Mapping Snakebite Risk in Ghana and Rwanda

Introduction

Background Snakebite was declared a neglected tropical disease (NTD) by the World Health Organization (WHO) in 2017, reflecting its global public health impact. Over 95% of cases occur in tropical settings, predominantly in low- and middle-income countries (LMICs) in South Asia, sub-Saharan Africa and South America and the greatest burden falls upon the most vulnerable populations, including those living in less robust housing, agricultural workers on low incomes, and children. Snakebite envenoming is thought to cause up to 138,000 deaths per year and non-fatal bites lead to considerable health burden through physical disability, such as severe scarring and limb amputation, psychological impact, and social stigma, isolation and economic loss resulting from these consequences. To help address the public health impact of snakebite, WHO launched its strategy Snakebite Envenoming - A Strategy for Prevention and Control in 2019, with an ambitious aim to halve death and disability from the condition by 2030.

Snakebite prevention and management is an evolving field. Significant work is needed to develop effective treatments against envenomation by the full range of medically important snake species at a cost that is affordable for the populations in need. Despite such challenges, approaches to help prevent and manage snakebite do exist and can be effective in reducing the burden of snakebite in high-risk communities. Prevention activities focus on building knowledge within communities on the dangers of snakebite envenoming, education regarding how to avoid snakes within the home and at work, including the use of protective footwear, and promotion of the importance of seeking effective medical treatment, if bitten, rather than relying upon traditional remedies. When a person experiences a snakebite, management of the victim may cover a range of measures including simple first aid techniques such as immobilisation and application of pressure bandages, supportive care in a medical facility, and, if needed and available, treatment with antivenom that is effective against the species of snake responsible.

High quality data on snakebite burden are essential for ensuring that the resources for treatment and prevention are targeted towards the populations most at risk. However, such data are sparsely reported, with data availability especially limited across countries in sub-Saharan Africa (SSA). Most snakebite data from SSA come from health facility surveys and reporting. However, snakebite is not widely designated a notifiable condition and routine health facility data are often either not reported nationally or are done so inconsistently. Where health facility data are available, they are likely to underestimate snakebite burden as they do not capture the many snakebite victims either not reaching health facilities in time or seeking traditional treatment in the community. High-quality cross-sectional community surveys can help in the assessment of snakebite burden at a fine scale, however very few have been conducted in SSA to date, and as they require significant time and resources and may be difficult to undertake in areas with inaccessible terrain, weak healthcare infrastructure or political instability, they do not provide a feasible alternative for estimating burden at the scale needed. Better data are needed to characterise snakebite epidemiology to inform risk mitigation, prevention or elimination strategies and without robust surveillance systems, an alternative approach to effectively identifying high risk areas is needed.

Problem statement and justification for the study

High quality estimates of the geographical variation in snakebite burden in SSA are needed to advocate for resources for snakebite prevention and management, and to align these with the populations most at risk. Recognising the significant deficiencies in routine data and the challenges to relying upon bespoke data-collection methods as outlined above, the WHO's Road Map for Ending Neglected Tropical Diseases 2021-2030 states that better data are needed on snakebite risk distribution in low-resource settings, and to this end 'new approaches and mapping tools are necessary to obtain a granular view of disease epidemiology'.

Spatial analysis and disease mapping techniques are increasingly recognised for their effectiveness in assessing the epidemiology of NTDs in areas in which the assessment and monitoring of disease burden face data availability challenges. Spatial analysis methods assist in the assessment of disease burden by utilising information on spatially varying explanatory variables, such as geographical, climatic, and population-based variables, and assessing their relationship with disease burden data collected at known locations. Such data is harnessed to develop geostatistical models to predict disease risk in unsampled locations.

Snakebite envenoming risk is influenced by the geographic distributions of venomous snake species and human populations and by factors affecting the likelihood and nature of their interaction. As such, snakebite prevention and management programmes could benefit from the application of spatial analysis techniques. There is limited availability of accurate data on snake species distribution at fine scale. Detailed datasets of factors affecting snake habitat suitability, including geographical features such as altitude, climatic variables such as rainfall and temperature, and demographic variables such as population density, poverty and urban-rural distribution have become increasingly sophisticated and publicly available. Many of these datasets also include historical data, allowing for the assessment of temporal trends. The availability of such covariate data and the utility of spatial analysis methods provide a valuable opportunity for harnessing the limited snakebite incidence data to develop models for predicting variation in snakebite risk more widely.

