Analyses for continuous, correlated outcomes in longitudinal studies
Bernard Rosner, Harvard Medical School, Boston, Massachusetts, United States; Harvard School of Public Health, Boston, Massachusetts, United States
DisclosureBlock: Bernard Rosner, None
Description
We will discuss the analysis of longitudinal ophthalmic continuous outcome data.
There are several types of correlation that need to be taken into account including:
(a) cross-sectional correlation between outcomes for fellow eyes at a specific visit
(b) longitudinal correlation between outcomes for the same eye over time
(c) longitudinal correlation between outcomes for one eye at a specific visit and the fellow eye at different visits
Both fixed effects, mixed effects, and GEE methods of analysis will be considered for this type of data.
Examples will be provided of change in visual acuity data from the CAPT study and the AREDS study, respectively.