Validation of three geolocation strategies for health-facility attendees for research and public health surveillance in a rural setting in western Kenya

STRESMAN, G. H.; STEVENSON, J. C.; OWAGA, C.; MARUBE, E.; ANYANGO, C.; DRAKELEY, C.; BOUSEMA, T.; COX, J.

A simple and operationally feasible way to identify the spatial occurrence of disease in rural areas where homes have no formalized address would be an extremely useful tool and could easily be employed as an operationally attractive approach to spatial disease surveillance in a wide range of settings around the world. A recent study has been conducted in Blantyre, Malawi in an urban setting [25] ; however, our study is, to our knowledge, the first attempt to examine different methods to geolocate health-facility attendees in a rural area and to gauge their precision. Although strategies are not directly comparable due to the different spatial scales, the current study showed that there are options available to obtain spatial information in areas where no formal postal network exists. Results have shown that it was possible to correctly locate close to 80% of participant compounds using either a participatory mapping exercise (to 2 × 2·5 km blocks with buffer) or by using information about the nearest primary school. This is similar to the level of detection of most geocoding strategies when applied in developed countries, although the spatial resolution is not as good [17, 32] . In this study, methods based on name-matching or participatory mapping to the 500 × 500 m cell level proved to be less accurate, but are capable of greater spatial precision.