Abstract
Despite the social and regulatory expectation that contemporary clinical trials for new pharmaceutical drugs will include a diverse set of research participants, achieving appropriate representation in clinical research remains tenuous. Of ongoing concern is how experts and regulators navigate decisions about drug approvals when presented with clinical studies that have limited demographic data. Drawing on regulatory discussions about Descovy, an HIV-prevention drug that was studied only in cisgender men and transgender women who have sex with men, we analyze a contentious debate over the meaning and impact of including and excluding certain populations from clinical trial design. Extending prior work in science and technology studies on how epistemological frameworks in clinical trials matter for concerns about the production of knowledge and social justice, we show how different conceptualizations of inclusion and equity (specifically, equity in data versus equity in access) come into tension in deliberations over pharmaceutical drug approvals. We argue that the discursive conflict over gender and inclusionary/exclusionary research practices that emerged in the case of Descovy points to an underappreciated feature of equity—temporality—that should be attended to when examining knowledge production in 21st-century clinical and regulatory landscapes.
Does an inclusionary research ethic infuse 21st-century biomedical research practices? For the past three decades, the inclusion of women and underrepresented minorities in clinical trials for new drugs has been a federal policy priority (NASEM, 2022); importantly, it is no longer considered appropriate to exclude from clinical trials the populations that are most in need of the drug under study (Epstein, 2007). Yet, diversity in biomedical research has not been fully realized (Bibbins-Domingo et al., 2022), and clinical studies that are funded by the pharmaceutical industry still tend to underrepresent women (Cottingham & Fisher, 2020). A pressing concern remains: What happens when inclusionary research expectations are not met, when the norms of knowledge production are flouted?
In the fall of 2019, the U.S. Food and Drug Administration (FDA) authorized the HIV-prevention drug Descovy, making it only the second drug to be approved by the FDA for the purpose of pre-exposure prophylaxis (known as PrEP). The first such pharmaceutical, Truvada, was approved in 2012 for all at-risk adults. By contrast, the approval of Descovy carried an important caveat: It explicitly excluded those ‘who have receptive vaginal sex’, meaning that the drug was not indicated for most cisgender women or people who were assigned female at birth. This outcome was not what the drug’s manufacturer, Gilead Sciences, had asked for during a 2019 meeting of the Antimicrobial Drugs Advisory Committee (AMDAC) of the FDA. 1 Despite presenting clinical trial data from studies that only included cisgender men and transgender women who have sex with men, Gilead requested an approval for Descovy that would include cisgender women and adolescents (FDA, 2019). Their request did not sit well with some members of the AMDAC, which was charged with discussing and making recommendations about Descovy. One expert member of the panel called the science behind the request ‘whimsical’, while others expressed how ‘appalled’ they were at being asked to consider a drug approval for cisgender women without available data on cisgender women (FDA, 2019). On October 4, 2019, the chief of infectious diseases at Massachusetts General Hospital, Dr. Rochelle Walensky (who would later lead the Centers for Disease Control and Prevention during the Covid-19 pandemic), told the New York Times, as an independent observer of the AMDAC, that the exclusion of women and other at-risk groups from the Descovy trials ‘should be unacceptable in these days and times’ (Mandavilli, 2019).
To be sure, the gender-exclusionary data that were presented to the FDA for a drug that could be a critically important prevention tool in the ongoing HIV/AIDS epidemic—in which women represent close to half of new cases globally (UNAIDS, 2023)—was shocking to many in the medical and scientific community. In the case of Descovy, cisgender women were not represented at all in the data—and, so, in asking the FDA to consider approval for a population that had not been studied in its drug trials, Gilead presented exclusionary data to make an inclusionary appeal for approval. The company put forward a unique argument based on a particular approach to the extrapolation of data, which set off an intense debate about the meaning and impact of excluding population groups from clinical trial design—especially groups that could stand to benefit greatly from the drug. Indeed, while most members of the FDA advisory committee expressed astonishment and dismay at the request, some decided to vote to extend the approval to all women, citing the inequitable consequences of not doing so. However, amid much debate, what emerged as paramount was the exclusionary data that were undergirding the FDA’s discussions, ultimately leading to formal approval of the drug only for cisgender men and transgender women who have sex with men. 2
Science and technology studies (STS) scholars have long been interested in pharmaceutical knowledge production, with much attention paid to whose bodies and which population groups provide the basis for such knowledge production (e.g., Cooper & Waldby, 2014; Cottingham & Fisher, 2020; Epstein, 2007; Fisher, 2009, 2020; Fisher & Kalbaugh, 2011; Peterson et al., 2015; Petryna, 2009; Sismondo, 2015; Shim et al., 2022; Sunder Rajan, 2017). In highlighting that knowledge production in clinical trials tends to be responsive to pharmaceutical interests, Sismondo (2008) has noted that we must pay attention to design biases that shape the formulation and execution of clinical trials. Moreover, the clinical datasets that undergird discussions about drug approvals are, like all datasets, socially produced and never crystal clear. As constructed entities, they reflect certain values and contain unresolved tensions. Because of this, the contours, merits, and shortcomings of the knowledge produced by a clinical trial come to matter greatly in the FDA decision-making process. Tracing the knowledge and practices that led to the approval of the first PrEP drug, Truvada, Davis (2020) shows how HIV clinical researchers deploy new techniques that shape how we come to understand the subjects and utility of clinical research. In doing so, Davis illuminates how such techniques of clinical knowledge-making rest on practices that exclude the most vulnerable from the clinical imaginary: specifically, individuals who are at high-risk for HIV but who are considered ‘poorly adherent’ to prophylactic regimens. What makes Descovy worthy of analysis is the exclusion of a different group (cisgender women) that is in great need of HIV prevention options, especially from the perspective of global public health, and the fact that this exclusion from the clinical research came shortly after a broader indication for Truvada. Building on the robust STS literature on knowledge production and exclusion/inclusion in medical research, we highlight and analyze a contemporary iteration of the inclusionary research era—the inclusion-through-exclusion approach proposed in the case of Descovy—that prompted acute reflections about risk, equity, and difference.
