Two-stage cluster sampling to assess SARS-CoV-2 seroprevalence without pre-enumeration: An example from Madagascar

Area of Science:

• Epidemiology
• Public Health
• Geographic Information Systems (GIS)

Background:

• Population-based surveys in resource-limited areas face challenges due to unavailable detailed population data.
• Pre-enumeration is often a logistical hurdle, increasing costs and time.

Purpose of the Study:

• To develop and implement a geographic cluster sampling method for surveys without pre-enumeration.
• To assess the feasibility and potential biases of this novel mapping and data collection system.
• To conduct a SARS-CoV-2 seroprevalence survey in Madagascar using this approach.

Main Methods:

• A cross-sectional observational study in urban Fianarantsoa, Madagascar (February-June 2021).
• Probability proportional to size sampling of clusters (fokontany) with randomly generated GPS coordinates.
• Mobile tablets with OpenStreetMap for real-time navigation, household identification, and data collection, functional offline.
• Integrated field mapping system for virtual household mapping during survey implementation.

Main Results:

• Households were successfully identified at 95.3% of GPS coordinates.
• High participation rates were achieved: 96.8% of contacted households (674/696).
• The sample demographics matched census data, indicating representativeness; a modest inverse correlation between participation and wealth was observed.

Conclusions:

• The integrated field mapping system provides a practical, cost-effective solution for two-stage cluster sampling without pre-enumeration in resource-constrained settings.
• This methodology simplifies logistics, allows for bias evaluation, and creates a valuable geo-referenced database.
• The approach is applicable to various public health surveillance activities beyond COVID-19 research.