Construction of a predictive model for the incidence of pulmonary tuberculosis in Nantong based on multivariate regression / 公共卫生与预防医学

ABSTRACT

Objective To establish a prediction model for tuberculosis incidence in Nantong area by multivariate regression analysis, and to provide theoretical support for the implementation of combined prevention work in this area. Methods A total of 37 338 registered patients with pulmonary tuberculosis in Nantong City from 2010 to 2021 were enrolled in the observation group. A total of 28,721 healthy people who underwent physical examination during the same period were selected as the control group. Results From 2010 to 2021, there were a total of 37 338 cases of pulmonary tuberculosis in central Nantong. From 2010 to 2015, more than 3,000 cases were reported annually, with the largest number (4 142 cases) in 2011, accounting for 11.09% of the total. The number of cases reported from 2016 to 2021 was all less than 3 000, and the number of cases reported from 2021 was the least , 1 803 cases, accounting for 4.83% of the total. The number of cases decreased each year in the past 12 years. The incidence of pulmonary tuberculosis in males was 70.97% (26 497 cases) and that in females was 29.03% (10 841 cases). In terms of age, the lowest incidence rate was 0.06% (23 cases) in the age group of 0-9 years old, and the highest incidence rate was 19.56% (7 304 cases) in the age group of 60-69 years old. Logistics regression analysis showed that male, age ≥60 years old, occupation as a farmer and smoking history were the risk factors for pulmonary tuberculosis (P < 0.05). ROC curve results showed that the AUC value of the risk prediction model for pulmonary tuberculosis in the Nantong area was 0.872, with a predictive sensitivity of 86.32% and a specificity of 89.21%. Conclusion There are many risk factors for pulmonary tuberculosis in Nantong area, and different factors interact and influence each other. The construction of a risk prediction model for pulmonary tuberculosis can better predict the clinical incidence, which is helpful to guide clinical diagnosis and treatment.