RESUMO
In medical research, longitudinal studies are widely used to describe the disease process and effectiveness of treatment. In these studies individual outcomes are recorded in different sections. Many of the medical findings are characterized by ordinal scale. Because of the relationship between ordinal responses and the structure of longitudinal studies, special statistical methods are used. In this paper, we describe and compare different statistical methods, and recommend the most appropriate method to analyze the effect of drug therapy for psychologically disturbed patients with oral lichen planus. In this study, the data of a clinical trial on 45 patients with oral lichen planus conducted in Mashhad Dental School was used. The outcome was defined as an ordinal scale using cumulative logit link function. Data was analyzed by fixed and random effects model. The results showed that random effect models were more accurate than the fixed effects model. Time and mental disorder showed significant effects on both the random effects model and fixed effects model. Treatment had no significant effect on random effects model; while it exhibited a significant effect on the fixed effects model. To analyze longitudinal ordinal data, random effects model is recommended rather than fixed effect model