Novel transdisciplinary methodology for cross-sectional analysis of snakebite epidemiology at national scale

Abstract

Background: Worldwide, it is estimated that snakes bite 4.5-5.4 million people annually, 2.7 million of which are envenomed, and 81,000-138,000 die. The World Health Organization reported these estimates and recognized the scarcity of large-scale, community-based, epidemiological data. In this context, we developed the "Snake-Byte" project that aims at (i) quantifying and mapping the impact of snakebite on human and animal health, and on livelihoods, (ii) developing predictive models for medical, ecological and economic indicators, and (iii) analyzing geographic accessibility to healthcare. This paper exclusively describes the methodology we developed to collect large-scale primary data on snakebite in humans and animals in two hyper-endemic countries, Cameroon and Nepal.

Methodology/principal findings: We compared available methods on snakebite epidemiology and on multi-cluster survey development. Then, in line with those findings, we developed an original study methodology based on a multi-cluster random survey, enhanced by geospatial, One Health, and health economics components. Using a minimum hypothesized snakebite national incidence of 100/100,000/year and optimizing design effect, confidence level, and non-response margin, we calculated a sample of 61,000 people per country. This represented 11,700 households in Cameroon and 13,800 in Nepal. The random selection with probability proportional to size generated 250 clusters from all Cameroonian regions and all Nepalese Terai districts. Our household selection methodology combined spatial randomization and selection via high-resolution satellite images. After ethical approval in Switerland (CCER), Nepal (BPKIHS), and Cameroon (CNERSH), and informed written consent, our e-questionnaires included geolocated baseline demographic and socio-economic characteristics, snakebite clinical features and outcomes, healthcare expenditure, animal ownership, animal outcomes, snake identification, and service accessibility.

Conclusions/significance: This novel transdisciplinary survey methodology was subsequently used to collect countrywide snakebite envenoming data in Nepal and Cameroon. District-level incidence data should help health authorities to channel antivenom and healthcare allocation. This methodology, or parts thereof, could be easily adapted to other countries and to other Neglected Tropical Diseases.

Fig 1. Example of household selection using random points and 10 m-radius buffer area.

The yellow dots exemplify the selected households. The size of the dots has been increased for clarity. Sample image from http://openaerialmap.org with similar viewpoint to the ones used in the methodology (licenced CC BY 4.0).

Fig 2. Percentage of household selection success in clusters, via satellite imagery in Cameroon.

Results of correctly and incorrectly selected household using high-resolution satellite background images in QGIS. The thick lines indicate the median, the diamonds display the mean, the box delimits the interquartile range (IQR) and the whiskers give the minimum and maximum values inside the 1.5 IQR.