摘要：Objective. We examined the feasibility and value of network analysis to complement routine tuberculosis (TB) contact investigation procedures during an outbreak. Methods. We reviewed hospital, health department, and jail records and interviewed TB patients. Mycobacterium tuberculosis isolates were genotyped. We evaluated contacts of TB patients for latent TB infection (LTBI) and TB, and analyzed routine contact investigation data, including tuberculin skin test (TST) results. Outcomes included number of contacts identified, number of contacts evaluated, and their TST status. We used network analysis visualizations and metrics (reach, degree, betweenness) to characterize the outbreak. Results. secondary TB patients and more than 1200 contacts. Genotyping detected a 21-band pattern of a strain W variant. No HIV-infected patients were diagnosed. Contacts prioritized by network analysis were more likely to have LTBI than nonprioritized contacts (odds ratio=7.8; 95% confidence interval=1.6, 36.6). Network visualizations and metrics highlighted patients central to sustaining the outbreak and helped prioritize contacts for evaluation. Conclusions. A network-informed approach to TB contact investigations provided a novel means to examine large quantities of data and helped focus TB control. The incidence of tuberculosis (TB) in the United States has declined annually since 1992, but the rate of decline is diminishing. 1 The national goal of TB elimination requires state and local TB control programs to increase efficiency with limited resources. 2 TB control in the United States relies on a costly, complex process known as contact investigation to record, locate, and medically evaluate persons recently exposed to contagious pulmonary TB patients. Such contacts are at risk of infection with Mycobacterium tuberculosis and are also more likely to progress to TB disease and continue transmission. 3 , 4 Thus, health department staff must meticulously elicit and locate contacts, screen them for TB symptoms, and administer a tuberculin skin test (TST), which requires a second encounter 48 to 72 hours later to interpret the test result. 5 If the TST results suggest M tuberculosis infection, a chest radiograph and additional clinical evaluation are necessary. Frequently, contacts of patients unlikely to be contagious are sought unnecessarily. 5 Methods to help prioritize TB contacts are needed to avoid fruitless expenditure of resources. A strategy that could also detect early evidence of ongoing M tuberculosis transmission would be especially useful. 6 – 8 TB controllers currently follow a paradigm known as the concentric circle approach to guide their contact investigations. 9 , 10 The duration of exposure to a contagious TB patient, type of relationship (close vs casual), and location of exposure (household, work and school, leisure) are considered when prioritizing contacts. Unfortunately, the current paradigm yields a collection of data from many separate contact investigations without placing the combined results into a broader context of community TB transmission. The outcomes of each contact investigation are often stored (usually on paper) with the TB patient’s records, with no systematic strategy to construct and examine linkages among TB patients, their contacts, and the places where these persons regularly aggregate. The science of network analysis is a mathematical strategy that includes visualization of nodes (people and places) and the connections among them. 11 , 12 For a respiratory infection spread via droplet nuclei, network analysis aims to identify the most critical nodes responsible for transmission and, based upon their location in the network, to predict which nodes are likely to be infected. As subgroups of TB patients and contacts converge, specific collections of nodes can be selected for screening prioritization. Network analysis can add to our understanding of individual-level variables, commonly explored through conventional biostatistical methods that assume independence and often fail to reflect complex links among cases, contacts, and the places they interact. Recent outbreak investigations have provided opportunities to explore various applications of this tool to TB control. 7 , 13 , 14 Our interest in network analysis is in understanding how it may complement, not supplant, health departments’ TB contact investigation practices. We sought to determine whether routine contact investigation data could be extracted from health department records and analyzed by commercially available network analysis software and to test the hypothesis that contacts prioritized with network analysis were more likely to be diagnosed with latent TB infection (LTBI) than nonprioritized contacts.
