Incorporating seasonal flooding into frameworks modelling geographical access to health services

Seasonal floods are an important barrier to healthcare access in low-income countries, with recognised implications of increased morbidity and mortality. Effective interventions require detailed understanding of how flood inundation alters patterns and magnitudes of inaccessibility through space and time. Despite recognition of the importance of seasonal floods, few have previously modelled their impacts on geographic access. This study developed a new framework for incorporating quantitative measurements of floodwater depth and velocity into network analysis and cost-distance algorithms. The new approach directly calculates flood hazard to vehicular and walking access as it changes through space and time, permitting more realistic representation of floods as a barrier without substantially inflating data requirements or relying on inappropriate assumptions from urban, high-income regions. The new approach was demonstrated in the Barotse Floodplain, a large African floodplain in Zambia. Floodwater variables were obtained from a hydrodynamic model, and all other inputs were sourced from freely-available datasets or manually delineated. Monthly scenarios were produced between October 2017 and October 2018, capturing the spatio-temporal impacts of one annual flood event on access to healthcare and specifically to maternal services. Access was shown to dynamically change through space and time as the seasonal floodwave passed through the floodplain, with substantial drops in access experienced at the peak of the floods. The new approach provided unprecedented, detailed information regarding the onset, duration, and cessation of flood impacts. Previous static approaches were shown to be unable to recreate the dynamism of access under floodwaters. The use of a hydrodynamic model is advantageous over satellite-derived flood extents and has the potential to predict future impacts of climate change on access. This research has demonstrated the importance of modelling spatio-temporal variability in access for regions affected by seasonal flooding and the framework can be applied to other regions to aid in intervention planning and build future health system resilience. Future research should continue to reduce the gap in knowledge of how floods affect access, integrate qualitative and quantitative data to address access through an interdisciplinary approach, and investigate what scale of data are needed for accurate results at different areal units.