A-16 Latent Item Response Theory Regression Using Neuropsychological Tests to Predict Functional Ability

Abstract

Abstract Objective Examine prediction of functional ability with neuropsychological tests using latent item response theory. Method The sample included 3155 individuals (Mage = 69.72, SD = 9.41; Median education =13.15, SD = 4.40; white = 92.81%; female = 62.03%; MCI = 25.13%; Dementia = 28.87%) from the Texas Alzheimer’s Research and Care Consortium who completed functional and cognitive assessments [Mini Mental State Examination (MMSE), Logical Memory (LM), Visual Reproduction (VR), Controlled Oral Word Association Test (COWAT), Trail Making Test (TMT), Boston Naming Test, and Digit Span]. Functional measures [Clinical Dementia Rating Scale, Physical Self Maintenance Scale, and Instrumental Activities of Daily Living)] were combined into a single outcome variable using confirmatory factor analysis. Item response theory (IRT) was used to fit the data, and latent regression to predict the latent trait score using neuropsychological data. Results All three functional scales loaded onto a single factor and demonstrated good construct coverage and measurement reliability (Supporting Figure). A graded response IRT model best fit the functional ability composite measure. MMSE (b = −1.08, p < .001), LM II (b = −0.58, p < .001), VR I and II (b = −0.09, p = .02 and b = −0.43, p < .001, respectively), COWAT (b = −0.10, p = .003), and TMT-B (b = −0.30, p < .001) all significantly predicted functional abilities, as did age (b = 0.61, p < .001) and education (b = 0.31, p < .001). Conclusions Global cognition, memory and executive function tests predicted functional abilities while attention and language tasks did not. These results suggest that certain neuropsychological tests meaningfully predict functional abilities in elderly cognitively normal and cognitively impaired individuals. Further research is needed to determine whether these cognitive domains are predictive of functional abilities in other clinical disorders.