Abstract:

Objective To provide scientific basis for the early detection of tuberculosis (TB) infection in close contacts though investigating the infection risk and model of students (the high-risk and susceptible family members) who were in close contact with elderly TB cases (referred to as indicator cases).Methods A total of 244 newly diagnosed elderly TB patients were registered in the Tuberculosis Prevention and Treatment Center in Wuhan from May 2016 to November 2017. By self-recommendation and according to the inclusion and exclusion criteria, 209 student close contacts were enrolled. A questionnaire survey and a purified protein derivative (PPD) test were conducted. Univariate analysis and multivariate logistic regression models were used to screen the risk factors for TB infection in student close contacts of senile tuberculosis and to study the key risk factors and establish a decision tree model.Results The positive rate of PPD test among the 209 student close contacts was 77.51% (162/209). Logistic stepwise regression analysis showed that the duration of contact with the indicator cases (OR=29.17, 95%CI: 4.65-183.04), site ventilation (OR=5.79, 95%CI: 2.14-15.67), and self-protection awareness (OR=3.56, 95%CI: 1.24-10.22) had statistically significant differences between the students with positive and negative PPD test results. Using the results of logistic regression analysis, the decision tree model was built with 3 layers and 7 nodes. The duration of exposure to the indicator cases had the greatest impact on the risk of TB infection in close contacts. The positive rate of student with a contact time of ≥2 h was 97.30% (72/74), which was higher than that of students with a contact time of <2 h (66.67% (90/135)). The self-protection awareness of close contacts was a secondary influence factor when exposed to indicator cases. Under the contact time <2 h node, the positive rate of close contacts without self-protection awareness was 94.12% (48/51), which was higher than that of contacts with self-protection consciousness (50.00% (42/84)). In the self-protection awareness node, the positive rate of contacts in good ventilation condition at the site of exposure was 25.00% (12/48), lower than that of contacts in moderate or poor ventilation condition at the site of exposure (83.33% (30/36)).Conclusion The decision tree model indicates the duration of contact with the indicator cases, the ventilation at the contact sites and the self-protection awareness as important influencing factors of TB infection among students close contacts. The degree of influence of each factor was as follows: contact time > self-protection awareness > contact point ventilation.

Key words: Tuberculosis, pulmonary, Aged, Contact tracing, Students, Decision trees, Proportional hazards models