Linear Mixed Effect Model with Application to Random Blood Sugar Data

Background aim: Diabetes mellitus is a common, chronic, metabolic syndrome characterized by hyperglycemia as a cardinal biochemical feature. Type1 diabetes is a continuing hormonal deficiency disorder that has significant shortterm impacts on health and lifestyle and is associated with major longterm complications like heart failure, kidney, hypertension, eye damage, etc. which reduced life expectancy. The main objective of this study was to assess the risk factor that increase prevalence of type1 diabetes mellitus and to determine their relationship with outcome of type1 diabetes mellitus over time. Methods materials: To address this objective linear mixed effect model was applied using the random blood sugar of 970 diabetic patient children during treatment period of 3 years at Hiwot Fana hospital which have been implemented in statistical packages STATA, SAS and R. #xD; Result: This study found that the mean progression of random blood sugar level of diabetic children was decreased over time after they starts their follow up and medications. The linear distribution also accounts 92 % variability of the data was explained by the covariates which were included in the study. The variable age, residence, family history, nutrition status, early diet, body mass index, electrolytes and renal function test had significant effect on the change of sugar level (p lt; 0.05). Conclusion: The cumulative incidence of type1 diabetes mellitus disease was increased due to presence of coinfections and decreased with pharmacological diabetes treatment. The linear mixed effects model fitted was appropriate for the estimation of sugar levels based on the risk factor variables for type1 diabetes mellitus patient children. ABBREVIATIONS:RBS = Random blood sugar; FH = Family history; UM = under malnutrition; OM = Over malnutrition; RFT= Renal function test; NS=Nutritional status