Although multiple population groups were excluded from clinical trial testing for Descovy, including transgender men and injection drug users, and although multiple potential indications were up for deliberation in the Descovy approval process (i.e., Descovy for cisgender men and transgender women, for cisgender women, and for adolescents), our focus in this article is on the intense social and regulatory conversation that emerged around possibly extending the drug’s approval to cisgender women—a debate that, as we discuss below, comprised the majority of the meeting and may have had the effect of foreclosing conversation about other population groups. To do so, we draw on multiple sources, primarily the 350-page transcript of the FDA advisory committee meeting on Descovy, as well as the associated public comments submitted to regulations.gov. 3 Participating in this crucial meeting were representatives from Gilead Sciences, the FDA, and the public (e.g., affected individuals, members of advocacy organizations, and other concerned constituents), as well as outside medical and scientific experts who were charged with reviewing the drug application and then voting on whether to recommend approval of the drug and whether to include cisgender women in the drug’s indication. Finally, we draw on statements from activist groups (e.g., PrEP4ALL; Treatment Action Group) as well as media reports.
Our concern in this article is not with the scientific accuracy or legitimacy of the extrapolation claim that was presented to the FDA’s advisory panel and that was used to make an inclusionary appeal based on exclusionary data, though it was contentious among panel members. Rather, we focus on how experts reacted to it, how their deliberations fit into the normative 21st-century narrative of knowledge-making around HIV prevention (and clinical trials for pharmaceutical drugs, broadly), especially regarding specific populations, and how their disputes reflected questions about the expectations of inclusion and exclusion in contemporary clinical trials. Extending prior work that focuses on moments of debate around PrEP research (e.g., Peterson et al., 2015), as well as work that shows how FDA advisory meetings and decisions reveal contentions among social worlds (e.g., Davis, 2020) and much contingency (e.g, Zhao et al., 2023), we ask: In an era of inclusionary research expectations, what does the debate over this case reveal about clinical trial epistemologies and principles of equity and justice in biomedical research?
We find that all stakeholders shared the desire to expand prevention options for all population groups; but, given the exclusionary nature of the data presented for inclusionary access, divisions emerged over the meaning of equity. Different framings hinged on a temporal dimension—whether equity was addressed early in the research process (inclusion in clinical trials) or later (inclusion in access). This conflict had as its backdrop the contemporary social and regulatory expectations of an inclusionary research ethic which, in this case, was falling short according to observers. Extending prior STS work on how epistemological frameworks in clinical trials matter for concerns about the production of knowledge and social justice, we show how exclusionary clinical trial practices paired with inclusionary appeals have a cascading impact on concerns about equity. Specifically, we demonstrate how different conceptualizations of inclusion and equity (equity in data versus equity in access) come into tension in deliberations over pharmaceutical drug approvals. In doing so, we discuss an underappreciated feature of equity—temporality—that should animate future STS studies about biomedical knowledge production.
The HIV/AIDS epidemic and new prevention tools
The HIV/AIDS epidemic is in its fifth decade. Globally, 1.3 million people were newly diagnosed with HIV in 2023, and cisgender women and girls comprise close to half of all new HIV infections (UNAIDS, 2023). Moreover, women and girls are disproportionately affected in certain geographical regions. In sub-Saharan Africa, for example, women and girls accounted for 62% of all new HIV infections in 2023. In the U.S., women accounted for 18% of new HIV diagnoses in 2022, but Black women are disproportionately affected, comprising 50% of new diagnoses among women (CDC, 2024). In addition, pregnancy is a time of heightened risk for contracting HIV (PHASES, 2020), which disproportionately affects women, and especially women in the Global South.
PrEP drugs are intended for HIV-negative people who are at risk of contracting HIV and are considered a crucial prevention tool for many people, including women. Dr. Demetre Daskalakis, who was previously the director of the division of HIV Prevention at the CDC, announced in 2021 that ‘PrEP is one of the most powerful tools we have to prevent HIV transmission. Expanding access to PrEP will be critical’ (CDC, 2021). PrEP is also considered to be an important tool for ending the epidemic by the year 2030 globally (HIV.gov, 2023)—an unrealistic goal, but one that nonetheless mobilizes prevention efforts in important ways. Clinical trials for new and improved treatment and prevention drugs are an integral part of the ongoing work to tackle the epidemic. Still, many population groups are left out of clinical studies, even though, over time, the inclusion of various population groups in HIV/AIDS clinical trials has been of foremost concern.
Inclusion, clinical trials, and knowledge production
Although randomized controlled trials (RCTs) have many limitations (Bothwell et al., 2016), they have long formed the basis of knowledge for drug safety and efficacy in the United States. Since the 1960s, the FDA has had broad oversight of the clinical trial process for pharmaceuticals (Carpenter, 2010; Greene & Podolsky, 2012; Waggoner & Lyerly, 2022). Coinciding with the FDA’s growing authority over clinical trials in the U.S. was the rise of the field of bioethics and the codification of ethical principles in medical research (Campbell & Stark, 2015). With the 1972 exposure of the U.S. Public Health Service Untreated Syphilis Study at Tuskegee, in addition to other research abuses, bioethicists were attuned to the need to protect population groups from the possibility of research harms. Thus, the foundational medical ethics principles that were promulgated in the 1970s and that remain central today—beneficence, non-maleficence, autonomy, and justice—emerged alongside a ‘protectionist’ ethic that did not foreground inclusion but rather underscored the notion of protection from research. This ‘protectionist’ ethic infused medical ethics and medical research throughout the latter part of the 20th century (Mastroianni et al., 1994), meaning that many population groups were excluded by design from clinical trials so as to protect them from encountering the risks that are inherent in clinical research studies.
