|Home | About | Journals | Submit | Contact Us | Français|
Studies investigating the effects of biological HIV prevention technologies have been reported with promising results for slowing the spread of HIV. Although prevention technologies can reduce the rate of HIV transmission at varying levels of efficaciousness, it is vital to anticipate the impact of HIV prevention technologies on subsequent sexual behaviors. Risk homeostasis theory posits that decreases in perceived risk, which will occur with access to HIV prevention technologies, will correspond with increases in risk taking behavior. Here we review the literature on risk compensation in response to HIV vaccines, topical microbicides, antiretroviral medications, and male circumcision. Behavioral risk compensation is evident in response to prevention technologies that are used in advance of HIV exposure and at minimal personal cost. We conclude that behavioral risk compensation be addressed by implementing adjunct behavioral risk reduction interventions to avoid negating the preventive benefits of biomedical HIV prevention technologies.
An important step in reducing HIV transmission involves understanding how individuals contemplate their own risk, particularly in the context of increasingly available biomedical technologies for reducing HIV transmission. Reductions in HIV transmission risks that result from HIV prevention technologies, such as vaccines, topical microbicides, antiretroviral medications, and male circumcision, have the potential to simultaneously lower perceptions of risk which in turn may alter risk-related behaviors. An inadvertent increase in risk behaviors following the application of a risk reduction technology is termed behavioral disinhibition or risk compensation . It is the evidence for potential increases in risk practices that can follow the introduction of risk reducing technologies that is the subject of this review.
Risk compensation threatens the potential benefits of newly developed and emerging HIV prevention technologies, no single one of which will likely exceed 60% protection. Because consistent use of condoms offers 90% protection against HIV, any significant reductions in condom use as well as increases in behaviors that risk co-occurring sexually transmitted infections (STI), can offset the protective value of a given prevention technology . For example, mathematical modeling suggests that even modest increases in sexual risk behaviors resulting from increased use of antiretrovirals will offset the potential preventive benefits of HIV treatments . Concerns that risk compensation threatens the protective benefits of HIV prevention technologies have been discussed , and there is now an emerging empirical literature on risk compensation in response to HIV prevention technologies.
Here we review empirical and mathematical modeling studies that have investigated risk compensation for HIV transmission. Our literature search was completed in March 2007, using several automated search engines, including PsychINFO, PubMed, and Google Scholar with key terms risk compensation, risk homeostasis, behavioral disinhibition, risk perceptions, HIV technologies, and HIV prevention. Manual searches through selected journals (e.g., AIDS, AIDS and Behavior, JAIDS) for relevant articles were also completed. For the purpose of this review, we define HIV prevention technologies as biomedical and non-behavioral strategies aimed to reduce HIV transmission, namely HIV vaccines, topical microbicides, antiretroviral medications including their use in pre-exposure prophylaxis (PrEP) and post-exposure prophylaxis (PEP), and male circumcision. For the present paper, we define risk compensation as an increase in risk-related behaviors, such as numbers of sex partners, unprotected sex acts, and injection equipment sharing in response to biomedical prevention technologies.
Knowing how individuals contemplate their perceived risk for HIV is a critical step in estimating the population impact of HIV prevention technologies. Wilde  developed risk homeostatsis theory to explain how individuals manage their personal risks. Risk homeostasis is defined as a system in which people accept a certain level of subjectively estimated risk to their health and safety in exchange for the benefits they expect to receive from that activity. Figure 1 shows Wilde’s model adapted for HIV risk compensation. As circumstances in the environment change, such as the availability of risk reducing technologies, estimates of personal risk will change, which in turn will result in risk compensatory behavior change.
In Wilde’s model, the first step in determining the level of risk an individual will accept for an adverse outcome, such as HIV infection, is to define the individual’s target set point. Determining the target set point generally involves an analysis of the costs and benefits of protective and risk behaviors. Once the target set point is selected a comparative process occurs between perceived risks and the target level of risk that an individual is willing to take. A discrepancy between the target set point of the amount of risk that one is willing to take and one’s perceived actual risk causes an adjustment in behavior that is aimed to reduce the discrepancy between the level risk willing to take and perceived actual risk. Maintaining constancy or homeostasis in perceived level of risk that is consistent with the target set point of risk requires self-monitoring and risk processing, resulting in a dynamic and closed-loop process . The application of a risk reducing preventive technologies theoretically reduces constraints on risk behaviors without concomitant increases in perceived risks. Potential risk compensating behaviors, however, likely vary among different HIV prevention technologies.
