Conducting epidemiological studies on snakebite in nomadic populations: A methodological paper.

Introduction: Research on snakebite has mostly been conducted on settled populations and current risk factors and potential interventions are therefore most suited for these populations. There is limited epidemiological data on mobile and nomadic populations, who may have a higher risk of snakebite. Methods and results: We conducted a scoping review to gather evidence on survey methods used in nomadic populations and compared them with contemporary survey methods used for snakebite research. Only 16 (10.5%) of 154 articles reportedly conducted on pastoralist nomadic populations actually involved mobile pastoralists. All articles describing snakebite surveys (n = 18) used multistage cluster designs on population census sampling frames, which would not be appropriate for nomadic populations. We used geospatial techniques and open-source high-resolution satellite images to create a digital sampling frame of 50,707 households and used a multistage sampling strategy to survey nomadic and semi-nomadic populations in Samburu County, Kenya. From a sample of 900 geo-located households, we correctly identified and collected data from 573 (65.4%) households, of which 409 were in their original locations and 164 had moved within 5km of their original locations. We randomly sampled 302 (34.6%) households to replace completely abandoned and untraceable households. Conclusion: Highly mobile populations require specific considerations in selecting or creating sampling frames and sampling units for epidemiological research. Snakebite risk has a strong spatial component and using census-based sampling frames would be inappropriate in nomadic populations. We propose using open-source satellite imaging and geographic information systems to improve the conduct of epidemiological research in these populations. Author summary: Nomadic populations are at increased risk of snakebites, but interventions will not be developed or improved without reliable data on this risk. Highly mobile populations require specific considerations in selecting or creating sampling frames and sampling units for epidemiological research, and using census-based sampling frames would be inappropriate. We conducted scoping reviews to gather evidence on survey methods used in nomadic populations and compared them with contemporary survey methods used for snakebite research, which we used to inform our methodology. We describe the creation of a sampling frame of 50,707 households, the selection and geo-tagging of a sample of 900 households, and the identification and enrolment of participants from 875 households between August 2019 and March 2020. We propose using open-source satellite imaging and geographic information systems to improve the conduct of epidemiological research in these populations. [ABSTRACT FROM AUTHOR]