Application of Bayesian logistic regression for determining the risk factors of diabetic retinopathy

Diabetes is one of the most common chronic diseases of this century. Retinopathy and makulopati are two most important implications of diabetes. In this study, Bayesian logistic regression is used to assess the factors affected on diabetic- retinopathy. Study population of this cross-sectional study contains all diabetic patients in Tehran of which 623 of them were selected using multi-stage cluster sampling. Age, BMI, hemoglobin, cholesterol, hypertension, duration of diabetes, etc were measured and the status of diabetes were assessed. Bayesian logistic regression was applied using SAS 9.2 software. Of 623 diabetes patients, 54.4% (339 patients) were female and 45.6% (284 patients) were male. In 38% (n=236) of patients diabetic retinopathy were occurred and mean (±SD) age of females and males were 59.5±11.05 and 60.5±11.65 years, respectively. Using Bayesian logistic regression, statistically significant associations were revealed between diabetic retinopathy and age, sex, type of insulin, duration of diabetes and macular edema. Estimates from Bayesian and classical logistic regression were almost similar in magnitude and direction, but, Bayesian model were provided shorter confidence intervals.