Application of finite mixture models to explore subpopulations in Crohn's disease patients.

Abstract

Background: Inflammatory bowel disease (IBD) commonly refers to ulcerative colitis (UC) and Crohn disease (CD), which are chronic inflammatory diseases of the gastrointestinal (GI) tract of unknown etiology. This study has been conducted to examine whether there are different components in the data, and if these components related to the treatment and the Inflammatory bowel disease (IBD) score at baseline.

Methodology: This is a clinical study which consisted of 291 subjects, who divided over four treatment arms and were measured during a seven-week period. The number of weeks in the period Week 0ne through Week seven was considered as the outcome of interest, as well treatment and IBD score at baseline were considered as predictors. Different statistical methods such as explanatory data analysis and finite mixture model were employed to explore the outcome of interest.

Results: From the finite mixture model, two components were obtained. Most of the patients, 196(67.4%), were classified in the first component (P1). The deviance for single component of the mixture model corrected for the covariates was 1049.3 and that of the two components was 948.8. The effect of ibdsc0 was significant in both subpopulations with p-value = 0.0001 for subpopulation1, and p-value = 0.0422 for subpopulation2, and Exp (0.01) = 1.01 and exp(0.087) = 1.09 are the amounts by which the mean count (µ) is multiplied per unit change in the ibdsc0 for subpopulations 1 and 2, respectively.

Conclusions: The two components are not related to the treatment, and as a result, the treatment does not completely explain the presence of potential clusters in the outcome. Ibdsc0 partially explains the presence of potential clusters in the outcome.