other) have thus shied away from collecting blood specimens or at best have collected unlinked, anonymous specimens, thus severely constraining the analysis and interpretation of results. In reality, the desirability of voluntary HIV counseling may be dictated by both the value of enrolling truly representative population samples in some intervention trials (and not only those persons who agree to receive their results) and host country regulations and standards. Indeed, large population-based HIV studies in countries such as Uganda and Tanzania have adopted voluntary testing strategies, with the proviso that HIV results be made readily available and that the programs provide information and motivation for subjects to receive their results. In Nairobi, Kenya, women were tested in perinatal HIV transmission studies after giving voluntary informed consent and were given an appointment one week later to collect their results. Among the 243 women who were told that they were infected, three-quarters did not report their HIV-positive status to their partner, 1 committed suicide, 7 were beaten, and 11 were replaced by another wife or expelled from their home. The investigators subsequently adopted a policy respecting women's right not to know their HIV test results (Temmerman et al., 1995).
Finally, although it is ethically important to provide an appropriate level of HIV education and prevention services to all study participants, in a randomized trial it is also important to ensure sufficient difference between treatment and control groups to allow interpretation of data. Studies that "overtreat" control-group subjects and thus do not meet this basic criterion are themselves of questionable ethical standing, as they may lead to false conclusions and subsequent ineffective programs or the dismissal of a useful approach to HIV prevention. Services provided to control-group subjects must at a minimum meet local standards of care; it is not necessarily appropriate for them to meet U.S. standards of care. To borrow an example from medical interventions, WHO recently concluded that comparing simplified intrapartum Zidovudine (AZT) regimens with untreated controls can be ethically appropriate in settings where AZT is not otherwise available, even if such a control group would no longer be appropriate in the United States (World Health Organization, forthcoming).
KEY RECOMMENDATION 1. Basic surveillance systems for monitoring the prevalence and incidence of STDs and HIV must be strengthened and expanded.
Good social science research is as dependent as public health and medical research on reliable and valid HIV/AIDS surveillance data. With the implementation of various interventions aimed at controlling HIV transmission, periodic