Longitudinal assessment of discriminability and response bias in Alzheimer's disease

The longitudinal progression of discriminability and response bias was examined using archival data from several studies on the effects of disease on cognitive function. Of primary interest was whether disease severity predicted the trajectory of discriminability and response bias in Alzheimer's disease (AD). Random effects modeling facilitated the identification of clinical, behavioral, and neuroanatomical correlates of discriminability and bias.; Data from a longitudinal cohort followed at a university medical center was used to examine the progression of discriminability and response bias. The baseline cognitive status of these individuals varied from normal, to mildly impaired, to demented; and at the time of the study all were diagnosed as probable AD. Discriminability for this cohort declined over the course of the follow-up period, independent of age at baseline. Level of cognitive impairment and a measure of premorbid verbal ability predicted discriminability at the initial assessment, but did not explain change in discriminability over time. Response bias declined slightly over time independent of age at baseline, shifting from a neutral to a more liberal bias. However, bias was not predicted by level of cognitive impairment.; To examine the relationship between brain structural magnetic resonance imaging (MRI) variables and discriminability and response bias, a mixed sample was drawn from patients recruited for a longitudinal project on the joint contribution of cerebrovascular disease and AD to cognitive impairment and dementia. The cognitive ability of these individuals ranged from normal to demented. At the first assessment, increased hippocampal (HC) volume predicted better discriminability and a less liberal response bias and increased white-matter hyperintensity (WMH) volume predicted poorer discriminability and a more liberal bias. Reduced initial HC volume was associated with greater decreases in discriminability over time; moreover, greater change in HC volume over time was associated with greater decline in discriminability. Reduced initial volume of HC and cortical gray matter (CGM) predicted a more liberal bias, but not change in response bias. Findings are discussed in terms of the extant cross-sectional literature on discriminability and response bias in AD and the neuroanatomical structures thought to subsume both behavioral measures in various groups.