OBJECTIVES : We hypothesize that multiple re-exposures to tuberculosis (TB) within a short period after initial infection confers greater likelihood of developing active infection within one year of initial infection, a phenomenon we refer to as complex contagion. The purpose of this research is to study the effect of treatment targeted to highly re-exposed infected individuals in a TB outbreak with complex contagion.

METHODS : We built and validated an agent-based simulation model (ABM) of a TB epidemic, parameterized with epidemiological data from a series of TB outbreaks in the 20^th century in western Canada. We developed a novel calibration procedure to fit two versions of the model: one without complex contagion (base model) and one with complex contagion, where the risk of developing active TB follows a sigmoidal function with a small number of re-exposures conferring low risk of activation and a high number of re-exposures conferring high risk of activation. We tested two treatment strategies: mass treatment for actively infected agents (untargeted strategy) and prophylactic treatment for highly re-exposed infected agents in addition to mass treatment for active agents (untargeted strategy).

RESULTS : We find that given the same treatment regimen, the incidence rate under the targeted strategy is 6% lower than that under the untargeted strategy in the year following treatment introduction. TB-specific mortality under the targeted strategy is 5% lower than under the untargeted strategy. Under the targeted strategy, 5% fewer active infections need to be treated to achieve the same incidence and mortality rate as the untargeted strategy. Including prophylactic treatment for highly re-exposed agents, 2% fewer agents are treated overall in the targeted compared to the untargeted strategy.

CONCLUSIONS : Targeted treatment resulted in greater decreases in incidence and mortality, with fewer agents receiving treatment, compared to the untargeted strategy. Our findings suggest that targeting highly re-exposed individuals may decrease treatment costs.