Construct validity of the mini mental state examination across time in a sample with low-education levels: 10-Year follow-up of the Bambú cohort study of ageing

Objective: The study aims to investigate whether longitudinal data on the structure of the mini mental state examination (MMSE) collected in an older Brazilian cohort support factorial invariance over time. Design: Analysis of 10-year data from a community-based cohort study was performed. Setting: The study took place in Bambú, Brazil. Participants: The study sample comprised 1558 (89.4%) of all eligible 1742 elderly residents. Measurements: A standard Brazilian version of the MMSE was used. Results: A five-factor solution (developed on the baseline of the cohort) either with no constraints or with loadings constrained to equality across time provided a reasonable fit for the MMSE. A comparison between both models suggested that the model with no constraints was superior. However, the five absolute goodness-of-fit indices suggest that the fully constrained model was also adequate and did not differ substantively from the model without any restriction. Conclusion: The structure of the MMSE remained relatively unchanged across the 10 measurement times, thus providing evidence for the good construct validity of the scale across time.