Failure to find a behavioral surrogate for STD incidence--what does it really mean?

Abstract

THE PAPER BY Peterman and colleagues,1 provides a clear illustration of, and strong evidence for, a point that has often been made, namely, that there is no simple linear relation between behavior change and sexually transmitted disease (STD) incidence. As we and others have pointed out, the effect of any given behavior change (e.g., increased condom use) on the incidence of an STD (including HIV) depends on many factors including such things as the prevalence of the STD in the population, sexual mixing patterns, degree of infectivity of the donor, characteristics of the host, and the type and frequency of sexual practices.2–5 As Schachter and Chow have argued, “It is naive to assume that a positive change in behaviors introduced in intervention trials necessarily will result in a detectable reduction in STD acquisition rates.”4 Unfortunately, like the earlier study by Zenilman et al,6 the Peterman et al study1 serves primarily to raise questions about (and cast doubt upon) the validity of self-reported behaviors and their utility as outcome measures in prevention trials. It is important to recognize, however, that the findings of Peterman et al, like those of Zenilman et al, probably have nothing to do with respondents' veracity or the validity of self-reports. On both theoretical and epidemiologic grounds, it is virtually a given that changes in behavior will not translate easily into STD acquisition.3 Even if there were some way of directly observing sexual and/or condom use behaviors, given the nature of their measures, Peterman et al would still have found that they “were unable to identify any behaviors that (used alone) could be surrogates for STD incidence in future trials.” The search of Peterman and colleagues for a “behavioral surrogate” for STD incidence confounds two distinct issues. On the one hand, there is the legitimate question of the relationship between behavior change and its impact on HIV and/or STD incidence. On the other, there is the assumption that, at least in randomized trials, STD serves as a valid surrogate for HIV. If this assumption is incorrect (i.e., if STD does not serve as a valid surrogate for HIV), then STD incidence may be neither better nor worse than behavior change as an indicator of successful HIV prevention. Clearly, if the purpose of an intervention is to reduce HIV, the most appropriate outcome measure is HIV incidence. Similarly, if the purpose of an intervention is to reduce some other STD, then the most appropriate outcome measures is STD incidence. But if the purpose of the intervention is to reduce HIV and if it is impractical or not feasible to measure HIV incidence, then, despite the claim of Peterman et al to the contrary, there is no reason to assume that reductions in STD incidence will reveal any more or less about changes in HIV incidence than will increases in protective behaviors. Peterman et al argue that, at least in randomized trials, reductions in STD can serve as a surrogate for HIV. Unfortunately, this simply is not true. The relationship between a given STD and HIV is no less complex than the relationship between a given behavior and HIV. Reducing STD incidence (like increasing condom use) affects the rate of transmission of HIV. But, as indicated above, the effects of decreasing transmissibility on HIV incidence depend on many other factors (e.g., HIV prevalence, sexual mixing patterns). Thus, just as there are conditions under which “true” changes in behavior will not be related to STD or HIV incidence, there are also conditions under which true changes in STD incidence will not be related to HIV incidence.7 As argued elsewhere,2,3 behavioral and biologic outcome measures cannot substitute for, nor validate, one another, and neither serves as a true surrogate for HIV prevalence or incidence. However, both behavioral and biologic measures are important outcomes for studying the efficacy or effectiveness of behavior change interventions-and both provide useful information. The critical question is this: under what circumstances can we expect to find a relationship between these two outcome measures? Unfortunately, by demonstrating the obvious, Peterman et al do not really address this issue; nevertheless, their paper does provide some important insights. First, the authors are to be commended for considering the number of unprotected sex acts rather than the percent of times a condom was used. Second, they do conclude that “people tend to have safe sex with risky partners and risky sex with safe partners.” Although the finding that people are more likely to use condoms with their occasional partners than with their regular partners is not new,8,9 the finding that people with regular partners are more likely to use condoms with these partners if they've been together less than 30 days than if they've been together more than 90 days, does provide additional evidence for the proposition that people behave very differently with safe than with risky partners. Most important, Peterman et al found that, among the participants in Project RESPECT,10 clients with 3 or more sex partners who had some unprotected sex with an occasional partner were significantly more likely to acquire a new STD (cumulative incidence = 20.8%) than those clients who had only a main sex partner with whom they had been for at least three months, who they believed was monogamous (cumulative incidence = 12.1%). Unfortunately, these findings were largely ignored because “although a relative risk of two indicates some correlation with STD, to be a valid surrogate change in measured behavior would have to fully capture the net effects of the intervention on STD.”1 Although this definition of a valid surrogate may be accurate, a measure that allows one to distinguish between clients who do or do not have risky partners would be a necessary first step in untangling the behavioral/biologic relationship. Peterman et al's inclusion of condom use (i.e., unprotected sex with occasional partners) as part of the risk definition is problematic if one is interested in understanding the relationships between condom use and STD incidence. It seems reasonable, however, to assume that those who have sex with either a new partner, an occasional partner, or a main partner who is having sex outside of the relationship are at higher risk than those who have sex only with a main partner who is believed to be monogamous. Consistent with this, participants in Project RESPECT with potential high risk partners (i.e., new, occasional, or nonmonogamous main partners) were significantly more likely to acquire an STD than those with potential low risk partners (18.5% versus 10.4%; relative risk (RR), 1.78; 95% CI, 1.46-2.17).11 Project RESPECT also assessed people's perceptions that their partners put them at risk for acquiring HIV. That is, as part of the baseline questionnaire, clients were asked to indicate on seven place likely (7)/unlikely (1) scales whether having unprotected sex with their main and/or occasional partners would increase their chances of acquiring HIV. Not