Microperimetry(MP) involving testing of multiple macula locations, is a visual field test for measuring macular sensitivity. The macular sensitivity is estimated as the mean sensitivity(MS) of all locations. Macular sensitivity may be used as an endpoint in clinical trials. For Stargardt disease (inherited juvenile macular degeneration), however, prior data have shown that MS is not a sensitive outcome measure for Stargardt trials. A clinical hypothesis is that locations where macular lesions have developed and are likely to expand, are locations where sensitivity is mostly like to decline. We are interested in using the point-level data to characterize the points that are mostly likely to lose sensitivity and to estimate their sensitivity changes. The data structure can be described as bivariate longitudinal multivariate outcomes, involving repeatedly MP tests for 68 points in each eye. With the complicated correlations, simple analysis using GEE or random effects model cannot address the clinical hypothesis. We use a hybrid modeling strategy based on Markov transition model together with pairwise composite likelihood for inference.