I am a physician, with some training in a variety of fields, who by accident found myself leading a series of clinical trials of pre-exposure prophylaxis, or PREP, including the iPREX trial for gay men and trans women, and the ADAPT trial of non-daily PREP. My experience led to some thoughts about evidence-based medicine and the social interplay of intentions, desires, and decisions.
In this paper, I describe an interpretative framework that is called “intention to treat,” which is one way for analyzing randomized clinical trials. Intention to treat focuses only on the perspective of the treater, imagined to be making a choice between 2 or more alternative therapies. This choice is embodied in the randomized assignment to receive one therapy or another, and whether this choice predicts the outcome of interest.
For example, in a PREP trial, the intention to treat analysis considers only the randomized assignment to offer PREP, or a PREP placebo, to determine whether that intention to treat alters the subsequent incidence of HIV infections. Information about the person and their social context is not considered in the analysis: Social context is reduced to site id numbers that are analyzed as random effects during the analysis if they are considered at all. Importantly, the intention to treat analysis considers only whether people were offered PREP or a placebo, and includes no information about whether the pills were taken frequently, occasionally, or not at all, whether there were missed opportunities to pick up more PREP tablets, or the circumstances in which these behaviors occurred.
Intention to treat analysis has some advantages. Such analysis is difficult to manipulate, so it appeals to regulatory authorities who work with parties who often have substantial interests in the outcome. Private industry has multiple concerns, and some are charitable, but the primary concern is profit. Academic researchers, like myself, have egos invested in the work, so the potential for bias is ever present. Having an interpretative approach that is difficult to fudge has real value.
The intention to treat analysis framework also makes the choice of treatments accountable for all consequences of the decision: whether the treatment works, whether it is used, and whether it is toxic. For those who feel equally poised between offering two alternative treatments, intention to treat analysis is satisfying.
Intention to treat analysis of randomized clinical trials has played pivotal roles at several junctures in the AIDS epidemic. Huge controversies arose around the perceived benefits and costs of AZT monotherapy, and such controversies served to give the perception of equipoise between using and not using AZT. Eventually, intention to treat analyses showed that AZT monotherapy increases survival by 9 months on average, but only 9 months, and there were toxicities. Amidst huge controversy, randomized clinical trials showed that antiretrovirals given to pregnant women decrease transmission to infants. Randomized trials also have had negative results that matter: some were convinced that therapy to suppress herpes would be valuable for prevention of HIV transmission, although the trials showed no benefit. A vaccine that induced a type of immune response, called cell mediated immune responses, was thought to be promising, but ended up showing increased rates of infection. In each of these situations, the trials shed light where there had been heat, better treatments became standard of care, and treatments that were ineffective or harmful were abandoned.
The fate of PREP remains to be seen. Many regard the intention to treat analysis of to provide “incontrovertible” evidence that PREP can prevent people from acquiring HIV. Evidence from these trials led to approval of the drug for prevention by the US FDA, guidance from the Centers for Disease Control and the South African infectious disease society, and WHO recommendations for demonstration projects in a diversity of settings to learn how best to provide PREP. San Francisco put the prevention use of the drug on its formulary, making it available regardless of insurance coverage or ability to pay. Private Insurance companies pay for PREP in the United States, although partly because they do not have an administrative mechanism to avoid paying for it.
Some remain very much in doubt about whether and how PREP should be used: health care providers are often reluctant to provide PREP and others are reluctant to ask for it.
Demand for PREP continues to be low almost everywhere, although it is growing in San Francisco. Typically, less than 10% of gay men have ever heard of the concept, although higher rates of awareness are emerging in regions where there has been access to PREP. PREP is used by less than 1% of MSM and transgendered women, and even less among heterosexuals. A few hundred Parisian gay men are still willing to be randomized to receive a placebo in a trial sponsored by ANRS. So far, PREP is not an important part of most people’s lives.
There are many barriers. I suspect that you have heard of some favorites. Questions are raised about PREP acceptability, the feasibility of providing prep in clinics already overwhelmed by treatment demands, cost effectiveness of PREP, drug resistance if PREP fails, adherence to PREP, fear of diversion of resources from treatment programs, profiteering by the pharmaceutical industry, and stigma. Each of these deserve careful consideration.
I will focus on one of the most widely cited concerns about PREP: so called “risk compensation” or “behavioral disinhibition”.
The idea of risk compensation reveals much about how people think about intentions, desires, and decisions. Both “risk compensation” and “Behavioral disinhibition” assume that the desire for sex is constant, if not instinctive, and that desire is contained primarily by shame or fear of HIV. In the theory of risk compensation, people are assumed to make sexual decisions in a rational manner: they are assumed to have a maximum tolerated level of risk of HIV, such that a feeling of safety afforded by use of PREP will result in taking of more risks. The notion of “behavioral disinhibition” is more historically situated in the dramatic changes that occurred in sexual practices in urban centers after the discovery of AIDS – the idea is that releasing people from fear of HIV will return everyone to the sexual practices of the 1970s. While the terms “risk compensation” and “behavioral disinhibition” are often used interchangeably by proponents of fear mongering, the two formulations imply very different concepts of decision making. This partly explains the staying power of these theories – they are shape shifters.
Regardless of formulation, the potential user of PREP is regarded to be taking destructive risks, and to be driven by dangerous desires and mistaken intentions. This is in marked contrast to attitudes about potential providers of treatments, including PREP, who are presumed to be taking constructive risks, and making decisions that are rational, beneficent, and guided by evidence.
