ETHICAL CONSIDERATIONS IN SURVEYS EMPLOYING RESPONDENT-DRIVEN SAMPLING

DeJong et al. discussed ethical considerations in surveys employing respondent-driven sampling with high-risk populations, including injection drug users (IDUs), in Lebanon.¹ While we commend the authors' work, several concerns bear comment.

First, almost all concerns raised as ethical problems in implementing respondent-driven sampling apply universally to HIV research with at-risk populations irrespective of the sampling methodology and are not unique to respondent-driven sampling.² Building on the Belmont report,³ procedures for protecting participants, assessing and differentiating between primary and secondary remuneration, and the timing of such remuneration are increasingly refined.^4,5 Respondent-driven sampling does not require reinventing procedures to protect participants but thorough implementation of relevant procedures (e.g., adequate formative research, pilot study, confidential coupon-management system, thorough screening for eligibility, adequate training of recruiters and staff). It is crucial to distinguish implementation concerns from regulatory and ethical concerns, addressing each using pertinent procedures.^5,6 Whenever ethical concerns converge synergistically, such as in HIV research employing respondent-driven sampling or other peer-driven sampling methodology with IDUs in stigmatizing environments, use of appropriate procedures is important.^5,7–9

Second, the article suggests that recruitment using respondent-driven sampling may expose participants to risk by disclosing illegal or stigmatized behavior. Respondent-driven sampling requires social network recruitment, during which participants recruit those whom they already know from the target population. Participants elect whom to recruit and whether they share information with others. The potential adverse effect of unwanted disclosure on IDUs' reputation appears unlikely when IDUs know they are at risk for HIV and because in most settings peer-to-peer IDU recruitment does not convey new information about drug use or HIV risk. Use of respondent-driven sampling with IDUs permits examination of social linkages across HIV status and ethnic groups. Safeguards against community-level harm to reduce further stigmatization are relevant, especially when HIV is concentrated in particular ethnic groups.¹⁰

Third, the article does not discuss adequately community-level benefits of biobehavioral surveys in advancing HIV/AIDS research, prevention, and treatment (e.g, use of data to plan interventions). Respondent-driven sampling provides access to hard-to-reach at-risk populations in high-stigma environments and improves statistical reliability and validity of outcome measures.¹¹ To provide appropriate balance, discussions and publication of ethical concerns need to be based on study data, because describing hypothetical harms could create the impression that harms were actualized, or that study-level protective procedures were inadequate.

We believe in the importance of accurately attributing ethical concerns to specific and correct factors to ensure scientifically valid research and use and publication of ethically sound safeguards in research.

A patient in a portable tent receives emergency medical care at the Wise County Fairgrounds in Virginia. From July 20–22, 2007, nearly 1200 volunteer physicians and students treated over 2500 Appalachians who could not afford health care. Printed with permission of Magnum Photos.