BACKGROUND
Thailand’s HIV epidemic is heavily concentrated among men who have sex with men (MSM) and surveillance efforts are mostly based on case surveillance and local biobehavioral surveys. We piloted Kai Noi, a web-based respondent-driven sampling (RDS) survey among MSM.
OBJECTIVE
To describe the design, coding, and procedures of a largely automated webRDS system that aimed to replace and fulfill all attributes of a physical and staffed RDS office
METHODS
We developed an application coded in PHP (Hypertext Preprocessor) language that facilitated all procedures and events typically for an RDS survey office, including eCoupon validation, eligibility screening, consent, interview, peer recruitment eCoupon issuance, and compensation. All procedures were automated; coupon ID numbers were randomly generated; participants’ phone numbers were the principal means to detect and prevent duplicate enrolments. Sampling took place across Thailand; residents of Bangkok were also invited to attend 1 of 10 clinics for an HIV-related blood draw and receive additional compensation.
RESULTS
Sampling took place from February to June 2022; seeds (21 at the start, 14 added later) were identified through banner ads, micro-messaging, and in online chat rooms. Sampling reached all six regions and almost all provinces. Fraudulent (duplicate) enrolment using “borrowed” phone numbers were identified and led to the detection and invalidation of 318 survey records. A further 106 participants did not pass an attention filter question (asking recruits to select a specific categorical response) and were excluded from data analysis, leading to a final data set of valid 1,643 participants. Only one record showed signs of straight lining (identical adjacent responses). None of the Bangkok respondents presented for a blood draw.
CONCLUSIONS
We successfully developed an application to implement an online RDS survey among MSM across Thailand. Measures to minimize, detect, and eliminate fraudulent survey enrolments are imperative in online surveys offering compensation. Efforts to improve biomarker uptake are needed to fully tap the potential of online sampling and data collection.