Screening for impaired cognitive function among the elderly in Spain: reducing the number of items in the Short Portable Mental Status Questionnaire.

The objectives of this work are to assess the agreement in classifying the 10th percentile of the population with impaired cognitive function resulting from different combinations of items in the Short Portable Mental Status Questionnaire, to evaluate the effect of literacy status on the new reduced versions, and to propose items less biased by education. Three data sets of representative samples of non-institutionalized elderly (65-year-olds and older) living in urban and rural communities in Spain were analyzed. Cognitive function was assessed using either the original or a modified Spanish version of the Short Portable Mental Status Questionnaire (SPMSQ). Different combinations of items were excluded from the scale and for each resulting reduced scale the lower 10th percentile of the distribution assessed a cut-off score. Cohen's kappa was used to test the agreement between the whole scale and the reduced scales in identifying the lowest 10th percentile of people with cognitive impairment. To test the effect of literacy status, the three samples were combined. New distributions were obtained for reduced versions of the scale including the items with the highest correlation with cognitive impairment controlling for literacy status. All three samples follow a similar distribution of errors. Most kappa values obtained when excluding one or more items from the original scales were between 0.80 and 0.95, while proportion of agreement varied between 94.2% and 100%. Lack of agreement is highest when eliminating all items that represent one dimension. Shorter versions of the SPMSQ are as reliable as the complete 10-item version in identifying those at risk for cognitive dysfunction. Cut-off points differed by literacy status but remained stable regardless of the number of items included.