We examined risk compensation resulting from HIV vaccines, topical microbicides, antiretroviral medications, and male circumcision. However, no studies were found that directly assessed risk compensating behaviors in relation to vaginal or anal microbicides. Table 1 summarizes the currently available evidence for behavioral risk compensation resulting from three sought after HIV prevention technologies: HIV vaccines, antiretroviral therapies, and male circumcision.
A safe and effective HIV vaccine is arguably the single most sought after of all HIV prevention technologies. Since the earliest days of the HIV epidemic, dozens of candidate HIV vaccines have proceeded to phase I clinical trials , and many of these vaccines have resulted in varying rates of success in terms of their ability to produce HIV specific immune responses. Thus far, only two vaccines have progressed to phase III clinical trials, neither of which, unfortunately, produced sufficiently positive results to justify moving forward. An effective HIV vaccine remains elusive because of HIV’s rapid and adaptive mutation rate. New approaches to HIV vaccines are continually being developed and hope remains that a safe and effective HIV vaccine will one day be available.
Studies have assessed the potential behavioral effects of HIV vaccines. Mathematical models have taken into account the possibility for risk compensation in response to an HIV vaccine. With increases in risk compensating behavior, there is a negative cumulative effect of vaccinating individuals. However, the degree to which compensatory risk behaviors offset the effects of a vaccine is a function of not only vaccine efficacy and the degree of risk compensation, but also the virulence of HIV in the population. Co-occurring increases in risk behaviors with partial protection from a vaccine can potentially eliminate the spread of the most attenuated virus, whereas the most virulent strains of HIV will continue to spread .
Qualitative research has shown that individuals often express their own concerns about increases in vaccine induced sexual risk behaviors. Focus group research has reported that commercial sex workers, for example, would have few fears of HIV once they have been vaccinated, diminishing their motivation to have customers use condoms . Individuals perceive HIV as less of a threat in the event that they were vaccinated and that they would likely increase their sexual risk behaviors following vaccination.
Risk compensation studies in the context of HIV vaccine trials are quite limited. In the few studies that have assessed risk compensation in relation to vaccine trial participation, increases in HIV risk behaviors have been observed, providing evidence that risk compensation occurs after receipt of an HIV vaccine (see Table 1). Increases in unprotected anal intercourse over the course of a placebo-control HIV vaccine trial has been reported , unprotected intercourse more than doubled over the course of the trial among participants who were vaccinated. Furthermore, hepatitis B vaccination, which can offer insight into HIV vaccination, among MSM has been associated with increases in unprotected anal intercourse .
Research is currently underway to develop chemical compounds to serve as topical microbicides, that when applied to the vaginal or rectal mucosa protects against HIV transmission . Some microbicides have attempted to provide a barrier that keeps HIV from reaching target cells; whereas other products have attempted to maintain the natural protection of the vaginal environment, and still other agents have attempted to prevent HIV replication once the virus enters cells. To date, there are more than twenty different microbicides in various stages of clinical development.
Thus far, there is limited data on the effects of microbicides on compensatory risk behavior. Some evidence for potential risk compensation comes from studies of men who have sex with men (MSM). MSM who are HIV positive and engage in unprotected anal intercourse show greater acceptance towards using a microbicide than men who do not engage in unprotected anal intercourse . Studies have also found that MSM who are unlikely to use condoms are more likely to use microbicides . There is evidence for risk compensation in response to early use of nonoxynol-9, an effective spermicide that was once thought to offer protection against HIV. Use of nonoxynol-9 among MSM became common and, in some instances, MSM were replacing condoms with nonoxynol-9 . Because microbicides will only be partially protective, along the order of 35% protective ; reductions in condom use in favor of microbicides could result in an increase in HIV transmission. However, use of microbicides by individuals who refuse to use condoms will surely have a protective benefit. The overall population protective benefits of microbicides is therefore unknown.
There are three major avenues under study for using HIV treatments for HIV prevention: treatment to reduce HIV infectiousness, PrEP, and PEP.
