Predictors of tuberculosis: Application of a logistic regression model

Abstract Since the patients who suffer from tuberculosis when referring to the physician, they are without any specific symptoms of isolation and treatment and prevention of transmission of the disease to other people. This study aimed to determine the variables affecting tuberculosis using the logistic regression prediction model. This cross-sectional study enrolled 378 people (189 TB patients as a patients group and 189 healthy individuals as a control group) from Ghaem and Imam Reza hospitals, Mashhad (Iran) during March 2011 to December 2014. The variables affecting TB patients such as age, sex, and marital status, AIDS, smoking, history of asthma, organ transplantation, body mass index (BMI), vitamin D3 level, Diabetes, hemoglobin, and malignant diseases were compared in two groups. The sensitivity, specificity, ROC curve (Receiver operating characteristic) and positive and negative predictive values were used to evaluate the predictive power of logistic regression model. Data analyzed using SPSS software version 22 through Logistic regression model and Chi-square test. And P-value  The sensitivity and specificity of this model in predict the tuberculosis were 78% and 68%, respectively. Also, the area under the curve (Roc) was 0.821. Variables; vitamin D3 (p = 0.01), hemoglobin (p = 0.01) and body mass index (BMI) (p = 0.01) significantly associated with tuberculosis. The results showed that the variables of vitamin D3, hemoglobin and body mass index (BMI) have a better prediction of TB in the logistic regression model.