Reassessing geographic bottlenecks in a respondent-driven sampling based multicity study in Brazil

Area of Science:

• Epidemiology
• Social Geography
• Public Health Research

Background:

• Respondent-driven sampling (RDS) is a valuable network-based method for reaching hidden populations, such as drug users.
• Geographic factors and social structures significantly influence the effectiveness and reach of RDS recruitment chains.
• Understanding spatial dynamics is crucial for overcoming bottlenecks in RDS studies.

Purpose of the Study:

• To analyze the spatial dynamics of recruitment chains in a respondent-driven sampling study among drug users in Recife, Brazil.
• To identify geographic bottlenecks and social geographical influences affecting RDS recruitment.
• To provide recommendations for optimizing RDS study designs in complex urban environments.

Main Methods:

• Temporo-spatial analysis was employed to visualize recruitment dynamics over time and space.
• Sequential mapping illustrated recruitment patterns based on participants' neighborhoods of residence and drug use locations.
• Poisson regression modeled recruitment rates in relation to neighborhood demographics and proximity to assessment centers.

Main Results:

• Recruitment success was negatively associated with the distance between participants' locations (residence/drug use) and the assessment center.
• Neighborhoods with a higher proportion of residents in informal settings showed a higher recruitment rate.
• The spatial distribution of drug users and social geography significantly impacted recruitment chain formation.

Conclusions:

• Recruitment chains in RDS studies are intrinsically linked to the social geography and demographic characteristics of the target population.
• To enhance RDS effectiveness, studies should strategically select initial participants (seeds) from diverse neighborhoods.
• The implementation of multiple assessment centers can mitigate spatial biases and improve overall recruitment coverage.