Complexity and evidence in health sector decision-making: lessons from tuberculosis infection prevention in South Africa.

To better understand and plan health systems featuring multiple levels and complex causal elements, there have been increasing attempts to incorporate tools arising from complexity science to inform decisions. The utilization of new planning approaches can have important implications for the types of evidence that inform health policymaking and the mechanisms through which they do so. This paper presents an empirical analysis of the application of one such tool-system dynamics modelling (SDM)-within a tuberculosis control programme in South Africa in order to explore how SDM was utilized, and to reflect on the implications for evidence-informed health policymaking. We observed group model building workshops that served to develop the SDM process and undertook 19 qualitative interviews with policymakers and practitioners who partook in these workshops. We analysed the relationship between the SDM process and the use of evidence for policymaking through four conceptual perspectives: (1) a rationalist knowledge-translation view that considers how previously-generated research can be taken up into policy; (2) a programmatic approach that considers existing goals and tasks of decision-makers, and how evidence might address them; (3) a social constructivist lens exploring how the process of using an evidentiary planning tool like SDM can shape the understanding of problems and their solutions; and (4) a normative perspective that recognizes that stakeholders may have different priorities, and thus considers which groups are included and represented in the process. Each perspective can provide useful insights into the SDM process and the political nature of evidence use. In particular, SDM can provide technical information to solve problems, potentially leave out other concerns and influence how problems are conceptualized by formalizing the boundaries of the policy problem and delineating particular solution sets. Undertaking the process further involves choices on stakeholder inclusion affecting whose interests may be served as evidence to inform decisions.