Binary Logistic Regression Analysis Using Stepwise Method on Tuberculosis Events

Abstract

Tuberculosis is an infectious disease caused by the bacteria Mycobacterium tuberculosis. Among all the districts/cities of North Sumatra province, Medan has the highest cases of tuberculosis sufferers with a total of 12,105 cases in 2019. This study aims to determine the factors that significantly influence tuberculosis. The factors analyzed were age, gender, occupation, education, BCG immunization, history of diabetes mellitus and HIV infection. This study uses secondary data for the period January 2019 to December 2020 obtained from the Sentosa Baru Health Center. With the help of SPSS, this study uses a stepwise method with forward selection and backward elimination as the method for analysis. Akaike Information Criterion (AIC) is used to select the best model in the stepwise method. With the AIC criteria obtained, the best model is forward selection because the AIC value is lower at 28,527 compared to backward elimination at 41,664. Of the 7 variables studied, there are 3 factors that have a significant effect, namely age, history of diabetes mellitus, and HIV infection so that the model g(x) = 2.802 1.056 X₁ 0.614 X₆ 2.477 X₇.