Do social networks attenuate the population-level impact of tuberculosis interventions? A mathematical modeling study

Abstract

Abstract Mathematical modeling is vital for tuberculosis (TB) goal-setting and program planning. Many TB models assume “all-to-all” mixing, i.e., that any infectious individual can transmit TB to any susceptible individual in a population. We compared the impact of TB treatment and vaccination in an all-to-all compartmental model versus a social network model that had identical TB disease assumptions, but with transmission only among social contacts. We found that low-coverage or low-efficacy treatment or vaccination had considerably less impact on TB cases when modeled using a social network. Treatment that shortens TB disease by 20% reduced new TB cases by 71±0.1% after one year with a social network, compared to 82±0.9% with all-to-all mixing. Effective vaccination for 30% of the population reduced new TB cases by 72±1.1% after one year with a social network, compared to 94±1.3% with all-to-all mixing. In contrast, high coverage-coverage and high-efficacy interventions had similar impacts in both models. Results were consistent across modeled population sizes (10,000 – 150,000) and average number of contacts per person in the network (12 – 60). Use of all-to-all transmission models may overestimate the impact of low-coverage and low-efficacy interventions, with implications for TB target-setting and program planning when only sub-optimal interventions are available.