MULTIFACTORIAL MATHEMATICAL MODEL IN CHOOSING DIRECTIONS TO REDUCE TUBERCULOSIS MORTALITY IN THE RUSSIAN FEDERATION

Significance. Studying factors influencing a reduction in the number of deaths from tuberculosis contributes to the Russia’s achievement of the political goals of the Strategy for Tuberculosis elimination, and will also make it possible to predict mortality rates from tuberculosis. In contrast to traditional studies based on a logistic regression model, the use of a regression model based on the analysis of absolute statistical indicators allows us to determine statistical indicators associated with the number of deaths from tuberculosis. The purpose of the study is to create a model describing the influence of various processes reflected in statistical forms on the number of deaths from tuberculosis in the Russian Federation. Material and methods. The influence of various indicators that can potentially affect the number of deaths from tuberculosis in the period from 2009 to 2023 was studied by including them in the Poisson regression model, giving preference to models with the lowest values of the Akaike information criterion (AIC). Results. The fitted regression model (AIC=201) includes an intercept (8.623775) and five statistically significant (p<0.05) parameters and their coefficients: the coding method for the deceased patients with HIV/TB co-infection (coefficient 0.137923); the number of patients with fibrous-cavernous pulmonary tuberculosis (0.000024); the number of patients with tuberculosis of the respiratory organs with multiple drug resistance to mycobacterium tuberculosis (0.000013); the number of incurable patients with tuberculosis (0.000023); and the number of persons examined to detect tuberculosis (-0.000005). The obtained discrepancies between the modeling results and the actual results are small compared to the number of deaths and do not exceed 2%, with the exception of the COVID-19 pandemic, when the estimated number of deaths in 2020 was 2.1% higher than the actual number, and in 2021 it was 3.6% lower than the actual number. The model also gives an acceptable result on earlier statistical results from 2005 to 2008. Conclusion and scope of application. The obtained model demonstrates correctness of the Russian approach to active detection of tuberculosis patients. The presented approach can be used to predict the number of deaths from tuberculosis. Keywords: tuberculosis; mortality from tuberculosis; mathematical model