Evaluating health status and risks among Native Hawaiian and Pacific Islander communities in Hawai'i: a respondent-driven sampling approach.

Abstract

Purpose: Respondent-driven sampling (RDS) is a sampling method that relies on social networks to recruit hard-to-reach populations, and reduces the bias from non-random selection. This study aimed to assess the efficacy of RDS in collecting health assessment data from underrepresented populations not captured by traditional sampling techniques.

Methods: An RDS study was conducted in Hawai'i between 2017 and 2018 of Native Hawaiians, Chuukese, and Marshallese participants. 1006 cases consisting of 352 seeds and 654 recruits were analyzed in conjunction with data from the 2018 Behavioral Risk Factor Surveillance System (BRFSS), filtered to include Native Hawaiian/Other Pacific Islander participants (n = 1564). Missing network size data was imputed by RDSAnalyst and determined by the sample median network size. Weighted samples were compared for differences.

Results: Chi-square testing revealed significant differences between the RDS and BRFSS weighted samples across sex, age, education, income, and colon/cervical cancer screening variables. Only BMI group and smoking status exhibited no significant differences. RDS methods recruited participants efficiently within one year.

Conclusion: The findings indicate that RDS offers an effective sampling methodology when trying to reach hidden populations and provides more insight into the social networks of underserved communities as the transfer/utilization of health information may be linked to social connectedness.