Understanding how PrEP is made successful: Implementation science needs an evidence-making approach

ABSTRACT After a decade of oral HIV pre-exposure prophylaxis (PrEP), the next generation of PrEP is being anticipated, including long-acting pills, injections, and implants. The unevenness of international PrEP implementation is increasingly recognised, with successful rollout in some settings and failure in others. There is a need to better understand conditions of PrEP implementation, and its localised (and sometimes unanticipated) effects. Implementation science explores how contexts and health systems shape the successful translation of health interventions. In this essay, we consider how PrEP is evolving and argue for an ‘evidence-making’ approach in relation to evidence and intervention translations. This approach emphasises how both interventions and their implementation contexts are co-constituted and evolve together. Unsettling the assumed universality of an intervention’s effects and potential in relation to its implementation contexts helps to harness the localised possibilities for what PrEP might become. As the next generation of PrEP offers renewed promise, we must explore how PrEP is put to use and made to work in relation to its evolving situations. We urge implementation science to consider implementation processes as ‘evidence-making events’ in which evidence, intervention and context evolve together.


Introduction
HIV pre-exposure prophylaxis (PrEP) involves the use of antiretroviral drugs to prevent new HIV infections.The first PrEP agent was approved in 2012, an oral combination of emtricitabine/tenofovir disoproxil fumarate, which held promise for ending HIV as a public health threat (Auerbach & Hoppe, 2015;Thomann, 2018).A decade later, this promise has been unevenly met, particularly in the United States where racialised and economic conditions structuring healthcare account for massive disparities in PrEP uptake (Bekker et al., 2022;Blackstock, 2021;Philbin & Perez-Brumer, 2022).Realising the promise of PrEP requires more than simply 'getting drugs into bodies' (Auerbach & Hoppe, 2015), and as rollout surges forward in some parts of the world and falters in others, social research is vital for understanding the conditions of successful PrEP implementation, and its localised effects.
In this commentary, we consider how PrEP is evolving.We emphasise how PrEP translates in multiple ways according to its local implementations and implementation contexts.A key focus of implementation science in the field of HIV research is to study how contexts shape the translation and implementation of evidence-based interventions (Cáceres et al., 2015;Geng et al., 2022;May & Finch, 2009;Peters et al., 2013;Shangani et al., 2021).We draw attention to some limits in how implementation science is commonly deployed.Taking PrEP as our example, we argue for an 'evidence-making intervention' approach to evidence and intervention translations (Rhodes & Lancaster, 2019).While implementation science proposes frameworks to examine how 'evidence-based' interventions (like PrEP) are implemented in different contexts (May et al., 2016), we propose a more relationally situated implementation science.Rather than taking an intervention (for instance, 'PrEP') as a pre-given, 'evidence-based' object, with universal effect potential, an evidence-making intervention approach attends to how the processes and practices of implementation work to (re)make both 'intervention' and 'context', as matters of 'becoming'.We call for research which attends to how PrEP is 'put to use' and 'made to work' in its situation of implementation (Rhodes & Lancaster, 2019;Smith et al., 2023), thus extending approaches to implementation science to better understand the varied relational conditions that make what PrEP becomes.Crucially, an evidence-making approach to implementation science neither holds 'intervention' nor 'context' as stable or separate from the other, but seeks to notice how PrEP evolves with its implementations.The focus then, becomes the 'implementation event' as an 'evidence-making' momenta moment in which evidence, intervention and context become together.

