimental state examination as a screening test for dementia. J Am Geriatr Assoc 1991;39:876-880.        [ Links ]

9. Fillenbaum GG, Hughes DC, Heyman A, et al. Relationship of healht and demographic caracteristics to minimental state examination score among community residents. Psychol Med 1988;18:719.        [ Links ]

10. Escobar JI, Burnam A, Karno M, et al. Use of the minimental state examination (MMSE) in a community population of mixed ethnicity. J Nerv Ment Dis 1986;174:607-614.         [ Links ]

11. Tombaugh TN, McIntyre NJ. The minimental state examination: a comprehensive review. J Am Geriatr Soc 1992;40:922-935.        [ Links ]

12. Fratiglioni L, JormAF, Grut M, et al. Predicting dementia from the minimental state examination in an elderly population: the role of education. J Clin Epidmiol 1993;46:281-287.         [ Links ]

13. Instituto Brasileiro de Geografia e Estatística. Anuário estatístico do Brasil contagem. Rio de Janeiro: IBGE: 1997 Tabela 2.98,p.2-165.        [ Links ]

14. Bertolucci PHF, Brucki SMD, Campacci SR, et al. O mini-exame do estado mental em uma população geral: Impacto da escolaridade. Arq NeuroPsiquiatr 1994;52:1-7.        [ Links ]

15. Altman DG. Practical statistic for medical research. London:Chapman & Hall, 1991:611.        [ Links ]

16. Chacon D. A criança e o jogo: estudo psicossocial do comportamento lúdico da criança do Recife. Recife: Secretaria de Educação e Cultura de Pernambuco, 1959:11-24.        [ Links ]



Received 6 June 2003, received in final form 1 October 2003.
Accepted 18 November 2003.



Dr. Paulo Roberto de BritoMarques Rua Sport Clube do Recife 280/616 50070450 Recife PE Brasil. E-mail:




Creative Commons License All the contents of this journal, except where otherwise noted, is licensed under a Creative Commons Attribution License