PROJECT SUMMARY Tuberculosis (TB), caused by Mycobacterium tuberculosis, is the leading cause of mortality from an infectious disease worldwide, with over 1.3 million TB deaths in 2017. Of increasing concern is the spread of extensively drug-resistant (XDR) TB, which is now present in over 100 countries. XDR TB is associated with high mortality rates and has been deemed a public health crisis by the World Health Organization (WHO). Transmission is now recognized as the major driver of this global epidemic. Thus, a better understanding of where XDR TB transmission occurs is urgently required. While transmission was previously thought to occur predominantly among close contacts, recent evidence suggests that most transmission is attributable to casual contact (brief interactions in the community), as would occur in shared spaces and congregate settings such as schools, public transport, and markets. Currently, there are large gaps in our understanding of casual contact transmission, including where it most commonly occurs. TB transmission risk (the probability of becoming infected) in an indoor space can be estimated using carbon dioxide (CO2) measurements as a proxy for rebreathed air. Incidence, or the number of new cases in a population, can also highlight areas of high transmission—although TB incidence demonstrates spatial dependency, wherein transmission is dependent on physical proximity to an infected individual. Current methods used by WHO to calculate TB incidence do not take into account spatial dependency and, thus, may not provide accurate estimates of local transmission. Our overarching hypotheses are: 1) casual contact transmission occurs mainly in congregate settings, and 2) improved estimates of local incidence can be calculated using Bayesian methods that account for spatial dependence. For Aim 1, we will measure CO2 concentrations in 15 congregate settings and calculate ventilation rates. Locations will be chosen from our group’s previous XDR TB transmission study in Durban, South Africa and location-specific risk of transmission will be modeled as a function of ventilation rates, number of individuals present, duration in location, and a range of infectious doses. In Aim 2, we will use Bayesian statistical methods to estimate local incidence in Durban. Data for this modeling analysis will come from the South African census and our current NIH-R01 cohort study including diagnosed XDR TB cases and geospatial information (e.g., home neighborhood and diagnosing clinic). We will estimate local incidence using autocorrelated regression models that account for spatial dependence. Quantifying the location-specific risk of transmission will allow for targeted public health interventions to improve ventilation and reduce transmission at these sites. Improved estimates of local incidence can also highlight neighborhoods requiring additional resources. With this research proposal and training plan, the applicant will gain a multidisciplinar...