Does a Reduction in These Behavioral Risks Lead to a Reduction in HIV?

The evidence is unequivocal that consistent and effective use of condoms and of sterile injecting equipment on the part of injection drug users is nearly 100 percent effective in protecting against HIV. Reduction in risky behavior leads to reduction in HIV to a degree that depends on context, particularly the local prevalence of HIV infection.

It is important to keep HIV seroincidence in mind as the ultimate outcome of interest for HIV prevention efforts. Seroincidence estimates also allow us to compare effectiveness and cost of different programs. Direct measurement of HIV infection is a feasible and desirable outcome variable for some programs. However, practical, ethical, and fiscal barriers often make reliance on measured seroconversion undesirable. In these instances, proxy indices-including other biological markers or modeled estimates of seroincidence based on behavioral outcomes-can be used to estimate the effects of prevention programs on seroincidence.

Study Designs That Lend Themselves to Using Seroconversion as an Outcome

To find reliable differences between intervention and control or comparison samples, one must expect a minimum number of seroconversions in the control sample within the timeframe of the study. These are found in populations where seroconversion rates are high, in large samples, or in studies with long followup. Only a limited number of situations have lent themselves to clinical trials and other studies on this scale.

Many studies using seroincidence as a measure of outcome were conducted in developing countries where HIV incidence is high and policy interventions or community-level programs have been implemented. Among these are studies from Tanzania and Bombay with comparison populations and from Thailand, where an historical comparison was employed. Few studies in the United States have used HIV or any biological measure as an endpoint for the reasons cited above. In the United States and elsewhere, seroconversion has been used to measure the effect of sterile injection equipment availability, bleach cleaning interventions, and methadone treatment with injecting drug users.

Constraints on Using Seroconversion Outcomes

Although seroconversion is a preferred standard for intervention efficacy, there are practical and ethical obstacles to its use. For example, there is a potential selective dropout of research participants who will not agree to repeated HIV testing. Furthermore, research costs can be greatly increased by pre- and post-test counseling and followup or referral for research subjects who are identified as HIV-positive in the study. Counseling and referral are, of course, required by ethical research practice. Nevertheless, where possible and feasible, it is important that behavioral and policy interventions be validated using seroincidence as an outcome.

Transmission Models To Estimate Effects of Behavioral Outcomes on HIV Infection Rates

When HIV seroconversion outcomes are not feasible, well-designed self-report behavioral outcomes have shown indications of being valid and reliable. These behavioral outcomes can be employed in transmission models to estimate the number of averted cases. The models have been developed from studies of HIV-discordant couples and epidemiological studies. Although use of these models requires assumptions about future prevalence and about relationships among variables being studied, a reasonable range of estimates about the probable impact of the intervention on HIV can thus be generated. In theory, estimates of HIV seroconversion during the study may be extended into the future under varying estimates of the maintenance of positive behavioral outcome. The models may also be extended to estimate the potential impact were the program more widely implemented in similar contexts. Finally, potential effects on seroconversion in field settings may be estimated, using these models, from data on behavioral outcomes from studies done in research settings. These models can estimate the impact on seroconversion using reasonable assumptions that the interventions will have less effectiveness in field settings.

Estimates of the effects of behavioral outcomes on HIV seroconversion are still relatively few and mostly retrospective. It should be possible to produce such estimates in advance of prevention trials, contingent on the targeted magnitude of behavioral outcomes and the expected prevalence of HIV infection in the local population. We recommend that such estimates be employed as an additional outcome measure for trials with behavioral endpoints whenever possible. Ongoing work on these models is needed to update and improve the database used to produce and validate them. Furthermore, there is a need to validate, by use of empirical data, the assumption that transmission rates based on naturally occurring behaviors are equivalent to transmission rates based on behavioral changes in response to prevention efforts. These models can also be used to estimate the validity of self-reports.

Other Biological Markers as Surrogates for HIV Seroconversion

Incidence of certain sexually transmitted diseases has been used as a plausible surrogate for HIV seroconversion. The same sexual behaviors are risks for HIV and some sexually transmitted diseases. Sexually transmitted diseases are a powerful potentiator of HIV seroconversion in exposed persons. The higher incidence of sexually transmitted diseases also makes detection of program effects more sensitive. Two ongoing multicenter randomized controlled trials for heterosexual populations have chosen incidence of sexually transmitted diseases as a biologic marker to study the efficacy of HIV prevention interventions, as have international studies such as the study in Tanzania. Unpublished results of a Centers for Disease Control and Prevention project show a decrease in the rate of sexually transmitted diseases to be correlated with a decrease in HIV-related risk behavior. Hepatitis C has been used effectively as a biological marker in studies involving injecting drug user populations, because of overlapping transmission routes. Sexually transmitted disease incidence, hepatitis C incidence, and other infectious disease incidence are reasonable markers for expected HIV exposure.

Back | Next
Table of Contents