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

by George O. Oluoch, Denis Otundo, Seth Nyawacha, Derick Ongeri, Monica Smith, Vivianne Meta, Anna Trelfa, Sayem Ahmed, Robert A. Harrison, David G. Lalloo, Ymkje Stienstra, Frank-Leonel Tianyi

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.

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.

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.