Search

Priority setting for infectious disease control has evolved beyond simple descriptions of costs and consequences of single interventions. Applications of economic evaluation alongside disease transmission modelling now include user-friendly models, which account for setting-specific variations in input prices and epidemiological characteristics, as well as optimisation routines. These developments allow the straightforward assessment of the local cost-effectiveness of new health technologies and rankings of multiple intervention options. At the same time, priority setting increasingly recognises that policymakers may be fulfilling multiple objectives alongside efficient resource allocation, such as pursuing equity in health outcomes, access to health care and financial protection, and that they are faced with a range of health system constraints in any given settings. These constraints may encompass physical input shortages on the supply side (e.g. lack of skilled human resources or disruptions in procuring supplies) and lack of uptake on the demand side (e.g. financial or other barriers such as hesitancy or stigma). Failure to take these into account can result in unfeasible health interventions being recommended and, ultimately, in economic evaluation evidence being disregarded. Different methods for incorporating constraints in priority setting have been put forward, both within the traditional cost-effectiveness analysis framework and alongside it. All these methods present strengths and weaknesses in terms of how they deal with different types of constraints and priority setting contexts, as well as in the extent to which decisions are arrived at algorithmically or through a more deliberative process. My PhD thesis was conceived during two years spent on a project advising the South African National Department of Health on tuberculosis (TB) control policy and implementation. In 2015, the South African Deputy President announced plans for a comprehensive TB screening programme to tackle one of the world's worst TB epidemics driven by HIV. A key question was how to implement such a complex and costly intervention as intensified TB case-finding (ICF) at full scale in an over-stretched health system. The aim of my thesis was therefore to explore and develop the methods for incorporating feasibility concerns, and specifically health systems constraints, in priority setting models both internally and externally to the traditional cost-effectiveness analysis framework, using priority setting around TB prevention and control in South Africa as a case study. The first step was to carry out a systematic review of the literature on the possible ways to restrict disease transmission model outputs to account for health system constraints that affect the achievable coverage and outputs of disease control interventions. I then carried out an incremental micro-costing exercise of the TB control interventions that the South African Department of Health was considering for inclusion in the latest National TB Plan. The costing covered all the resources needed to deliver the intervention at scale, including the costs of the extra resources needed to relax health system constraints, such as hiring additional clinical staff and budgeting for additional diagnostic equipment. These constraints were identified in consultation with experts on the South African TB Think Tank. Intervention costs were then attached to disease transmission model outputs to generate incremental cost-effectiveness ratios under three different scenarios: (1) without considering the constraints to implementation; (2) considering the constrains; and (3) including the costs of 'relaxing' the constraints to achieve unconstrained coverage. This exercise showed that the cost-effectiveness ranking of interventions is substantially affected by considering health system constraints. It also provided valuable information for policymakers on the practical feasibility of the proposed interventions. Lastly, the use of group model building, a qualitative system dynamics modelling technique, was explored to elicit information on the health system constraints that apply to a given setting and set of interventions. This approach was found to be superior to the unstructured expert elicitation usually employed to generate unit cost and quantities assumptions in economic evaluation, as it takes into account the dynamic interactions between the intervention and the health system. The approach was also more likely to identify high level health system constraints that are difficult to incorporate in quantitative analyses. Information on these constraints might be best presented to decisionmakers either alongside, but externally to cost-effectiveness analysis results; or in the form of disease transmission model 'exemplary' scenarios where intervention effects (but not costs) are restricted.