Objectives: To investigate the validity of previously suggested dimensions underlying the Mini-Mental State Examination (MMSE) and differences in associations of these dimensions with sociodemographic and health characteristics in an older Latin-American community sample with low levels of education.
Design: Secondary analysis of baseline data from a population-based cohort study.
Setting: Bambuí, Brazil.
Participants: Of 1,742 total residents aged 60 years or older, 1,558 (89.4%) participated at this study.
Measurements: A standard Brazilian version of the MMSE.
Results: A five-factor solution (Concentration, Language/Praxis, Orientation, Attention, and Memory) for the MMSE was generated from Principal Components Analysis, and the five-factor solutions proposed in previous studies of developed nation samples were tested in this sample by Confirmatory Factor Analysis. In the adjusted linear regression models, MMSE factors varied in their correlates: for example, female gender was associated with higher concentration, orientation, and attention but lower Language/Praxis; increased age was inversely associated only with language and attention; and activity of daily living impairment was principally associated with lower Language/Praxis.
Conclusion: This study provides support for the cross-sectional equivalence of the MMSE, suggesting that most of the items and underlying constructs remain meaningful after alteration and translation in a low-education sample with lower overall distribution of scores.