Because HAART decreases HIV viral load in blood, it is common to believe that individuals are less infectious when their viral load is undetectable. Evidence for risk reduction resulting from lower viral load comes from several sources, including the reduced risk of mother-to-child transmission when HIV positive pregnant women are treated with antiretrovirals. Also, HIV viral load has been identified as a critical predictor of HIV transmission among heterosexual couples in Uganda, where a 1-log increase in an HIV positive partner’s viral load corresponds to a risk ratio of 2.45 for HIV transmission . Having an undetectable viral load is often associated with increases in unprotected intercourse [17–19]. Moreover, taking protease inhibitors is associated with decreases in unprotected intercourse among MSM .
In a meta-analysis of the HIV treatment and risk behavior literature, Crepaz et al.  examined the relationship between HAART, viral load, and sexual risk behaviors. Individuals who believe that HAART reduces the likelihood of HIV transmission or that HIV transmission is less of a concern because of HAART are more likely to engage in unprotected sex. In addition, individuals are more likely to engage in unprotected sex if they endorse the belief that an undetectable viral load adds protection against transmitting HIV.
Unfortunately, peripheral blood viral load is not a reliable indicator of HIV infectiousness. In fact, the correlation between HIV viral load in blood plasma and semen ranges between .20 and .60 (p<.05 [22, 23]), with some studies finding no association between blood and semen viral loads . Beliefs regarding the protective benefits of having an undetectable viral load based on blood plasma are, therefore, potentially erroneous. In addition, viral load tests are based on detecting copies of free viral RNA in blood plasma and do not quantify cell-associated HIV, which is a major source of HIV in genital secretions . Perhaps most important are non-HIV STIs that increase HIV viral shedding in the genital tract, again causing HIV in semen to spike well beyond what would be expected based on blood plasma viral loads . Finally, different antiretroviral agents have varied degrees of genital compartment penetration, so drug levels are not necessarily the same in all compartments at any given time, especially when adherence to regimens is low. Nevertheless, there is strong evidence that beliefs that HAART offers protection against HIV transmission are prevalent and contribute to risk compensating behaviors.
PrEP is an HIV prevention method in which antiretrovirals are administered protectively to individuals who are known to be HIV negative. The rationale for PrEP is to prevent HIV from replicating following exposure to an HIV transmission event, therefore decreasing the likelihood of HIV establishing permanent infection . The efficacy of PrEP, as an HIV prevention technology, is currently under study. To date, studies with reported results have been limited to non-human animals, which do demonstrate promise [28–29].
Because PrEP as a HIV prevention method is relatively novel, there is limited information on its effect on risk compensation. Nevertheless, there are risk compensation concerns about the use of PrEP, most noteworthy that PrEP could have a similar effect on behavior as HAART . Considering that HAART is not generally prescribed for reducing the likelihood of HIV transmission and PrEP is, increased risk behavior seems particularly likely. At present, the efficacy of PrEP for preventing HIV transmission is unknown and its effects on compensatory behaviors are also unknown.
PEP is prescribed after potential exposure to HIV. Most PEP regimens consist of taking one to three different types of antiretroviral drugs for up to a period of two weeks. While the use of PEP is typically considered a last resort to avert an HIV infection, there is evidence that high risk individuals are more likely to rely on using PEP and that PEP could inadvertently increase an individual’s sexual risk behavior .
However, in a cohort study that followed participants longitudinally and measured behavioral outcomes, there was no overall increase in sexual risk behaviors among participants who were provided an advance supply of PEP . Additionally, as shown in Table 1, individuals who have used PEP have not shown evidence for increases in risk behavior .
The lack of evidence for risk compensation resulting from PEP suggests that, as suggested by Risk Homeostasis Theory, risk compensation is influenced by the perceived costs and benefits of a prevention technology. PEP offers individuals uncertain protection from HIV infection after a transmission event at the expense of considerable side effects and financial cost. These issues are well represented in qualitative research reported by Korner et al. . Participant quotes reported by Korner et al. include the following “Well I would say that it’s not going to make me less safe in the future…I don’t want to have to put myself through that…having to take the pills regimented [sic] and the worry and the stress that it causes. I mean, I would rather not have to do that. [PEP] sort of gives you a taste of what it would be like to be HIV-positive and have to take those drugs continuously…when I went to the pharmacy and I had three big bottles of tablets that I had to get through and the way it makes you feel, it makes you think twice about doing anything risky (pages 881–882). Thus far, the availability of PEP has not been associated with risk compensation.