This project aims to understand whether spatial analysis and disease mapping techniques can be used to help understand geographical variation in snakebite risk and aid the prediction of risk in unsampled locations, thereby addressing some of the challenges generated by the poor availability of snakebite risk data in SSA and reducing the reliance upon traditional data collection approaches and the need for repeat surveys across the continent. It specifically aims to address the following research questions:

1\. To what extent can information from routinely available snakebite incidence data and spatially referenced covariates be used to predict geographical variation in snakebite risk?

To enable the development of geostatistical models for predicting snakebite risk, sufficient data on snakebite incidence, and social and natural environmental data of the region of interest are needed. A limited number of existing datasets on snakebite burden, collected through previously conducted community and health facility studies have been identified and will be formally requested for analysis. Alongside this, there is an increasing availability of high-quality open-source datasets on spatially referenced potential explanatory variables, such as environmental (including land cover, altitude), climatic (including rainfall, temperature) and sociodemographic (including poverty and population density) factors, identified through literature review, are available.

Analysing available historic data on snakebite incidence and potential explanatory variables will provide a better understanding of where to target available resources for effective primary data collection in order to validate or refine the geostatistical model.

2 To what extent can the combination of primary and secondary data be used to validate and refine the geostatistical analysis, and what is the relative utility of the different data sources available to the development of this geostatistical model?

Further snakebite data at selected locations will be collected to validate or refine the assessment of the association between snakebite incidence and potential explanatory factors.

Existing data collection platforms will be utilised to collect data on snakebite incidence. At this stage, a collaboration will be planned with the Global Health and Infectious Diseases research group at the Kumasi Centre for Collaborative Research in Tropical Medicine (KCCR), Kumasi, to conduct snakebite surveys alongside ongoing data collection in some of the selected study districts. There is also the potential to collaborate with the Rwanda Biomedical Centre in their upcoming village-level health assessment if feasible within the study period. In addition, stand-alone snakebite community surveys at selected complementary locations will be conducted.

The collection of additional data will enable refinement of the geostatistical model and improve the assessment of the association between snakebite risk and spatially referenced explanatory factors. The aim is to contribute to the development of a predictive risk model for snakebite that can be applied within the study locations. In future work, and with both new or existing data, this would have the potential to be expanded to and applied in further locations across SSA. The outputs of these predictive risk models, in the form of snakebite risk maps, will provide much needed knowledge of geographical variation in snakebite risk so that resources for treatment and prevention can be effectively targeted to populations in need.

Through this work, an assessment of the relative utility of more readily available data sources, such as health facility data and data collected through existing survey platforms, compared to a gold standard (data collected through community-based cross-sectional surveys) will be conducted, so that the value of these data sources in the prediction of snakebite risk can be better understood.

3\. Do changes in environmental variables such as climate, flood and land use (agriculture, roads, dams, building infrastructure) affect the incidence of snakebite?

Environmental factors such as climate, floods, and land use may alter the habitat, distribution, and behaviour of snakes, leading to an increased likelihood of human-snake contacts in some areas. Other factors including humidity, temperature, elevation, rainfall, and normalized difference vegetation index (NDVI) have also been linked to geospatial variation in snakebite incidence, but the evidence is sparse in resource-limited settings, including Ghana and Rwanda.

Using open source, spatially referenced natural and social environmental data, the study will assess the association between time-varying environmental data and snakebite occurrence to understand the temporal variations in snakebite incidence. Understanding the temporal variations in snakebite incidence is crucial as it could help predict seasons of heightened snakebite risk and inform the timing of preventive interventions. The results would therefore practically inform environmental management and risk mitigation measures in high-risk areas and be instrumental in forecasting future trends of the spatio-temporal variation in snakebite incidence across the study locations and other unsampled locations.

4\. How can epidemiological data assist in the appropriate distribution of antivenom?

Within a model-based geostatistics framework, epidemiological data from community surveys can provide reliable estimates of the burden and risk of snakebite. By mapping both areas of increased snakebite incidence and predicting seasonal variation (periods of heightened risk), such epidemiological data could help identify seasonal variation in hotspots where they may be increased demand for antivenom. To answer this research question, available data on health facility locations and antivenom distribution will be used to generate a map of the current strategy for antivenom distribution. Snakebite ris