Abstract: In the interdisciplinary realm of statistics, genetics, and epidemiology, longitudinal sibling pair data offers a unique perspective for investigating complex diseases and traits, allowing the exploration of the dynamic processes of gene expression over time by controlling numerous confounding factors. Missing-not-at-random (MNAR) data are commonly used in such types of studies, but no statistical methods specifically tailored have been developed to handle MNAR data in complex longitudinal data in the literature. Here, we propose a new statistical method by jointly modeling longitudinal data from sib-pairs and MNAR data. Extensive simulations demonstrate the excellent finite sample properties of the proposed method.