Multilevel analysis for tuberculosis mortality during treatment

Tuberculosis is an infectious disease with social determinants and is associated with individual and regional factors. In this study, multilevel logistic regression analysis was used to identify regional factors affecting mortality during treatment considering the individual level and regional level factors together in the early period (2012-2015) when the Korean tuberculosis PPM cooperation system was introduced. When PPM general hospitals, public health centers, and clinics were selected as treatment institutions, there was no difference in mortality according to regional PPM support rates. Non-PPM hospitals (slope 0.016, p<0.001) and non-PPM general hospitals (slope 0.017, p<0.001) showed a higher mortality rate in regions with higher PPM support rates. When PPM hospitals were selected as treatment institutions, the higher the PPM application rate, the higher the mortality rate (slope 0.033, p<0.001). Due to the institutional situation of adopting a treatment system centered on a small number of large hospitals, each region has a diverse distribution of regional PPM coverage depending on the circumstances, and accordingly, the type of healthcare providers selected creates a different context and leads to a difference in mortality during treatment.