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.

Given adequate address information, automated geocoding software packages can generate accurate spatial coordinate data for a large proportion of individuals [16, 17] , thereby providing a basis for the spatial analysis of disease transmission [18] [19] [20] . In circumstances where formal address data are unavailable or privacy concerns limit the use of precise spatial locations, other approaches have been used to obtain geographical information on incident cases. Catchment areas of, for example, community pharmacies or general practitioners have been used for describing spatial patterns in disease occurrence [6, 15, [20] [21] [22] . In areas with well-developed public health infrastructure, catchment areas tend to be well defined and sufficiently small to allow a meaningful attribution of localities to clinical cases based on the facility they attended [20, 22] . Geolocation approaches are likely to have less utility for resource-poor settings where formal address systems are commonly unavailable and where health-facility catchment areas are relatively large and poorly defined [5, 23, 24] . Alternative approaches to geolocation strategies are needed in such settings.