Longitudinal Methodology Series III – ‘Are time in study and age exchangeable as metric of time when modelling change in age heterogeneous longitudinal studies?’ – Dr Graciela Muniz-Terrera, MRC Lifelong Health and Ageing Unit at UCL
In cognitive ageing research, investigators often model cognitive decline fitting longitudinal models that describe change as a function of some metric of time. For instance, when modelling preclinical dementia trajectories, dementia diagnosis is often used to define the time metric and change is modelled as a function of time to diagnosis. When interested in understanding changes before death, the metric of time often considered is time to death. Researchers using these metrics of time assume that changes in the process being modelled is driven by the event used to define the time metric (dementia or death in the previous examples).
Yet, most often, the focus of ageing research is developmental change. In this context, cognitive changes are often modelled as a function of time in study or age and both metrics of time are used almost interchangeably in the literature.
In this presentation Dr Muniz-Terrera discussed some issues that arise when modelling change as a function of time in study and as a function of age.
She illustrated the different points modelling change in perceptual speed scores from participants of a longitudinal study of cognitive ageing.