Filling a methodological gap of tuberculosis modelling: accurately capturing the duration of active disease

Tuberculosis (TB) is the world’s leading infectious cause of death. Although effective treatments and vaccines have been available for many decades, progress in TB control has been hampered by the relatively poor epidemiological understanding of TB. The very slow dynamics of the disease make traditional field investigations and trials very difficult to implement, which explains the popularity of mathematical modelling in current TB epidemiology and control.
TB modelling is hampered by extensive uncertainty regarding the most fundamental characteristics of the disease. A particularly important example is the duration of active disease. TB models are often parameterised based on the WHO’s broad indication that disease duration ranges between zero and two years in notified individuals and between one and four years in non-notified individuals. Despite such uncertainty, there has been no attempt to calibrate TB models using patient data on disease duration.
In this project, I will provide a detailed characterisation of the time from disease onset to outcome (cure or death) in individuals with active TB and assess the ability of past and current TB models to capture this duration accurately. The time to reaching the outcome of a disease episode depends on the natural history of TB and on the time to detection when individuals are detected.
An investigation of TB natural history will provide a detailed characterisation of the time to TB-related death or spontaneous recovery. This exercise will be based on published data from the pre-chemotherapy era. Next, data collected during a recent TB randomised controlled trial will be used to profile the time to detection in individuals with TB.
Finally, the collated data on TB natural history and time to detection will be used to assess the ability of compartmental models identified from literature review to replicate the empirically observed dynamics of TB.