How much are notions of risk compensation based in evidence? In reviews of the literature conducted by my group, and separately by others, evidence for risk compensation and behavioral disinhibition is scant. Despite fears and expectations, almost all biomedical interventions engender other prevention efforts. PEP, PREP, circumcision, herpes suppression, vaccines all have inspired safer behavior whether or not the intervention worked or was considered promising. In fact, just enrollment in a cohort offering nothing more than a computer based interview every few months was associated with safer behavior. Such reports of safer behavior do not satisfy the devotees of “risk compensation theory,” who assume that reports of safer behavior are inventions to avoid shame. However, we see similar trends toward safety in the incidence of syphilis and acute HIV infection in the iPREX trial, and these laboratory measurements are hard to fudge. Further, we are learning from ethnographic studies that participation in any prevention cohort brings attention to HIV as a problem, highlights solutions, and builds communities interested in those solutions – all this fosters protective behavior.
So the theory of risk compensation and behavioral disinhibition have little evidence, and the premise that sexual practices arise from rational decision making seems absurd to anyone who has recently been on a date or read the latest scandal in the newspapers. Yet, these theories are regarded to be self-evident, and are used to guide study designs, set research agendas, and make public health policy. The theories serve to reinforce the notion that those at risk for HIV have dangerous intentions and desires.
What are the practical implications of a strict focus on intention to treat in medicine and public health? Staying with the example of PREP, the differing interpretative frames for trial results come to markedly different answers to the question: does prep work? In the intention to treat analysis of iPREX, the PREP trial in gay man and trans women, the efficacy of PREP was estimated to be 42%. When we look people who took PREP daily, and compare them to the placebo group and non-users, HIV risk was reduced by an estimated 99%. Indeed, so far no gay man or trans woman has become infected whose blood levels of drug indicate use of PREP 4 or more times or more week. The two interpretative frames give different answers that are relevant to differing intentions – that of the treater and the treated.
Which interpretative frame is discussed? Typically, the intention to treat information is always cited. 42% is number that potential users of PREP have heard, this is the number displayed in medical talks, this is the number cited by advocacy organizations, who desire credibility among scientists to support their policy agendas. 42% is the number cited by ministries and potential users who say that 42% is not good enough for implementation.
There are scientists who insist that the intention to treat estimate of efficacy is the only valid estimate. They claim that people who use PREP are safer in many other ways – imagining that adherence to PREP regimens is linked to adherence to condoms, commitment to ideals of monogamy, and careful surveillance for sexually transmitted diseases. The adherent person is deemed to have a well organized personality, a strong sense of social responsibility, and a consequent capacity to keep sexual cravings in check. The non-adherent person is none of these lauded features. Hence, the nearly complete protection associated with use of PREP is deemed not to have anything to do with PREP, rather their PREP use is an emblem of their adherent, organized, controlled and safe life.
Are PREP users really using more condoms and having less intercourse partners as presumed by the champions of intention to treat? That is not what what PREP users told us. In fact, use of prep was associated with sexual practices that are regarded to be more risky – receptive anal intercourse without a condom. This suggests that PREP works well in people who use it despite sexual practices that could otherwise transmit HIV.
In summary, uncontroverted evidence that PREP is effective has not led to PREP use. Most users have not heard anything about PREP, and those who have heard about PREP have heard the benefits are low, cutting risk of HIV by only 42%. The information that PREP was nearly completely effective when used has not been disseminated partly because of an exclusive appreciation of intention to treat, and suspicion of the intentions of PREP users, including reference to people as “truvada whores” as appeared in a Huffington Post Blog. The lack of demand for prevention services, packaged as they are in negative regard, is taken to be more evidence that people who are exposed to HIV cannot be trusted. This situation has reinvigorated focus of prevention research on directly observed interventions that are controlled by the treater – such as vaccines and PREP regimens based on injections every 3 months.
What is there to do now? People who are exposed to HIV are not helpless victims. We are already seeing signs of resistance. Young people presumed to be too disorganized to take a pill a day even to avoid a deadly disease, somehow manage to appear perfectly adherent to the study protocol. Their pill counts perfectly match their self-reported pill use in interviews which perfectly match what they report on computer assisted structured interviews – everything looks good except the complete absence of any detectable drug in the bodies. People sign up for the benefits of being a trial participant, even if they choose not to take a placebo. When investigators casually referred to study participants as being “high risk” they complained to the ethical review committees, saying that they were not high risk, only popular and friendly. Questions about the intentions of treaters, including profiteering, and hiding evidence of adverse events arise partly in retaliation for the presumption that people exposed to HIV have dangerous desires and intentions.
PREP practices, or non-practices, arise from the intentions, desires, and decisions that play out between treaters and the treated. Appreciation for the intentions of the treater has provided clean uncontroverted evidence. Abjection of the intentions of potential users has made this evidence useless. In practice, the intentions are socially constructed such that the intention to treat is driven by largely by demands for treatment. Similarly, desire for sex and decisions to have sex arise from social interactions, rather than arising from individual risk/benefit calculations. Consideration of the desires, intentions, and decisions of PREP users requires that there be PREP users, at least few of whom are brave enough to participate in the discussion. Such PREP users are appearing, and their intention to take PREP is clear. We will see what happens next.