|
Abstract:
|
Many statistical challenges exist in longitudinal studies of cognition. For example, cognitive decline is associated with debilitating health consequences, such as death, making the drop-out mechanism informative. Further, participants experiencing cognitive decline might be more/less likely to agree to neuropsychological testing making the observation times informative. In many longitudinal settings a linear mixed effects model is used to estimate cognitive decline, but this method ignores informative drop-out and informative observation times leading to spurious and biased associations. Recently, joint models have become a promising solution to these issues by combining the linear mixed effects model with a Cox proportional hazards model, to take into account informative drop-out, and a frailty model, to take into account informative observation times. In our study we propose a novel joint model specifically tailored to modeling cognitive decline.
|