This was certainly the case with women, who were systematically denied access to participate in clinical trials after the FDA’s 1977 recommendation that all women of ‘childbearing potential’ be excluded from early-phase drug trials (FDA, 1977). As part of the broad project of the ‘protectionist’ ethic that aimed to avoid introducing harm to ostensibly ‘vulnerable’ groups, many other population groups were regularly excluded from clinical trials as well, including individuals from racial and ethnic minority groups, gay men, and children. Over the last three decades of the 20th century, advocates for broader inclusion for population groups in clinical trials increasingly argued that key ethical principles—like justice—in fact require protection through research, not protection from research (Johnson & Fee, 1994; Mastroianni & Kahn, 2001; Waggoner & Lyerly, 2022). Bioethicists, regulators, and policymakers were alert to the fact that research was predominantly conducted on heterosexual white cisgender males and, thus, that many approvals for drugs that were indicated for and used by diverse groups of people were based on limited data.
It was advocacy from gay activists during the early years of the HIV/AIDS epidemic that changed how clinical trials are conducted, as this advocacy spurred long-lasting conversations about what proper inclusion looks like in clinical trials and challenged epistemological foundations of the ‘gold standard’ randomized controlled trial. By questioning statistical endpoints of efficacy and advocating for access to unapproved, but potentially lifesaving, drugs, HIV/AIDS activists urged people to consider the relationship between the bodies that are studied in a clinical trial—including restrictions about what such bodies ‘can’ and ‘cannot’ do as trial participants—and the people who will be the end users of the drug (Epstein, 1996). Activists argued that exclusionary data are insufficient in vital and meaningful ways, and chief among concerns about this impoverished knowledge base was the problem of extrapolation. Experts argued that medical knowledge cannot simply be extrapolated from one population group to another (Epstein, 2007). Women’s health advocates also crucially surfaced the problem of exclusionary data and extrapolation (Johnson & Fee, 1994; Levine, 1993).
This advocacy work culminated in major regulatory changes that occurred in the early 1990s, such as a new National Institutes of Health (NIH) policy that obliged federally funded projects to include women and minorities as subjects in clinical research as well as revised FDA guidelines that removed exclusions on the participation of women in clinical trials for new drugs (Epstein, 2007). Despite thoughtful concerns from STS scholars about how testing drugs in different population groups can legitimize and essentialize difference (Epstein, 2007; Shim, 2000), the governing scientific and regulatory consensus that emerged was that bodies from distinct population groups may process and metabolize drugs differently, and that to truly—and ethically—assess the safety and efficacy of drugs, testing in specific populations must occur.
However, shifts to inclusionary practices in biomedical research did not fix the knowledge gap problem for all population groups. Scholars and ethicists soon questioned the continued exclusion of children from clinical trials (arguing that children are not just little adults), as well as the persistent exclusion of pregnant women from clinical trials (arguing that pregnant people are not just people with protracted bellies) (Lyerly et al., 2008). Even today, women remain underrepresented in clinical trials (Abbasi, 2023; Cottingham & Fisher, 2020). Black women, especially, are not typically centered in research, prevention, or treatment concerns (Fletcher et al., 2021; Watkins-Hayes, 2019), and pregnant people remain ‘protected’ from research, despite great need for a better evidence base for therapeutics and biologics during pregnancy (Waggoner & Lyerly, 2022).
When it comes to HIV/AIDS clinical trials, women have not always been well represented (Matthiesen, 2016). Interestingly, pregnant women have historically been included in HIV/AIDS clinical trials because of ardent desires to reduce maternal-to-fetal transmission of the virus (Faden et al., 1996). But while pregnant women have been included in antiretroviral trials, they have been left out of PrEP trials (see Sullivan & Lyerly, 2017). With the vast number of women globally, especially in the Global South, who are at-risk of contracting HIV while they are pregnant, this oversight seems significant. Exclusion matters because the way that clinical trials are conducted shapes the scientific knowledge generated from the research. In the case of pregnant people, for example, their exclusion from clinical research has resulted in wide gulfs of knowledge about their care (Lyerly et al., 2008).
Today, the NIH, FDA, and others tout inclusion and representation in clinical trials as a way to build equity for women and underrepresented groups (NASEM, 2022). And yet, as is the case with Descovy, there are instances where the very meaning of inclusion vis-à-vis equity becomes contested.
The case: Expanding the PrEP ‘armamentarium’
On August 7, 2019, the Antimicrobial Drugs Advisory Committee (AMDAC) of the U.S. FDA convened to discuss a supplemental new drug application (sNDA) for Descovy, emtricitabine 200 mg and tenofovir alafenamide 25 mg, submitted by the pharmaceutical company Gilead Sciences and proposed for pre-exposure prophylaxis (PrEP) to reduce the risk of sexually acquired HIV-1 infection among individuals who are HIV negative and at risk for HIV. 4 Dr. Jeffrey Murray, then Deputy Director of the Division of Antiviral Products at the FDA, provided the opening remarks at the meeting and expressed: ‘We’re happy to be talking about expanding the HIV prevention armamentarium today.’ 5
The sNDA presented to the AMDAC was based on phase 3 of the DISCOVER trial, an international clinical trial conducted between 2016 and 2019 to assess the safety and efficacy of Descovy for HIV prevention. DISCOVER was a double-blind, active comparator, non-inferiority trial that compared Descovy to the already approved standard of care for prevention, Truvada (emtricitabine 200 mg and tenofovir disoproxil fumarate 300 mg). The trial enrolled over 5,000 cisgender men and transgender women who have sex with men, and eligibility criteria were designed to ensure that the study population was at high risk of HIV infection. Participants were required to report at least one of the following criteria: two or more episodes of condomless anal sex with more than one partner in the 12 weeks prior to enrollment, or a diagnosis of a sexually transmitted infection (gonorrhea, chlamydia, or syphilis) in the 24 weeks prior to enrollment. Gilead reports that DISCOVER recruitment and study sites were mostly urban and specifically chosen to be in locations with a high HIV incidence. The trial included 94 sites—hospitals, private practices, and local STI clinics—in 11 countries across North America and Western Europe. Each recruitment site was responsible for determining the best recruitment practices within the community, a process from which activists were excluded from participating. 6
Across the 11 North American and European countries, 84% of participants self-identified as white, 9% as Black, 25% reported being of Hispanic or Latinx ethnicity, and 74 participants, about 1-2% of the study population, were transgender women. As some AMDAC committee members and members of the public remarked—and which we discuss in more detail below—the demographic makeup of the DISCOVER trial significantly diverged from HIV incidence rates in the United States. In 2022, for example, African American individuals accounted for 39% of all new HIV diagnoses in the U.S., despite comprising only about 13% of the overall population (CDC, 2024). And, as mentioned earlier, women comprised 18% of new HIV diagnoses in the U.S. in 2022 (CDC, 2024), but made up around 50% of new diagnoses globally (UNAIDS, 2023).
