Mapping system dynamics shaping neonatal infection prevention and antimicrobial stewardship in Botswana

Abstract Background Healthcare-associated infections and antimicrobial resistance threaten hospitalised neonates in low-resource settings. Infection prevention and control (IPC) and antimicrobial stewardship (AMS) are shaped by interacting system constraints that linear approaches may not capture. Qualitative system dynamics (SD) offers tools to capture feedback mechanisms, but applications frequently rely on group model-building workshops and limited qualitative rigour, undermining credibility. Methods We developed and applied an interview-led qualitative SD workflow in a tertiary neonatal unit in Botswana. Data comprised 67 semi-structured interviews across stakeholder groups spanning the unit, hospital services, and health system governance, complemented by a one-day group model building (GMB) workshop. Interview data were analysed using hybrid inductive-deductive thematic analysis. Themes were translated into SD variables and causal relationships, mapped into ‘seed’ causal loop diagrams (CLDs), then integrated into a comprehensive CLD. Findings and diagrams were validated through expert review and stakeholder feedback. Results The workflow produced ten seed CLDs and an integrated CLD comprising 62 variables and 92 causal relationships. Four dominant reinforcing feedback structures explained persistent IPC/AMS challenges: (1) overcrowding-transmission-length of stay, (2) workload-burnout-turnover, (3) workload-reduced IPC adherence-infections, and (4) IPC support-shared accountability-communication norms. Material shortages and governance gaps amplified harmful feedback loops. Conclusion A qualitative SD approach, grounded in extensive stakeholder interviews and established qualitative analytic practices, can generate credible CLDs whose variables, causal relationships, and feedback structures are traceable to qualitative evidence when repeated workshops are not feasible. The resulting feedback structures highlight leverage points for strengthening neonatal IPC and AMS in comparable resource-constrained hospital settings, while providing a replicable methodological template for future systems-informed implementation research.