In the Field

In April 2024, the SMART Initiative and Action Against Hunger piloted the U.S. Centers for Disease Control and Prevention’s GPSSample application. The purpose of this pilot was to examine the feasibility and acceptability of GPSSample to facilitate statistical sampling and household selection during a SMART[1] survey. The pilot was completed in the Northern sub-counties of Meru County, Kenya during an Integrated SMART survey that assessed the prevalence of malnutrition among children (6-59 months) and women of reproductive age (15-49 years).

The SMART survey featured a cross-sectional design with a two-stage sampling approach based on probability proportional to population size. The sampling procedure featured the lowest administrative level as the primary sampling unit, while households represented the basic sampling unit. GPSSample was used to generate the sampling frame and randomly select households for inclusion in the survey. Fifty-five clusters were mapped, and 3779 households were listed using GPSSample. Using the criteria outlined in the study configuration, n=3618/3779 households were eligible for selection. From the 660 households selected for inclusion, n=596 households consented and n=286 children were assessed for malnutrition.

_______

“First, it’s a game changer, it has enabled the reduction of selection bias because now it’s an app doing it. As colleagues have said, ‘Why aren’t we choosing the households with children?’, if left to humans to choose, we would have conveniently chosen the households that have children…” 

– Survey Supervisor

_______

Overall, focus group discussions revealed that survey staff enjoyed using GPSSample and felt that it had great potential for use in future SMART surveys.  The process of mapping and listing all households within a cluster, rather than relying exclusively on a village guide to provide a paper list of households, helped to improve the quality of the collected data and reduce sampling bias. Improvements in data quality are further facilitated by not only removing the responsibility of sampling from the field survey teams but also the proximity warnings that ensure surveys are not launched from households that were not selected. Survey staff felt the application would be well integrated into contexts where community-based partners already keep track of, and map, public health activities. Leveraging existing systems that allow cluster mapping and household listing to be completed before the survey will be key to GPSSample’s use and success in future SMART surveys.

For questions, please contact Kaitlyn Samson, Research Project Officer at Action Against Hunger Canada: ksamson@actionagainsthunger.ca

Coming soon