P3‐316: CALIBRATING LONGITUDINAL COGNITIVE PERFORMANCE ACROSS DIVERSE NEUROPSYCHOLOGICAL BATTERIES AND DATASETS

• DOI: 10.1016/j.jalz.2014.05.1409
• Publication Date: 2014-09-06T15:12:35Z
• Authors: Alden L. Gross; Shubhabrata Mukhe...; Richard Sherva; John Kauwe; Stephen Jeffrey N...; Leo Waterston; Richard Norman Jones; Robert Green; Paul Crane
• Keywords: Categorical variable; Cognitive Decline; Neuropsychological test

To identify the optimal approach for calibrating longitudinal cognitive performance across studies with different neuropsychological batteries to use as a common phenotype in genetic research on Alzheimer's disease (AD). We examined four approaches calibrate eight large of AD (N=10,997): (1) selecting test datasets, (2) standardizing and averaging together all tests, (3) confirmatory factor analysis (CFA) continuous indicators, (4) CFA categorical indicators. Datasets included Disease Neuroimaging Initiative (ADNI), Rush Memory Aging Project, Religious Orders Study, Cache County Study Aging, Myriad Tarenflurbil phase III clinical trial, AddNeuroMed, Adult Changes Thought, National Coordinating Center. compare precision, we determined minimum sample sizes necessary detect 25% decline 80% power using each approach. criterion validity, association change changes over up six years whole brain cortical thickness hippocampal volume available MRI data from ADNI joint process latent growth curve models. Summary measures were highly correlated (all r's>0.88). indicators required lowest size (N=253) compared (N=283), standardize average (N=300), (N=329) approaches. Associations strongest (Z=7.0; Z=5.2, respectively) (Z=6.2; Z=4.5), (Z=5.5; Z=3.7), (Z=6.1; Z=4.1) demonstrated best precision thus greater change. It also was more strongly biological markers than other This has wide applicability directly studies, making it good operational analyses AD.