Longitudinal data analysis of fasting blood sugar level of adult diabetic patients In Jimma University Specialized Hospital 2006: An application of semi-parametric mixed model.

Abstract

Diabetes mellitus is a metabolic disorder of multiple aetiology characterized by chronic hyperglycaemia with disturbances of carbohydrate, fat and protein metabolism resulting from defects in insulin secretion, insulin action, or both. Fasting blood sugar level is a quantity used to detect the diabetes status of patients. There were several factors for the change in their fasting blood sugar level of adult diabetic patients. Objectives: The main aim of this study is to investigate the fasting blood sugar level of adult diabetic patient's overtime who have been under follow up from 2004 to July 2006 in Jimma University specialized hospital. Methods: This study used data obtained from a retrospective cohort follow up study from adult diabetic patients who have been under follow up from 2004 to July 2006 in Jimma University Specialized Hospital. All patients included in this study were those diabetic patients whose age is 18 years and above and who have been followed at least two times. This thesis deals with the regression analysis of repeated measurements taken at irregular and possibly subject-speci c time points. Therefore, we proposed a Semi-parametric Mixed-e ects model (SPMM) using smoothing spline to investigate the evolution of FBS overtime. Result and Discussion: From 534 adult diabetic patients, 200(35.27%) were females and 367(67.73%) were males. Twenty ve percent were type I diabetic and seventy ve percent were type II diabetic patients. The mean age of patients at baseline was 45.16 years and that of the mean fasting blood sugar level(FBS) level of patients at baseline was 204 mg/dl. In both linear mixed models and semi-parametric linear mixed models FBS level declines over time, but the predictive ability of the SPMM model is increased. The semi-parametric mixed model further suggested that the decline was non-parametric. Follow up time (p-value=0.000) and weights (p-value=0.000) are the only signi cant factors for the change of fasting blood sugar level. Conclusion and Recommendation: Semi-parametric mixed e ect model with quadratic time random e ects ts the data appropriately to the continuous measurement of fasting blood sugar level in mg/dL of adult diabetic patients.