Developing a representative community health survey sampling frame using open-source remote satellite imagery in...

Area of Science:

• Geospatial analysis
• Public health
• Demography

Background:

• Sub-national population data is vital for effective health interventions in low- and middle-income countries.
• Lack of up-to-date census data hinders accurate planning and evaluation.
• Remote satellite imagery offers a novel solution for developing population distribution estimates.

Purpose of the Study:

• To develop a representative community survey sampling frame using remote satellite imagery in Sofala Province, Mozambique.
• To evaluate the impact of a 7-year health system intervention.
• To address the absence of recent census data for sub-national population distribution.

Main Methods:

• Digitized buildings from satellite imagery in Sofala and Manica provinces, depositing data in OpenStreetMap.
• Created a probability proportional to size sampling frame by gridding provinces and counting buildings per grid square.
• Utilized a random walk method within selected grid squares to identify households for surveys.

Main Results:

• Successfully created a provincial-level sampling frame using digitized building data from satellite imagery.
• Collected data from 1549 households in Sofala and 1538 households in Manica.
• Despite civil conflict, the methodology provided a robust sample for health intervention evaluation.

Conclusions:

• Remote satellite imagery is a cost-effective method for developing representative sampling frames in data-scarce regions.
• This approach can support health systems planning and intervention evaluation.
• Other research teams can adapt these methods for population distribution tracking.