Structure and agency in the economics of public policy for TB control

Globally, Tuberculosis remains a devastating disease, despite the availability of treatment. The disease is associated with poverty, and those with the disease incur a high cost of accessing care, while simultaneously experiencing income loss due to a loss in productivity. A key challenge in TB programmes remains the accurate diagnosis of the disease, especially in people who are HIV positive. Diagnosing TB can be very resource intensive and the accuracy of diagnosis is dependent on a range of disease, health service organisation and provider behaviour factors. This thesis seeks to enhance understanding of how the behaviour of healthcare workers mediates the value of TB diagnostic algorithms, and how this may affect the costs, outcomes as well as the economic burden associated with the disease in South Africa. The work presented is based on empirical work done alongside a pragmatic cluster randomized control trial. Empirically, it examines the longitudinal economic burden of TB diagnosis and treatment in South Africa. The discrepancies between the time at which patients incur the greatest cost and income loss, and the available social protection are highlighted. Based on empirical work, a purpose-built state-transition mathematical model of TB diagnosis and treatment was developed to estimate the cost-effectiveness, from the perspective of the health service and the patient, of health systems interventions to strengthen TB diagnosis. Recognising healthcare workers as those who ultimately express policies, the behaviour of healthcare workers was included in the cost-effectiveness analysis by 1) using data from a pragmatic trial reflecting routine practice and clinical decision-making at the time of the study; 2) developing a conceptual framework of the relationship between behaviour at decision points and disease outcomes; and 3) investigating how these interactions may influence the value of the diagnostic algorithm. Possible public policy levers to improve TB diagnosis in healthcare facilities, as well as the potential mediators of costs and effects were explored. The thesis concludes with recommendations for further methodological work to expand on the approach explored in this thesis to improve how heterogeneity in estimates of cost-effectiveness is presented to decision-makers