Representatives at the meeting all commented on the importance of having ‘another tool’ in the arsenal of HIV preventatives and the importance of allowing individuals to have a choice about which HIV prevention option is best for them. Crucially, Descovy for PrEP followed Gilead’s 2016 approval of an HIV treatment drug—also called Descovy and with the same drug formulation—that was indicated for all adults. Moreover, Descovy for PrEP was following on the heels of Gilead’s approved PrEP drug, Truvada, which was indicated for all at-risk adults. While clinical trials for Truvada were contentious (see Davis, 2020), they did include safety and efficacy data for men, transgender women, and cisgender women (see also FDA, 2019). The safety and efficacy data for men and women paved the way for the eventual 19-3 vote in 2012 in favor of a broad indication for Truvada for the prevention of HIV.
This, of course, was not the case with Descovy for PrEP. Because Gilead requested approval for cisgender women as well as men, extrapolation emerged as the key issue for debate in the FDA advisory meeting on Descovy. When asked, the FDA acknowledged in the meeting that extrapolation happens all the time in drug applications, but that the particular extrapolation approach advocated by Gilead was novel (i.e., using non-systemic pharmacokinetic data to extrapolate from men to women): ‘I don’t think we’ve ever made a regulatory approval decision based on [this argument]’, said one FDA participant.
During Gilead’s presentation to the AMDAC, they explained that cisgender women were not included in the DISCOVER clinical trial because the HIV incidence rates in the sites where DISCOVER was conducted were 13 times lower in women compared with men in those same locations. Gilead explained that a ‘trial would require enrollment of about 22,000 women in the high incidence regions. This would require approximately 8 to 10 years to conduct’ (FDA, 2019). In other words, Gilead’s explanation for the exclusion of cisgender women rests on the exclusive reliance on randomized controlled trials for developing and testing HIV prevention technologies. Randomized trials that rely on a percentage of participants contracting HIV as an endpoint (whether that be an intervention group compared to a placebo group or an intervention group compared to an already approved alternative intervention) require high incidence or a large number of participants to demonstrate statistical significance of the intervention—a feature of RCTs that some scholars have challenged (Adimora et al., 2017). 7
When Gilead presented this reasoning to the FDA committee, people were nevertheless still shocked by the exclusive nature of the trial data. The more expansive range of data that had previously been presented for Truvada led people in the FDA meeting about Descovy to remark at just how different the Descovy conversation was from the one that was had with Truvada only seven years earlier. Dr. Jeffrey Murray of the FDA reminded AMDAC committee members of this fact in his opening remarks at the August meeting to discuss the sNDA for Descovy: ‘I remind you the applicant submitted two clinical trials, a clinical trial in MSM [men who have sex with men] and transgender women, iPrEx, and a trial in discordant heterosexual couples, Partners PrEP, which allowed for a broad indication [for Truvada] among at-risk populations. Today we’re dealing with one clinical trial’ (FDA, 2019). One expert member commented, ‘I feel like we’re moving backwards from the 2012 meeting that approved Truvada, in which we had data on women, and now we don’t have data on women at all.’ Another committee member asked the FDA representatives to comment on ‘how we should view the appropriateness of an application for a drug that intended all the while to have an indication in men and women, coming in with clinical trial data only for men with an expectation to apply to women?’
This quote gets right at what rubbed so many in attendance the wrong way—that Gilead expressly excluded women despite the fact that women would be likely users of the drug. In fact, one committee member noted that Gilead, in its presentation to the AMDAC, mentioned that it is planning studies in women, highlighting that an inclusive trial is, indeed, possible. As discussed above, ethical clinical research design should keep in mind the end users of drugs. And this case of exclusion, amid a request for an inclusive indication, was seen by some as a strange way to go about equity concerns in an era of inclusion, especially if such an approach carries implications for precedence around inclusion and equity in clinical research design and drug approval decisions.
To be sure, while the controversial approval process for Descovy is but one case among many drug approvals that come before the FDA, it affords insight into the era of inclusion in biomedical research. Every discussion about a drug approval reflects current thinking about inclusion, safety, efficacy, and precedent at the FDA, and how experts, sponsors, and publics respond to new ideas shapes policy. And while the FDA may make an approval decision that is not in line with an advisory committee’s recommendation, the agency is nevertheless concerned with how its deliberations and approvals will set the stage for future decisions. In the case of Descovy, the FDA acknowledged in the meeting that the argument being presented by Gilead was novel, setting up the committee’s deliberation and the agency’s eventual decision as potentially precedent-setting concerning expectations around inclusionary practices and the extrapolation of exclusionary data. Advisory committee members were attuned to this, and they openly worried about whether an inclusion-through-exclusion approach would set a dangerous precedent.
The specific claim used by Gilead to include cisgender women thus produced a contentious scientific and ethical discourse, one that panel members worried would have far-reaching implications for inclusion, equity, and clinical trials, and that crucially shaped the committee’s discussion about the latest addition to the PrEP toolkit. In the case of Descovy, the charge for the FDA advisory committee was how to navigate the possible approval of a drug that had limited and exclusionary data, and to do so at a particular moment in time: well after inclusion had become the normative expectation in clinical research and well after calls for more equitable inclusion in clinical trials for women and pregnant people (especially regarding HIV/AIDS).
We see this case—the sNDA presented to the FDA and the acutely social debate that ensued—as a test of the fortitude of the inclusionary research ethic that has been in place since the 1990s. In what follows, we show that the exclusionary data were received by many in the FDA meeting as a reflection of poor scientific knowledge production that failed populations in need, particularly women. We then detail the debate over equity that emerged, as meeting attendees struggled to decide the best course of action amid a paucity of evidence. In doing so, we explicate committee members’ and members of the public’s varied emphases on the temporal nature of equity in biomedical research, with implications for notions of inclusion, difference, and ethics in contemporary clinical trials.
