Razi Journal of Medical Sciences

Background: Longitudinal studies are widely used in medical and social sciences. According to repeated measurements in these studies, independence assumption is not observed and therefore suitable models should be selected. In this study, application of marginal and transition models for analyzing the longitudinal data related to hygroscopic expansion of composite is shown. 

Methods: In this longitudinal study, laboratory data from three common composites (         ,   ) is used. These composites were kept in two different environments (distilled water and natural mouth saliva) for three months and their cylinder length was measured 18 times after preparation of 540 samples. Statistical analysis was done using marginal and transition model and programming with R software.

Results: Results show that in marginal model, composite effect was significant (p=0.025, p=0.045) and in transition model, both composite (p<0.001, p=0.037) and environment (p=0.046) were significant.

Conclusions: Results show that transitional model can be a good alternative to common methods of analyzing longitudinal data. As composite was significant in both models, it is better to use suited composites which have lower hygroscopic expansion during the time (like    ).