This is Not Your Facebook Social Network

BACKGROUND:   In October 2012, the North Dakota Department of Health, Grand Forks Public Health Department and Altru Health System confirmed three cases of active tuberculosis in Grand Forks County. Through contact tracing and testing of close contacts, 12 additional active tuberculosis cases were discovered. A complex social network was identified through contact interviews, including shared social settings such as households, prisons, schools and bars.

METHODS:   A social network map was constructed using NodeXL, a template for Microsoft® Excel® that allows the construction and analysis of network graphs. This program allows all cases to be linked to their named close contacts and shared social settings contacts to provide a visual map of the social network. Cases and their contacts were assigned individual visual properties, such as color, size and shape based on whether they were confirmed cases, contacts of cases, or shared social settings and testing status. Degree centrality, closeness centrality and betweenness centrality were calculated by NodeXL.

RESULTS:   Epidemiological links were found for 21 active TB cases since 2010 and 583 contacts were identified through contact investigations. A social network map was developed to provide a visual description of the outbreak and identified how cases were epidemiologically linked.  Individuals who shared a household or were in prison with a confirmed case had significantly higher betweenness centrality (p=0.0001) than those who did not.

CONCLUSIONS:   A social network map can help identify sources of the outbreak and additional individuals that may have been infected. It shows the complex connections between cases due to a shared social network and helps identify high-risk contacts. NodeXL builds off of Microsoft® Excel®, making it quick to learn and utilize during an outbreak. NodeXL was also an efficient tool for calculating centrality measures, which is important for identifying individuals most influential in disease transmission. Individuals that shared a household or were in prison with a confirmed case were more likely to have a betweenness centrality measure of 100 or greater, which indicates the extent to which individuals act as channels in the transmission of disease. Identifying individuals with high betweenness is important for targeting interventions, because they are likely to become infected early in the epidemic and have the potential to spread to a greater number of individuals in the network.