Failing science, failing populations in need
It became readily apparent in the FDA meeting that all parties—from Gilead to the advisory committee members to advocacy groups—cared deeply about improving PrEP drugs and access to them. It also became clear that attendees were very concerned that inclusionary research expectations were not being met. As representatives from Gilead made the case for extrapolating data from cisgender men to cisgender women, some flagged the harms of an inclusionary appeal based on exclusionary data. To many, the lack of inclusionary data represented a failure in scientific knowledge production and set a potentially harmful precedent for both the FDA and for populations in need.
To explain the problem, a representative from the AIDS Vaccine Advocacy Coalition (AVAC), an international non-profit that advocates for the ethical development and equitable delivery of HIV preventatives, succinctly stated during the public comment portion of the meeting that ‘today we sit in somewhat of an evidence-free zone’ (FDA, 2019), which is notably a distinct scenario from that of the meeting that was held on Truvada seven years prior. Despite the lack of evidence, though, the community felt that the discussion about a broad indication was still necessary. The same AVAC speaker went on to explain: ‘But the data are the data, and we must act on that most urgent data point presented.’ This speaker stressed how important it is to ‘always be clear that safety and efficacy matter’, but also how worrisome it would be if the indication were not extended, because then ‘the only group that will suffer and pay the price for that decision are women at risk of HIV infection’ (FDA, 2019). Of course, it would be hugely problematic if Descovy turned out to have an unacceptable safety or efficacy profile for people who have receptive vaginal sex. Committee members did discuss the possibility of failed extrapolation, especially in terms of safety and efficacy, and many comments had to do with concerns about risk, safety, and efficacy of the drug (which was, in fact, the charge of the committee).
Indeed, the conversation was made all the more controversial and pressing due to the lack of safety data in cisgender women. Some of the data presented to the FDA suggested that Descovy may be safer and more tolerable than Truvada in certain ways (all hypotheses), yet there were not clear data on all the populations that Gilead wanted to include in the drug’s indication. The combination of the drug’s potential importance with the lack of data for key populations emerged, then, as a socially combustible analytic and practical space. As discussed above, Gilead asserted that both safety and efficacy for ciswomen and adolescents ‘can be inferred from [the] DISCOVER’ trial (FDA, 2019). Whether or not their reasoning was convincing, their argument promoted the stance that safety is something that can be inferred. It is this stance that did not pass muster with many in attendance at the meeting and among some who submitted public comments beforehand.
Of course, imperative in the drug approval process is the establishment of a risk profile for the drug under review, and so concerns in the meeting were not just about the failure of knowledge production but also about regulatory and practical implications. Many comments revolved around the fact that risk was not taken on in the clinical trial phase for certain populations, thus producing risk for some populations that might use the drug. Moreover, people worried about what a broad approval would signal for the FDA’s goals of protecting the public from the harms of unsafe drugs. Consider the following quote from a public comment submitted by the executive director of AIDS Action Baltimore: If Descovy is approved for HIV prevention in [women], it will be with insufficient safety and efficacy data. For me, the answer is clear. I earnestly believe that it would be a violation of the agency’s clearly defined statutory mandated duty to approve Descovy at least for cisgender women without sufficient safety and efficacy data.
In another letter submitted to the FDA by PrEP4ALL and Treatment Action Group, the activists claimed that Gilead’s novel approach ‘defies logic’. While the activists were careful to hedge that they do not oppose a broad indication, they expressed fear that an approval without any efficacy or safety data ‘would establish a disturbing precedent discouraging other new HIV prevention technologies from being tested in cisgender women.’ Bluntly, they stated that ‘Gilead’s failure’ puts advocates into an impossible quandary—either advocate for a delay in the approval of Descovy PrEP in cisgender women and effectively deny them the ability to choose which PrEP formulation they prefer, or advocate for the approval of Descovy PrEP in cisgender women while establishing a precedent that will disincentivize future PrEP modalities from being tested in this key population.
These organizations clearly articulated the conundrum facing the meeting attendees given the limited data presented and the high stakes that any decision would have for at-risk populations. The general perception was that certain populations were being neglected by this failure in scientific knowledge production.
Members of the FDA advisory committee shared the concerns of advocacy organizations about how the data presented did not reflect populations in need. Speakers specifically noted the lack of attention to and data on transgender individuals and African Americans. An administrator for the HIV/AIDS Bureau at the U.S. Department of Health and Human Services who voted ‘no’ on extending an indication to cisgender women shared the following as she explained her ‘yes’ vote on the indication for men and transgender women: ‘I am disturbed that this far into the epidemic and this many clinical trials, we still can’t do trials in the people most at risk in this country, representatively, and that we really do need to be looking at African Americans … and transgender women’ (FDA, 2019). One letter from an activist noted that ‘the racial and ethnic make-up of the DISCOVER trial did not reflect the makeup of the populations most vulnerable to the HIV epidemic in the United States.’ The letter cited HIV infection rates by race and claimed that the minimal inclusion of people of color, particularly African Americans, is ‘particularly concerning’.
Although other excluded population groups were certainly mentioned in the committee’s discussion, as just illustrated, it was the exclusion of women that was seen as the ultimate failure, especially insofar as it led to the production of ignorance about the drug and its potential users. A committee member who voted ‘no’ on the indication for women said that ‘women in underserved populations deserve our best efforts to make sure drugs are effective and safe for them … before we start recommending it in lieu of one that has demonstrated safety and efficacy.’ A different voting member stated that ‘we have failed women’, underscoring that ‘to be at this point and not have the data to guide decision making is a shame on all of us’. A consumer representative and voting member of the AMDAC characterized their vote as ‘a strong no’—‘There’s about eight women on the committee, and that the agency and the applicant would present insufficient data to support’ the indication in women, they said, is ‘appalling’. Finally, another ‘no’ vote came from a member who revealed the compounding equity problems that result from the initial exclusion of cisgender women from the clinical trials: ‘I feel like we have failed women by letting this application come in without data from women to begin with, and I fear we’re failing them again by having approval for use in men and not women’ (FDA, 2019).
