Longitudinal time to event analyses use Cox proportional hazards models for dichotomous outcomes (e.g., incident AD, incident MCI) [92]. Recently, we have used models for discrete (tied) data [93]. Mixed models are used to examine the relation of risk factors to change in cognitive function over time [94]. These models assume that individuals follow the mean path of the group, conditional on a set of explanatory variables, except for random effects which allow for the initial level to be higher or lower and the rate of change to be faster or slower. Other paths of change can be modeled in other ways such as change point models [95].
