Identifying risk for functional impairment using cognitive measures: An application of CART modeling

The relationship between cognitive test performance and functional status was assessed in 289 persons with a cognitive impairment diagnosis and in 307 controls from the Alzheimer's Disease Registry at Mayo Clinic. Classification and regression tree (CART) procedures were used to create decision rules (using cutoff scores) to identify persons most at risk for functional impairment. The results suggest that scores from the Dementia Rating Scale, the Mini-Mental State Examination (MMSE), the Perceptual-Organizational factor of the Weschler Adult Intelligence-Scale-Revised (WAIS-R), and age and education may be used in combination to identify persons at risk for functional disability. Decision rules to predict performance in specific functional domains are also presented. The researchers concluded that CART has several advantages relative to traditional linear methods in establishing cognitive-functional relationships.