Aim. To develop a test of cognitive performance in persons with moderate-to-severe dementia.Background. Various instruments are used to assess the course of dementia and to evaluate treatments in persons with dementia. Most neuropsychological assessments are inappropriate for measuring cognitive abilities in persons with severe dementia, because these persons perform at floor level in such measurements.Design. A cross-sectional research design.Methods. The test (Clinical Evaluation of Moderate-to-Severe Dementia; Swedish acronym: KUD) was developed from a pool of 25 test items with the final KUD consisting of 15 items. Reliability and validity were established using 220 subjects (with various dementia diagnoses) with scores of Mini-Mental State Examination between 0–20. Approximately two weeks after the first test, 116 of the original 220 subjects were retested.Results. A factor analysis with the 15-item scale revealed an interaction factor comprising three items and a cognitive performance factor with 12 items. The internal consistence reliability was 0·93 for the KUD (Cronbach’s alpha). Test–retest reliability was also high (0·92) and correlation between the KUD and the MMSE (≤20) was high (r = 0·80).Conclusion. The KUD seems to be a valid, reliable performance-based assessment scale for measuring cognitive performance in persons with MMSE score below 12 or 15 points.Relevance to clinical practice. It is of outmost interest that cognitive performance can be easily followed for persons with moderate-to-severe dementia in, for example, drug therapies and other therapies, but also in terms of treatment of and support to the person based on his or her abilities.
