Adequate surveillance of hard-to-reach and ‘hidden’ subpopulations is crucial to containing the HIV epidemic in low prevalence settings and in slowing the rate of transmission in high prevalence settings. For a variety of reasons, however, conventional facility and survey-based surveillance data collection strategies are ineffective for a number of key subpopulations, particularly those whose behaviors are illegal or illicit. This paper critically reviews alternative sampling strategies for undertaking behavioral or biological surveillance surveys of such groups. Non-probability sampling approaches such as facility-based sentinel surveillance and snowball sampling are the simplest to carry out, but are subject to a high risk of sampling/selection bias. Most of the probability sampling methods considered are limited in that they are adequate only under certain circumstances and for some groups. One relatively new method, respondent-driven sampling, an adaptation of chain-referral sampling, appears to be the most promising for general applications. However, as its applicability to HIV surveillance in resource-poor settings has yet to be established, further field trials are needed before a firm conclusion can be reached.
From the aFamily Health International, Arlington, VA, USA
bCenters for Disease Control and Prevention, Atlanta, GA, USA
cCornell University, Ithaca, NY, USA.
Correspondence to Keith Sabin, MS E-30, 1600 Clifton Road NE, Global AIDS Program, Centers for Disease Control and Prevention, Atlanta, GA 30333, USA. Tel: +1 404 639 6314; E-mail: firstname.lastname@example.org