Abstract B52: Using respondent-driven sampling to study healthcare and screening behaviors among undocumented immigrant women in the U.S

Abstract

Abstract Introduction: Despite reductions in cervical cancer (CC) incidence in the US due to screening, CC continues to disproportionately affect certain populations, including immigrant women. Immigrant women are significantly less likely than US-born women to receive Pap test screening for CC. Barriers to screening are heightened among undocumented immigrants (without valid US visas or residency papers), who face additional obstacles that limit access and use of healthcare services. While undocumented status is a known barrier to CC screening, additional research is needed to explore causal factors for CC screening disparities among undocumented and legal immigrants. Methodological barriers to research among undocumented immigrants arise from the lack of a sampling frame from which to draw probability-based samples and from individuals’ reluctance to participate in research due to privacy concerns. Respondent driven sampling (RDS) is a new probability-based method that overcomes these barriers by accessing members of such “hidden populations” through their social networks. Despite its success in multiple populations, RDS has had minimum application among undocumented immigrants in the US. The purpose of this pilot study was to determine the effectiveness of using RDS to study healthcare behaviors among undocumented Central American immigrant women in Houston, Texas. Central Americans are the fastest-growing sector of the US Hispanic immigrant population and have high rates of CC. Methods: Recruitment was initiated by 3 non-randomly selected “seeds.” Seeds and eligible participants were given 3 serially-numbered coupons to recruit peers. Women were eligible to participate if they were from Guatemala, Honduras, or El Salvador, ages 18 to 50 years, and currently living in Houston without a valid US visa or residency papers. Participants received monetary compensation for completing the interview and for recruiting peers. RDS relies on key data to generate population-based estimates from the sample: 1) the size of each participant's social network (SN); and 2) recruitment patterns (who recruited whom). Recruitment patterns and homophily were used as an indicator of social clustering by country of origin and number of years of residency in the US (≤5 years vs. >5 years). Homophily (H) is a measure of the likelihood that recruiters recruit individuals like themselves; scores range from −1 to 1, where 1 indicates 100% homophily and 0 indicates random recruitment. Attainment of a stable equilibrium sample composition in regard to demographic and healthcare characteristics was used as an indicator of RDS's ability to generate population-based estimates. Analyses were conducted using RDSAT 6.0. Results: Beginning with 3 initial participants, we recruited a sample of 226 immigrant women over 16 weeks. Participants adopted the recruitment system with reasonable ease (46% recruited ≥ 1 peer) and SNs were dense (mean SN size=20). Homophily was moderate by country of origin (Guatemalans: H=0.52; El Salvadorans: H=0.42) and low by number of years of residency in the US (H≤0.25). Equilibrium was attained for all demographic and healthcare characteristics. Conclusions: This study is the first to evaluate RDS in a US immigrant population. SNs in this population were dense, allowing recruitment to be sustained. While recruitment was moderately influenced by country of origin, women did not affiliate exclusively with those like themselves. This sociometric diversity allowed the sample to attain an equilibrium composition independent of initial participants. Overall, RDS was easy to implement, attained a large sample in a relatively short period of time, and reached an otherwise hidden population. RDS is an effective method for recruiting undocumented Central American immigrant women for research on healthcare and CC screening behaviors. Citation Information: Cancer Epidemiol Biomarkers Prev 2011;20(10 Suppl):B52.