Abstract:
Many longitudinal studies generate a dataset having two or more longitudinal repeated biomarkers measurement, which often depend on each other. For example, in Gestational hypertension study the two important markers, gestational systolic blood pressure (GSBP) and diastolic blood pressure (GDBP) are collected simultaneously from a pregnant woman every visit time. In such studies, evolution of the biomarkers over time and the association between them are commonly of interest. Often Univariate analyses using a mixed effects model are performed and are well developed for a single outcome variable. However, separate models are overly simplified because they do not consider the association between two components of such data and so produce misleading conclusions. In this study, we propose a joint random-effects model which enables two or more longitudinal repeated biomarker measurements to be modeled together while taking account of association between them. We apply these methods to a pregnancy induced hypertension among antenatal care follow up pregnant woman in Jimma University specialized hospital. The aim of the analysis was to determine joint evolution and association of pregnancy induced systolic and diastolic blood pressure over time and determining their associated risk factors. The association among the two sequences is captured by correlated normal random effects included to account correlation between two outcomes. Besides, correlation of error terms is given a great consideration. Both Separate and joint modeling results are consistent. But, fit statistics shows that joint modeling with uncorrelated error is the best to fit the data. Under joint analysis, two aspects of the relation were investigated: the association between the evolutions and the evolution of association. Results of the joint model suggested a very strong association between the evolutions of GSBP & GDBP and a slowly decreasing evolution of the association over gestational age. Sex of fetus, family history of pregnancy induced hypertension, gestational age, age of mother and number of Gravida are identified as associated risk factors. Joint model is able to address the same questions as separate model with more accuracy by addressing additional questions that may be of great interest to the researcher, such as the association of evolution and the evolution of association of the responses.
