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.

There were some limitations to this study. First, it was only feasible to trace 27% of participants to their compounds. Although this provided a large sample, it is possible that if we could have traced all individuals, the results and the conclusions on the applicability of the techniques tested may have been different. However, as the sample was a random selection, the impact on the results is expected to be minimal. Similarly, spatial coordinates were only available for the government-run primary schools in the area, thereby restricting the sample to those residing near these schools. The limited number of school locations that were available as well as the lack of covariates such as size or perception of academic rigour to include as part of delineation of catchment areas likely influenced the size of catchment areas as calculated by both approaches. However, although altered catchment area boundaries would impact both the precision and accuracy of the results, this is not likely to have a significant impact of the results.