Subjects and sampling method

This study used a respondent-driven sampling (RDS) method (Heckathorn, 2014) to recruit participants who were current consumers of medical cannabis. RDS is a probability-based sampling method where sampling procedure starts with a convenience sample of well-networked population members, referred to as seeds. After enrolment and completing the interview, seeds receive a fixed number of coupons to recruit members from their social network (Johnston, 2013). The researchers keep track of who recruits whom and their numbers of social contacts. By recruiting long respondent chains, biases related with the initial convenience sample of seeds are detached from the final sample. The RDS method thus produces samples that are independent of the initial subjects from which sampling begins. It also combines the breadth of coverage of network-based methods with the statistical validity of standard probability sampling methods, making it possible to draw statistically valid samples of hard-to-reach population groups (Heckathorn, 1997). Although cannabis use for medical purposes was allowed in Thailand at the time of the survey, medical cannabis clinics had not yet opened in all MoPH hospitals, thus legal access was limited. Most consumers of cannabis were still considered as illegal consumers and a “hidden population”. RDS was therefore justified as the method of choice for recruiting participants.

Consumers of medical cannabis in this study refer to individuals who had been using medicinal cannabis products (including raw plants) to treat or relieve their symptoms or health conditions within the past 12 months of the study. This definition does not imply that the cannabis products were indicated or prescribed by a health professional.

Four parallel recruitment sites were included: Chiang Mai, Khon Kaen, Bangkok, and Songkhla, representing the Northern, Northeastern, Central, and Southern regions of the country, respectively. Identical RDS procedures were used across the four sites. In each site, 3–4 seeds who were well connected to and trusted by the target population were identified through local contacts. In an attempt to recruit representative participants from various socio-demographic groups, the seeds were selected to include both males and females, three age-groups: young or middle adult (18–44 years), late adult (45–64) and elderly (>65), those who received medical cannabis from legal and illegal sources, and those who used it for different conditions (cancer and non-cancer patients). Participants were eligible for the study if they were a current consumer of medical cannabis, aged 18 years or over, and currently lived in one of the four study regions. Exclusion criteria included those who were intoxicated, cognitively or mentally impaired, or too ill to be interviewed; however, no subject was excluded due to any of these reasons. No more than three participants were allowed to be recruited from each recruiter. We aimed to recruit 120–125 participants from each site. This sample size was calculated assuming a design effect of 2 and was sufficiently powered to estimate an assumed medical cannabis use prevalence of 20% with an absolute precision of 10%.