surprisingly those who felt it was likely that unprotected sex would increase their chances of acquiring HIV (i.e., those with scores of 5, 6, or 7) were, in fact, significantly more likely to acquire a new STD (19.4%) than those who perceived that their partner(s) did not put them at risk (i.e., those with scores of 1, 2, 3 or 4; STD incidence = 12.6%). Equally important, and consistent with expectations, those who perceived they were at risk were significantly more likely to always use condoms (23.4%) than those who perceived they were not at risk (13.1%). Somewhat surprising, clients perceptions of the risk status of their partners were not highly correlated with whether they were actually with a potentially safe or dangerous partner (r = 0.11, P < 0.001, df = 3244). Nevertheless, both risk scores independently relate to STD acquisition. More specifically, those who are high on both actual and perceived risk are almost 4 times as likely to acquire a new STD than are those who are low on both actual and perceived risk. Those with mixed patterns (i.e., high/low or low/high) are intermediate in STD acquisition.11,12 So, in contrast to the Peterman et al conclusion, it does appear possible to not only find behavioral measures that distinguish between those who are more or less likely to acquire a new STD, but equally important, it appears possible to identify measures that distinguish between those who are, or are not, having sex with partners who are potentially placing them at high risk for acquiring an STD. Now let us assume that this combined measure of actual and perceived risk is accurate in identifying whether a person is having sex with a risky (i.e., an infected) or a safe (i.e., a noninfected) partner. If this were the case, then whether one did or did not use a condom should make no difference with low risk partners, but condom use should make a major difference with high risk partners. More specifically, correct and consistent condom use should significantly reduce STD incidence among those having sex with potential high risk partners. But note the use of the term “correct” as well as consistent. As we've pointed out elsewhere,2,3,13 consistent condom use does not necessarily equate to correct condom use, and incorrect use does permit the transmission or acquisition of STDs, including HIV. Thus, one must go beyond looking simply at the difference between the number of sex acts and the number of times condoms were used, and consider whether condom use did or did not protect against STD transmission and acquisition. Interestingly, among the Project RESPECT participants who were asked about incorrect condom use, over 70% report condom use that was not always fully protective during the past year (i.e., they report slippage, breakage, starting sex with a condom but removing it and continuing sex, starting without a condom but putting one on later, etc.).2,12 Thus, these respondents are all potentially capable of acquiring or transmitting an STD. Clearly, looking only at consistent condom use is not enough. It is only when one uses condoms consistently AND correctly that significant decreases in STD incidence can be expected. When “condom errors” occur, 100% condom users are probably not very different from those who use condoms only some of the time. The above hypotheses are strongly supported by the data from Project RESPECT.11,12 More specifically, while condom use was unrelated to STD incidence among those with low risk partners, correct and consistent condom use significantly reduced STD incidence among those at high risk. Not only are these findings important for understanding the relationship between condom use and STD incidence, but also they provide evidence for both the validity and utility of self-reported behaviors. Indeed, it is now safe to say that if behavioral interventions significantly increase correct and consistent condom use among people who perceive they are at risk and/or who have a new, an occasional, or a nonmonogamous main partner, there will be a significant reduction in STD incidence. On the other hand, if, among this high risk group, we only increase consistency of use without increasing correct use and/or if we only increase condom use among those at low risk, then we will see little or no reduction in STD (or HIV) incidence. Clearly it is time to stop using STD incidence as a gold standard to validate behavioral self-reports and to start paying more attention to understanding the relationships between behavioral and biologic outcome measures. Although Peterman et al perhaps are correct in arguing that many people have taken behavior change as a surrogate for HIV incidence, it should be realized that most behavioral scientists do NOT view behavior as a surrogate for STD or HIV. Instead, they view behavior change as a legitimate end point in and of itself. Clearly, HIV prevention behavioral interventions are designed to increase condom use, promote abstinence, or reduce numbers of partners because the epidemiologic evidence suggests that these changes will affect HIV incidence. However, it is important to recognize that the intervention per se is directed at increasing condom use, promoting abstinence or reducing partners. Such an intervention is quite different than an intervention designed to reduce STD incidence. That is, if we were asked to design an intervention to reduce STD incidence, it would, among other things, contain an STD control program, and a major component of the intervention would be directed at increasing the likelihood that people would come in for both screening and early treatment. While attempts to increase correct and consistent condom use would certainly be an important component of the intervention, it would not be the only one. Given that there is no simple relationship between behavior change and STD incidence, it is time to stop asking behavioral interventions to do more than they were designed to do, and to accept behavior change as a legitimate outcome measure in HIV prevention trials when it is impractical or not feasible to assess HIV incidence. Perhaps more important, it is time to start designing behavioral interventions to reduce STD incidence rather than to change one or more epidemiologically identified behaviors. And while we recognize that from a public health perspective, the most important question is not whether we have changed behavior but whether we have reduced disease transmission, the data clearly indicate that at least under certain circumstances, behavioral change can reduce disease transmission.13–15 Moreover, there are circumstances where, even though we may not see any immediate impact on disease, increasing correct and consistent condom use is an important public health goal in and of itself. Thus the real question is not whether behavior change is or is not a surrogate for STD or HIV, but rather what are the circumstances under which a behavior change will lead to disease reduction. Unfortunately, even when we are able to demonstrate that behavior change does lead to reductions in STD incidence, we may still know very little about the impact of the intervention on the HIV epidemic.