Relation between education and dementia: the role of test bias revisited.

Several authors have suggested that dementia screening tests may be biased against low levels of education, whereas others find that a low level of education is a genuine risk factor for dementia. The present paper attempts to reconcile these conflicting views by examining item bias and test bias indices of the mini mental state examination (MMSE). Psychometric calculations and receiver operating characteristics (ROC) analyses of sensitivity and specificity as performed by earlier studies were replicated and extended from the database of the Amsterdam Study of the Elderly. This is a population survey on cognitive decline and dementia (age range 65-84). Subjects with a low level of education (primary school) were compared with better educated subjects (at least some secondary education). Cases were matched by age and sex. The results indicate that the MMSE is not educationally biased as far as item characteristics, reliability, and construct validity are concerned. Yet its predictive validity as a screening test for dementia is educationally biased. This bias will effectively be eliminated with a two point higher cut off score for the subjects whose education extends beyond primary school. Even after such score correction, a low level of education probably remains a genuine risk factor for dementia.