Premium
[P4–535]: ATROPHY IN DISTRIBUTED BRAIN NETWORKS CORRELATES WITH PERFORMANCE ON MEMORY TESTS IN AD PATIENTS
Author(s) -
Buss Stephanie,
Fried Peter J.,
Padmanabhan Jaya,
PascualLeone Alvaro
Publication year - 2017
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1016/j.jalz.2017.07.697
Subject(s) - temporal lobe , psychology , recall , neuroscience , atrophy , hippocampus , neuropsychology , episodic memory , default mode network , cognition , cognitive psychology , medicine , pathology , epilepsy
Charlestown, MA, USA; Massachusetts General Hospital, Boston, MA, USA; Department of Neurology, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA; VU University Medical Center, Amsterdam, Netherlands; Department of Epidemiology and Biostatistics, VU University Medical Center, Amsterdam, Netherlands; Alzheimer Center andDepartment of Neurology, VUMedical Center, AmsterdamNeuroscience, Amsterdam, Netherlands; Alzheimer Center and Department of Neurology, VU University Medical Center, Amsterdam Neuroscience, Amsterdam, Netherlands. Contact e-mail: Lrabin@brooklyn.cuny.edu IMAS/DMAS 271 117 106 72 34 MYHAT 2093 85 69 41 28 Background: In its conceptual framework for research on SCD in Alzheimer’s disease (Jessen et al., 2014), the Subjective Cognitive Decline Initiative (SCD-I) working group highlighted the lack of common assessment procedures, which hampers the SCD construct and ability to compare research across studies and settings. To address this critical problem, we aim to link measures of subjective cognition using item response theory (IRT) across international aging studies.Methods:We combined item-level subjective cognitive data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI), Monongahela-Youghiogheny Health Aging Team (MYHAT), Indiana/Dartmouth Memory and Aging Study (IMAS/ DMAS), and Harvard Aging Brain Study (HAB). We performed pre-statistical harmonization by assigning each item to one of eight cognitive domains by three independent raters. We dichotomized items due to limited or absent responses in some response categories and dropped items: (1) where fewer than 5 participants fell in a dichotomized response category; (2) that demonstrated a common factor loading qualitatively different from theoretical expectations; and (3) with arbitrarily large common factor loadings as evidence of local dependency. Multidimensional (bifactor) models consistent with IRT capturing a general SCD factor and specific cognitive domains were estimated using a robust maximum likelihood estimator with probit link function. We designated ADNI, with post-stratification weights to normalize the distribution by baseline diagnostic group, as the reference