Tuberculosis (TB) is a deadly disease caused by the airborne pathogen Mycobacterium tuberculosis. Active TB disease can occur either as a consequence of a recent infection, generally within one year from the infection episode, or upon reactivation of a latent TB infection (LTBI) occurred even several decades before. In low-burden countries, TB control in general populations is based on a combination of passive diagnosis and treatment of cases, and active screening and treatment for close contacts of index cases. The effectiveness of active case finding through contact investigation activities for reducing the incidence of TB has not been assessed yet, partly because of inherent difficulties in the implementation of targeted epidemiological studies. In this work, we propose an individual based model of TB transmission dynamics within a realistic, time-evolving socio-demographic structure that includes control activities implemented according to the guidelines established by the United States Center for Disease Control (CDC). Four alternative scenarios are considered for comparison with this baseline protocol in the hopes of indentifying an optimal one: (a) passive diagnosis alone (no contact investigation); (b) passive case finding rate improved by 10%, contact investigation as in the baseline; c) baseline with improved contact investigation performance indicators (e.g., coverage), set to CDC performance goals for 2015; d) an extension of the baseline covering also contacts of TB cases with negative smear sputum microscopy (who are less infectious than smear-positive). The model is calibrated using data on TB incidence over time and by age from Arkansas, USA, relative to the period 2001-2011. The calibrated model predicts with high accuracy four independent data sets: i) percentage of clustered vs. non-clustered cases, both total and disaggregated by age group); ii) distribution of cluster sizes; iii) prevalence of secondary TB cases and iv) prevalence of LTBIs in contacts of TB cases. Both iii) and iv) are measured in households and workplaces. According to model output, the implemented contact investigation protocol (baseline) avoided about 15.2% [95% confidence interval: 9.2-24.0%] of TB cases and 19.4% [13.5-32.6%] of TB deaths overall with respect to passive diagnosis alone (scenario (a) ) in the time span considered. The most effective scenario was the one extending investigations to contacts of sputum smear-negative cases (d), with about twice the number of avoided cases and deaths. Treatment of LTBI in contacts of TB cases is predicted to avoid only about 3.1% [1.8-4.5%] of all TB reactivation cases within the next 60 years, in the baseline scenario. This percentage increases only slightly even in the most optimistic case (d), with 5.3% [4.6-6.5%] avoided cases. Contact investigation programs have avoided a substantial number of TB cases and deaths, but margins for improvement exist, both through better implementation of current protocols and by its expansion to less infectious TB cases. In addition, this study suggests that LTBI treatment of contacts is not an effective strategy for the prevention of TB cases, and its continuation as a control strategy should be evaluated against potential risks for the emergence of drug resistance in incompletely treated infections.

Tuberculosis; Mycobacterium tuberculosis; Contact investigation; Individual Based Model