There has been substantial evidence for the biological plausibility and epidemiological likelihood that male circumcision reduces risks for HIV infection for some time [33, 34]. Researchers have long recognized that circumcision has the potential to offer protective effects from STI including HIV infections. Most likely, the benefits of circumcision result from removal of highly susceptible cells for HIV transmission located in the foreskin as well as cornification of epithelial cells, allowing the skin to develop a protective keratin layer . Initially, multiple studies found a correspondence between HIV prevalence and the number of circumcised men in a given population, particularly in sub-Saharan Africa .
There are now three large randomized clinical trials (RCTs) that were halted when interim analyses showed more than 50% reductions in HIV infection among men who were circumcised [36–38]. Williams et al.  suggested that increased coverage of male circumcision among men in southern Africa could avert as many as 2 million HIV infections and 300,000 deaths over a 10-year period. Kahn et al.  also suggested that the protective value of circumcision will be cost saving. The evidence for potential HIV protective benefits of male circumcision is therefore compelling.
Although research from observational studies has been mixed , there is evidence that risk compensation occurred among men receiving circumcision in two of the randomized trials that tested male circumcision for HIV prevention. For example, Auvert  asked men to report on a series of questions relating to sexual risk behaviors before and following circumcision. Participants who were circumcised reported more sexual contacts through the course of the study with the exception of the healing time period immediately following the circumcision. Bailey  also reported that circumcised men demonstrated significantly greater sexual risk than non-circumcised men over the course of their follow-up. In this trial, consistent condom use did increase by 63% among circumcised men, but this was significantly less of an increase compared to the 95% increase observed in uncircumcised men. Similarly, unprotected sex decreased by 19% among circumcised men, which was significantly less than the 25% decrease observed in men who were not circumcised. Although these differences in risk behavior actually bolster the HIV protective effect observed among circumcised men - circumcised men were less likely to become HIV infected despite their higher risk behavior – the apparent risk compensation raises concerns over the long term should men contract other STIs over time. Of particular concern is how decreased condom use among circumcised men could increase incidence of ulcerative STI and therefore amplify HIV transmission. STIs are associated with a two to five-fold increase in HIV transmission . Herpes simplex virus-2 (HSV-2) is particularly important in HIV transmission and male circumcision does not reduce the risk for HSV-2 . In fact, in one of the circumcision RCTs, 28% of men were HSV-2 positive and HSV-2 was the only baseline predictor of HIV seroconversion . Thus, male circumcision is a major HIV prevention breakthrough, the only biomedical HIV prevention technology with demonstrated effectiveness to date, and its overall impact on HIV prevention could be diminished by risk compensation .
The current paper reviews the extant literature regarding HIV prevention technologies and risk compensation. Based on our review, there is evidence for risk compensation occurring as a result of HIV prevention technologies. Clearly, HIV prevention technologies are greatly needed and offer promise for reducing HIV transmission rates. However, we suggest, based on the literature, that in order to maximize the benefits of these technologies risk compensation must be incorporated as a main focus in risk reduction interventions.
Moreover, the degree to which individuals will compensate their risk behaviors is currently unclear. Likewise, the degree to which different prevention technologies affect behavior is unclear, but would most likely vary as a function of the perceived lowered risk associated with that technology. Thus, considering changes in perceived risk may offer insight into subsequent behavioral changes and further improve upon strides being made in biomedical HIV prevention efforts. Most studies included in this review are representative of initial stages of emerging HIV prevention technologies. Because these technologies are still developing, and published clinical trials investigating microbicides and sexual risk behavior are not yet available, it is difficult to definitively state the impact of HIV prevention technologies on behaviors. Nevertheless, risk compensation needs to be a priority in studies researching the effectiveness of HIV prevention technologies.
To ensure optimal protection offered by HIV prevention technologies, brief risk reduction counseling should be concurrently implemented. Brief risk reduction counseling has been shown to be effective in reducing risk behaviors and subsequent STIs [54–55]. Existing risk reduction counseling models can be modified to address changes in perceived HIV transmission risks in response to prevention technologies, with the goal of curtailing increases in risk behavior post receipt of new HIV prevention technologies. Such counseling may focus on the calibration processes that reconcile risk perceptions and target set points for willingness to take risks. Adjusting behavioral risk reduction to offset risk compensation will therefore increase the preventive benefits of biomedical HIV prevention technologies.
This research was supported by National Institute of Mental Health grants R01-MH71164 and T32- MH074387.