Putting respondent-driven sampling on the map: insights from Rio de Janeiro, Brazil

Area of Science:

• Epidemiology
• Geographic Information Systems (GIS)
• Social Network Analysis

Background:

• Hard-to-reach populations with high HIV prevalence are critical in restricted epidemics like Brazil's.
• Respondent-driven sampling (RDS) is a key strategy for studying these populations and informing policy.
• Understanding population bridges is vital for controlling HIV epidemic dynamics.

Purpose of the Study:

• To geocode and visualize the spatial distribution of a Respondent-driven sampling (RDS) study.
• To analyze the geographic patterns of 605 heavy drug users in Rio de Janeiro in 2009.

Main Methods:

• Utilized Audio Computer-Assisted Self Interview (ACASI) to collect residence location and characteristics.
• Geocoded interviewee residences and visualized data using network graphs and thematic maps.

Main Results:

• Interviewee distribution was highly heterogeneous across Rio de Janeiro.
• Recruitment chains expanded geographically over 11 waves but slowly.
• Certain key geographic areas remained excluded from the study's scope.

Conclusions:

• The study identified significant network bottlenecks in a complex urban setting with structural violence.
• Small sample size and structural constraints posed challenges for RDS effectiveness.
• These bottlenecks present a formidable challenge for network-based methods in similar urban environments.