© 2012 the Nordic Societies of Public Health DOI: 10.1177/1403494812465028 Scandinavian Journal of Public Health, 2012; 40: 725–729 Background

Respondent-driven sampling (RDS) was introduced in the 1990s as a new method of recruiting hard-to-reach populations in research, mainly in studies focusing on HIV risk behaviour. This method of sampling originates from a smaller number of non-randomly selected study persons, usually called ‘seeds’, who are interviewed and included in the study. Thereafter, these individuals recruit a deter-mined number of other subjects known to them, often three, and in this way, the study sample grows with each wave of recruitment [1–3]. The newer RDS technique adds important features to older chain-referral methods; clients obtain incentives both for their own participation and for successful recruitment of peers, and their peers are identified only when they volunteer for the study, instead of being identified by the index client. Recruitment is to continue until equilibrium is reached for central variables, i.e. when the distribution of relevant characteristics becomes independent of how these characteristics are distributed among the original seeds [1]. Since the introduction of this technique, it has been applied in a large number of studies, assess-ing several potentially hard-to-reach populations, including injecting drug users, sex workers and men who have sex with men [4]. Although many of these studies have been assessing injecting drug users [4], few have started the sampling process specifically Respondent-driven sampling in a syringe exchange settin