Regression Analysis for Analyzing Life Style Risk Factor of Diabetes in Pakistan

Diabetes is a worldwide metabolic disease. In Pakistan prevalence of diabetes is increasing day by day. This research aims to analyze the risk factors associated with the patient in the Malakand division, KPK Pakistan. The data is collected from four districts of Malakand division District Headquarter Hospital for the period year 2018. The insignificant risk factors are eliminated using the backward Elimination method for the Binary logistic Regression model and for a best-fitted model, the AIC technique is used, while the logistic Coefficient is tested with help of Wald statistic and Hosmer-Lemeshow is performed for the Goodness of fit test. The positive and negative association among risk factors with diabetes is checked with the help of a Chi-square and odds ratio. Based on P-value at 5% level of significance the risk factors Age, cholesterol level, Hypertension, Family History, and Obesity are sensitive risk factors to develop diabetes. The AIC also show same best-fitted model while according to Hosmer- Lemeshow 0.844 indicating a better fit and these risk factors are associated with diabetes for the combine data of Malakand division. In each districts, the significant risk factors that affect to develop diabetes are Age, Cholesterol level and Obesity while according to AIC the best-fitted model is that in which the risk factors Gender and Occupation Status are involved the risk factor obesity show low level of precision based on 95% Confidence Interval and Chi-square statistic shows these factors are associated with diabetes.