While those in attendance shared concerns about the lack of adequate safety information for women, and while some voted no on extending the indication to women for that very reason, not all committee members voted ‘no’ despite sharing similar concerns. For example, one member who voted in favor of extending the indication to cisgender women stressed that approving it in men would result in an inequitable system; they explained:
‘Creating a two-tier prevention treatment will not be helpful, and we should remind ourselves there are more women living with HIV in the world than there are men, and that the risk of new infection is significantly higher among women if we look at this globally’ (FDA, 2019).
For this panel member, it was a ‘terrible failure’ to not have inclusive data about how the drug is tolerated by the bodies of people from different population groups.
To wit, no matter whether there was a unique and valid scientific argument to be made about Gilead’s use of extrapolation, it did not work smoothly in a regulatory arena where details are discussed among a diverse group of social actors who are attuned to inclusionary expectations. Worries about scientific knowledge production, failing at-risk populations, and equity were front and center in the committee’s deliberations. In what follows, we discuss the role of equity in scaffolding deliberations around (un)ethical scientific knowledge production and its products. While all parties involved evinced desire to expand PrEP access, important divisions emerged, most prominently over issues of inclusion and equity. It was different temporal framings of equity that shaped committee members’ reasoning for their eventual votes on whether to extend approval to cisgender women.
Framing inclusion, debating equity
In a discussion that hinged on retrofitting inclusion with exclusionary data, the scope of the meeting became about more than the details of the clinical trial—it became about how best to think about inclusion and equity when exclusion is the backdrop. As discussed above, many meeting participants agreed that the data were inadequate to the question at hand and that populations in need were not being served. Everyone, including the representatives from Gilead, openly acknowledged the need to include cisgender women in future trials. Amid this consensus on the lack of appropriate data and the importance of the drug’s availability, different framings of inclusion and equity emerged among meeting participants, voting committee members, and members of the public. A temporal concern was key: Should equity be privileged early in the research process (i.e., during clinical trial design, recruitment, and execution), or later in a drug’s timeline (i.e., access to drugs by individuals following regulatory approval)? Following other STS scholars who have written about how scientific professionals navigate and reconcile ethical dilemmas in their work (e.g., Bliss, 2012; Markens, 2013; Waggoner, 2017), we show here how shared ethical principles come into conflict in practical cases.
Privileging equity in access
Speaking for PrEP4All, James Krellenstein identified the ‘catch-22’ that was at the very heart of the discussion, highlighting the ‘difficult decision’ with which the committee was faced: deny the broad indication for cisgender women and other populations, or deny the population the drug. Even though he asked, ‘why don’t we have better data on cisgender women?’, Krellenstein argued that ‘the right choice is to extend the indication to women’ (FDA, 2019). In Krellenstein’s view, one way to approach equity in this case would be to focus on an outcome: expanding drug access to cisgender women. Despite raising concerns about how equitable practices had failed early in the research and clinical trial process, emphasis was placed on equity when it comes to access—that is, inclusion after drug approval.
Others agreed. For example, one committee member voted ‘yes’ on the indication for ciswomen because of a concern about equitable access, saying: ‘I felt it was appropriate to include cisgender women in the indication, especially given the equity issues that have been discussed during this committee hearing’ (FDA, 2019). Although many worried about the inequity that was baked into the clinical trial design, foregrounding equity when it comes to drug access allowed some meeting participants to accept the use of exclusionary data for an inclusionary appeal. Presumably, part of an equitable and ethical approach was to allow a choice of PrEP options and access to a new drug notwithstanding concerns about the data. Indeed, multiple stakeholders deployed a discourse of ‘choice’ during the AMDAC meeting. A member of the Gilead applicant team, for example, emphasized choice in varied ways, illustrated by the following exhortations: I can’t stress enough how detrimental it would be to give men a choice of a safer medicine but not offer the same choice to ciswomen. Ciswomen should have the same options that would be available to cismen and transwomen. Please at least allow women to have that choice. Let them decide if the safety advantages outweigh the concerns about efficacy. Allow them the option to choose what will be best for them. (FDA, 2019)
The above poignantly illustrate the discursive coupling of choice and the privileging of equity in access.
Mobilizing the logic of choice in medical spaces can shift the burden of care to the patient (Mol, 2008) and has the effect of individualizing risk. If Descovy were indicated according to this logic, then risk would shift to individuals and providers as they potentially take and prescribe drugs (off label) without knowledge of the drug’s effects in different population groups. This too, though, may be overstated, as professionals and patients need knowledge in order to propose and make choice-based decisions. In this vein, Gilead’s use of ‘choice’ discourse resembles a strategy to countenance the production of ignorance. The rhetorical moves around choice proved to resonate with activists and researchers in this space and served to bolster committee members’ explanations of their votes (see also D’Amours, 2023). In other words, choice and access were clearly tied to issues of equity for many of the committee members, and especially the activists and global commenters. For example, two scientists shared in a public comment that, To have a global public health impact, a choice of PrEP options is needed, from which persons who are at risk and motivated to use biomedical HIV prevention can choose the option that best fits their situation and preferences. [Descovy for] PrEP offers another PrEP option, one that should be available to men and women equivalently.
One speaker from the Abounding Prosperity organization proffered the equity argument that Descovy needs to be approved immediately for ciswomen, highlighting that ‘additional tools’ (i.e., expanded choice) would ‘empower’ Black women, MSMs, and trans individuals: ‘Women need this drug, and it will not be in the interest of public health to have this drug approved without including women.’ Or more generally, ‘everyone deserves the same choices of prevention options’ (FDA, 2019).
When faced with what was described by some as an impossible decision—whether to vote to extend the approval of a drug to a population for which no safety or efficacy data were collected—some of those present, including Gilead representatives themselves, employed a discourse of inclusion via choice to ultimately privilege equity in access. Emphasis on equity later in the temporal timeline of pharmaceutical drug development has consequences, particularly regarding the individualization of safety and risk, as well as the proliferation of liberal allusions of choice, to which we return in the discussion.
