MODELING RISK FACTORS FOR DIABETES AMONG OUT-PATIENTS OF UNIVERSITY COLLEGE HOSPITAL IBADAN, USING BAYESIAN STRUCTURAL EQUATION MODELS

Abstract

The consistent increase in the national prevalence of diabetes remains a public health concern due to its mortality effect on adults. Many models were built to identify the risk factors of diabetes in Nigeria including the use of traditional regression but there is limited research adopting structural equation modelling to identify the risk factors of diabetes in Nigeria. Hence, this study is aimed at assessing multidimensional relationship between risk factors and diabetes in Ibadan, Nigeria using structural equation model. Methods Secondary data on total of 979 patients were obtained from adult outpatients of the general out-patient directory (GOPD), University College Hospital (UCH), Ibadan using a cross sectional study design. Missing values were examined through the use of ‘mice’ package in R- software and the data contained no notable outliers. Descriptive statistics including mean, standard deviation, frequency and percentages were used to describe the study participants. AMOS statistical package in SPSS version 20 was use to draw the path diagrams and analyse the data, the analysis were performed at 5% significant level. Results Diabetes was prevalent (10.7%) in out-patients and diabetic patients (55.66 ± 14.34 years) were relatively older than non-diabetic patients (49.46±14.76 years). Bayesian SEM model provide an acceptable fit to the dataset with ppp-value=0.47 while the two structural models also confirm the model fit established by the Bayesian SEM model with the following measures of fit indices for model 2; CGF= 3.338, GFI= 0.961, NFI= 0.958, CFI=0.926, RMSEA= 0.029, and model 3; CGF= 3.223, GFI= 0.973, NFI= 0.932, CFI=0.943, RMSEA= 0.031. All the constructs have significant positive influence on diabetes meaning that the chance of participants being diabetic is higher when exposed to a low socio-demographic status, unhealthy lifestyle factors, unmonitored metabolic and anthropometric factors. Socio-demographic construct have an indirect positive influence on diabetes through lifestyle and anthropometric constructs meaning that lifestyle and anthropometric variables can positively mediate the effect of socio-demographic status on diabetes. Conclusion This study identified the consistent rate of increasing prevalence of diabetes in Nigeria as it increases with age with many undiagnosed cases. It was also established that all the constructs are significantly and positively related to diabetes, hence the need for government and health practitioners’ intervention by organizing enlightenment programs for adults on risk factors and prevention of diabetes is suggested.