Objective: To compare logistic and bilogistic models to describe the pattern of cognitive decline in the preclinical phase of Alzheimer disease (AD).
Methods: We conducted mixed effects modeling of Mayo Cognitive Factors Scores to determine the longitudinal pattern of cognitive decline in the period 10 years prior to and 5 years following a clinical diagnosis of AD. Our analysis included 199 people that eventually received a diagnosis of clinically probable AD. Participants had at least two neuropsychological evaluations including one before the evaluation at which they received the AD diagnosis.
Results: A bilogistic model, including terms for a plateau in the course of cognitive decline, better fit longitudinal memory scores than a simple logistic model. On average the plateau began about 4 years prior to the clinical diagnosis of AD and ended with a decline that probably contributed to the clinical diagnosis of AD. A similar plateau was not evident in four other cognitive domains.
Conclusions: The current findings may support proposed compensatory hypotheses involving redundant memory systems, up-regulation of neurotransmitters, or recruitment of other neural networks.