Privileging equity in data
While the above framing of equity in terms of access was powerful and persuasive to many in the meeting, it did not supplant other concerns. By a vote of 10-8, the committee ultimately voted not to extend the approval to women, and we argue that this outcome was related to equity considerations. By voting ‘no’, committee members were adhering to the need for evidence, which thereby effectively privileged equity early in the research timeline. In other words, we posit that those voting members who prevailed underscored the degree to which equity concerns matter not only during discussions about inclusionary access but also throughout the research process itself.
The chairperson of the committee quite bluntly highlighted the temporal concerns when he explained his ‘no’ vote on expanding the indication of Descovy to women: ‘the question was do we have substantial evidence of safety and efficacy data. There are no efficacy data presented’, and ‘there are side effects to our interventions. Our interventions are not benefit with no risk.’ Another voting member shared this sentiment when she explained her ‘no’ vote: ‘I do not believe the data support TAF/FTC efficacy as PrEP for women, as it’s not been studied in that population.’ Another committee member explained: ‘The absence of actual clinical data in this group … do not allow me to recommend labeling this drug as effective in cisgender women.’ And another panelist expressed: ‘It was a strong “no” for me, no wavering on the fence’ (FDA, 2019). A committee member from the CDC reflected back on Truvada—for which there was data on cisgender women—and noted the following during the Descovy meeting: The decision has been made that we’ll do the trial in MSM, and then we’ll figure out what it means for women rather than studying women themselves. I find that bad science, and that’s why I voted no, but I also find it disrespectful and an issue of … research equity. Women deserve the same quality of data about the safety and efficacy of the drugs that they’re exposed to that men get. (FDA, 2019)
The above quotes highlight how for many on the committee, equity in access cannot be realized because the scientific data collection had failed—and without establishing equitable practices early, it is not ethical to extend an indication.
That is not to say that those same ‘no’ voting committee members did not share concerns about increasing PrEP options and equitable access to drugs. On the contrary, they expressed pressing equity concerns about access: ‘I do not like the idea of approving a drug for a single population’; ‘this will only further inhibit implementation of PrEP among women’; and, ‘I really wanted to vote yes. … I think that we’re in this position is absolutely horrible’ (FDA, 2019). The slight majority of committee members therefore acknowledged what we call the cascading impacts of equity—that is, when equity is not prioritized early in the drug development and research process, serious concerns and implications about equity may follow.
This privileging of equity in data came into tension with concerns about equity in access, even amid the shared ethical goal among all stakeholders of expanding access to another option for HIV prevention. As we have shown, arguments emerged from committee members and public commenters that perhaps it is satisfactory to pursue inclusion in the post-marketing setting (rather than in the clinical trial setting); such individuals suggested that it is morally acceptable to set aside the issue of exclusion from the clinical trial in order to allow for inclusionary access and avoid the inequitable arrangement of a two-tiered prevention system. The idea here was to highlight outcomes, thereby promoting equity in drug access even if inclusion was not part of the research design. Yet, as we have illustrated in this section, others emphasized process, thereby calling social, scientific, and regulatory attention to the cascading impacts of equity. Temporality—when equity is addressed and the extent to which it has a short-term or long-term impact—emerged as an animating component in the committee’s framings of inclusionary practices and ethical knowledge production.
Discussion and conclusions
In 2022, the National Academies of Sciences, Engineering, and Medicine (NASEM, 2022) highlighted Descovy as an example of how the lack of representation in clinical trials can lead to inequitable drug access and compromise health equity goals. We have here unpacked this case to analyze how inclusion and equity were framed, and with what implications. Although inclusionary practices are expected in clinical trials today, such practices are far from uniform. Whether or not they ultimately voted to extend the drug’s indication to ciswomen, many participants in the FDA advisory committee meeting for Descovy were aghast that ciswomen were not initially included in the clinical study. When it came time to vote for a broad inclusionary approval, participants were divided on how to think most productively about the meaning and upshot of gender-exclusionary data. For some, certain ethical framings of equity (i.e., equity in access) smoothed out the problem of limited demographic data: Even if the use of exclusionary data was troubling, concerns about equitable access shaped the viability of an inclusionary appeal. For others, a different framing of equity (i.e., equity in data) magnified the problem of exclusion from the beginning and undercut any subsequent inclusionary appeals.
In the end, by not approving Descovy for those who have receptive vaginal sex, the committee agreed, as stated by the chairperson of the AMDAC, that ‘there needs to be actual data’ (FDA, 2019). This outcome broadly reveals that it is not considered enough to try to leverage exclusionary data to secure broader indications for other populations; rather, what is expected now in the inclusion era is actual inclusion at every step of the clinical trial process, including the collection of representative data. The committee as a whole could not countenance the fact that women were not included in the research phase, and this concern derailed the extrapolation argument that the drug’s sponsor presented. There were just too many unknowns in the data and quite a bit of uncertainty. In such a situation, inclusion in clinical trials (which produce data on the relevant populations) is now the social expectation and research ethics norm, even as ethical concerns such as equity come to matter differently to different stakeholders at different points in the drug approval process.
This case is also important in terms of showing what conversations did not emerge, how the science and its framings prompted certain silences. Cisgender women were excluded from the clinical trial informing the FDA’s meeting, and there was subsequently no mention of pregnancy. As discussed earlier, pregnancy is a heightened time of risk for acquiring HIV infection, and pregnant women globally—especially in the Global South—are in great need of better options for HIV prevention (PHASES, 2020; Sullivan & Lyerly, 2017). When Truvada was approved, the topic of HIV prevention and pregnancy was part of committee deliberations. With Descovy, the lack of cisgender women in the study group cut off any discussion of pregnancy. In addition, it is possible that the committee’s astonishment at the exclusion of cisgender women from clinical trials foreclosed discussion of other excluded population groups, including people who inject drugs and transgender men, that would potentially benefit from new interventions. These oversights show the cascading effects of coupling exclusionary data with inclusionary appeals, where key discussions go missing, just like the data.
It is additionally worth noting how this case may shed light on how experts are making social sense of bodies in the inclusionary era. In the vote for approval for cisgender men and transgender women who have sex with men, many committee members noted the poor data for transgender women and minority populations but still voted for approval for MSM and transgender women. Here, certain framings of calls for inclusion may inadvertently be ‘doing difference’, by affirming an inherent or biological difference between raced or sexed bodies—the same kind of thinking that has historically led to devastating results. As Epstein’s (2007) work shows, inclusion in biomedical research, as part and parcel of what he calls the ‘inclusion-and-difference paradigm’, can produce sticky understandings of race and gender. Calls for racial inclusion in clinical trials, for instance, assume a biological difference between social categories of race. As one example, the FDA’s 2005 approval of the drug BiDil for African American patients had the effect of essentializing racial difference in pharmaceutical testing and perpetuated the troubling use of race as a biological category (Roberts, 2011). With regard to sex and gender, biomedicine and regulatory authorities have meaningfully wrestled with variation and variability even while often reifying entanglements between sex and gender as well as a male/female binary (Pape, 2021; Zhao et al., 2023). In the case of Descovy, we see the FDA advisory committee doing purposeful and generative work to accept a social understanding of gender while, at the same time, maintaining a biological coherence between cisgender men and transgender women who have sex with men. This is evident in the committee’s approval of indication for both cisgender men and transgender women despite the critical underrepresentation of transgender women in the DISCOVER trial. Moreover, the extrapolation argument advanced by Gilead (i.e., using non-systemic pharmacokinetic data to extrapolate from men to women) may align with a non-essentialist understanding of sex and gender (i.e., a body is a body, and a drug will not discriminate). Future research could more attentively examine expert constructions of social versus essentialist views of bodies at the intersection of inclusionary discourses, including the seemingly contradictory constructions that emerged in this one case.
Examining the shortcomings of narrow epistemological frameworks in HIV clinical trials, STS scholars have foregrounded the interactions between concerns about scientific knowledge production and concerns about social justice (Davis, 2020; Epstein, 1996). We emphasize that myopic clinical trial frameworks preclude matters of ‘data justice’ that are advocated for in the contemporary HIV/AIDS space (Buchbinder et al., 2022; Molldrem & Smith, 2020). With regard to health and medicine, many calls for data justice center on issues of public surveillance or prevention practices, as well as on the systematic exclusions of populations from datasets (Shaw et al., 2024). Our case underscores the social and regulatory desire for equity in data and the pursuit of data justice in a much earlier time period—the biomedical research context that provides the landscape for datasets in the first place. The research frameworks that are regularly deployed in HIV prevention clinical studies—especially insofar as they leave out or ignore entire populations, from those who are not adherent (Davis, 2020) to pregnant people (Sullivan & Lyerly, 2017) to cisgender women—should become a key component of future discussions about HIV data justice, for it is these clinical research frameworks that influence which prevention drugs become available and to whom.
This case also illuminates the discursive strategies that may be deployed—and the implications of doing so—when the temporal dimensions of equity are not considered. In this case, we see Gilead and some advocates adopt a strategy—a discourse of choice—to counter the production of ignorance resulting from the exclusion of women from the clinical trial. STS scholars have theorized about the production of ignorance in medical knowledge-making, showcasing the struggles over ignorance within medical specialties and medical science (Almeling, 2020; Whooley, 2019). Non-knowledge can obviate various ways of thinking about medical problems or causation, and it can be strategically deployed in debates over medical science and pharmaceuticals (Decoteau & Underman, 2015; McGoey, 2012; Whooley & Barker, 2021). We argue that production of non-knowledge can also forestall equity down the line (in terms of drug access), overstate the extent of ‘choice’ in the clinical moment, and further individualize risk. Without safety and efficacy data to make informed choices, how much choice is there? The fallback to choice rhetoric in this case may therefore be more indicative of what other scholars have called ‘agency without choice’ discourses (Mann & Grzanka, 2018). Additionally, when risk is not adequately studied in the clinical trial phase, inclusionary appeals shift risk to individuals and providers as they potentially take and prescribe drugs without knowledge of the drug’s effects in different population groups. Risk is offloaded onto patients and providers—left to be toyed with downstream (see MacKendrick, 2018; also Waggoner & Lyerly, 2022).
Pharmaceutical drug approvals are not trivial or uncontested processes. They are the result of both scientific and social deliberations. Our analysis in this article contributes to research that focuses on moments of ‘failure’ in HIV-prevention research, such as when research protocol or drug indication requests are rejected (see Peterson et al., 2015). Such moments can sometimes reveal the impact of broader political concerns. The advisory committee meetings and public correspondence we analyzed reflect a space that is concerned with the pertinent scientific questions as well as the social ones, where different versions of ethical concerns are played out and debated. The creative persistence of exclusionary justifications in clinical trials should continue to be analyzed by sociologists of science and medicine. Although inclusionary norms are entrenched in 21st-century clinical trials, they remain fluid and contested and are often unrealized in medical science. Future studies could fruitfully use publicly available information, as we have done here with FDA meeting transcripts, to further analyze how drug regulation and social values interface with medical and pharmaceutical research practices.
Going forward, to make inclusion ethical and palatable in contemporary clinical trials, equity has to be considered from different temporal vantage points. Equity must come to the foreground of discussions about inclusionary clinical trials in all phases of drug development, research, and approval. Our analysis suggests that considering equity throughout the clinical research process would facilitate data justice and the goals and values of everyone involved. This case revealed that a lack of inclusion puts stakeholders in impossible situations, as it forces notions of equity (equity in data vs. equity in access) to collide and compete in an era where they should more readily coexist.
Footnotes
Acknowledgements
We are grateful to Norah MacKendrick and Susan Markens for their helpful comments on an earlier version of this paper. We also thank Wen-hua Kuo, Sergio Sismondo, and three anonymous reviewers for their thoughtful feedback and deep engagement with our work.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This material is based upon work supported by the National Science Foundation under award numbers 2409180 and 1945448. The content of this article is solely the responsibility of the authors and does not necessarily represent the official views of the sponsor.