Implementation science and PrEP
Implementation science has captured the attention of public health and HIV scholarship (Cáceres et al., 2015;Geng et al., 2022;Lobb & Colditz, 2013).It is a growing discipline with varied approaches to conceptualising how interventions are implemented within the messy social settings and contexts of healthcare, with the goal of understanding how interventions may be implemented successfully.Implementation science moves beyond determining the effectiveness of interventions in controlled experiments and trials to evaluating how interventions are translated into 'real-world' settings.For example, implementation scientists may approach evaluating 'real-world' implementation through studying the acceptability, adoption, appropriateness, feasibility, fidelity, implementation cost, penetration, and sustainability of an intervention (Proctor et al., 2011).Prominent models and frameworks for implementation science also include the 'RE-AIM model' (Glasgow et al., 1999) and the 'PRISM model' (Feldstein & Glasgow, 2008).In doing so, implementation science frames public health interventions as complex and recursive adaptations in social systems (May & Finch, 2009;Peters et al., 2013).However, while 'context' is recognised as important in implementation science, context is operationalised as a static and stable backdrop 'against which' an intervention is translated into, rendering context as a problem for the successful implementation of an intervention.
In the field of PrEP, investments in implementation science have been characterised as vital for maximising PrEP's effectiveness, including to track progress in PrEP implementation and to minimise what are framed as social, political, and logistical barriers that may impede successful rollout (Cáceres et al., 2015;Geng et al., 2022).While there is optimism regarding the possibilities of implementation science in HIV, there are divergences in fidelity to implementation science frameworks.For example, it appears that PrEP implementation science research in the US is often conducted without a clear implementation science theoretical framework, and that most studies have focused on the acceptability and feasibility of PrEP, with few studies assessing other evaluation criteria (Shangani et al., 2021).In contrast, Geng et al. (2022) advocate for a broad agenda to study the implementation of HIV interventions, putting emphasis on embedding implementation science throughout the process of implementation and adapting implementation to the contexts of interventions rather than prescribing a fixed a priori implementation science methodology.We agree that adaptation is key, but suggest that what is needed is a less mechanistic approach to implementation which attends to the transformations of PrEP in its evolving situations (Rhodes & Lancaster, 2019).In the following section, we frame what an evidence-making intervention approach offers.

Implementation science and evidence-making interventions
Although there are a diversity of approaches in implementation science, they are unified by treating interventions as 'separate from, yet shaped by, their implementations; as pre-existing but travelling objects made prior to their situated implementations' (Rhodes & Lancaster, 2019, p. 3).We question whether implementation science approaches go far enough in appreciating how both interventions and their contexts can be seen as relational effects of their entangled implementations (Rhodes & Lancaster, 2019).Implementation science needs to attend not only to how interventions are enacted differently, and multiply, in their implementation practices, but that it is the situation of implementation, as much as the 'intervention' in 'translation', that constitutes success or failure.Put another way, public health interventions are never separate from their implementationas if pre-existing, evidence-based and ready for transferbut rather are always 'made', and re-made, as an effect of their situation (Rhodes & Lancaster, 2019).This is the crux of an evidence-making intervention approach.While implementation science focuses on evaluating how an intervention object is adapted into a system, an evidence-making intervention approach attends to how interventions and their situations of implementation are co-constituted and made in practice.This means that interventions are transformed, becoming something different in their local settings and sites of practice.This does not reduce the study of interventions and their effects to a 'context trap' (Geng et al., 2022) but rather hones attention to precisely how things evolve, adapt, and are made multiple and different.Crucially then, both interventions and context are fluid and emergent in an evidence-making intervention approach (Rhodes & Lancaster, 2019), whereas most implementation science, including as deployed in the HIV field, tends to presume a relatively stable context separate to adapted intervention.
If we are to fully appreciate the broad social and political contexts and logistical issues which situate interventionswhich is the promise of implementation science (May et al., 2016) we must treat PrEP as always 'lively' and 'becoming' with its situation, and not as simply 'evidencebased' with assumed effects and potential.For example, debates over the expenditure of public funds on subsidising PrEP in some settings such as England has required contending with a framing of (queer) health as a discretionary 'lifestyle choice' amidst broader health and social welfare austerity measures imposed by governments (Maine, 2020;Nagington & Sandset, 2020;Paparini, 2021).While PrEP was thoroughly trialled and its potential well evidenced in England, the situation of implementation has required tinkering around the system to provision PrEP, including 'DIY' (self-imported) PrEP and administering implementation trials as access points, resulting in precarious access to PrEP and intensive efforts amongst advocates to ensure it is provisioned (Nagington & Sandset, 2020;Paparini et al., 2018).
An evidence-making intervention approach helps us to see and harness the possibilities for what PrEP becomes, enabling implementation science to trace how the thing we call 'PrEP' is variously put to use and made to work in relation to its evolving situations.This accentuates the 'implementation event' as the site of study to trace how evidence, intervention and context come together locally (Race, 2016;Rosengarten & Michael, 2009).In the following sections we elaborate on these approaches and consider how PrEP is being put to use and made to work in different ways.

Expanding PrEP or multiplying PrEP?
As PrEP continues to be introduced in different countries and populations, the focus of implementation has expanded from engaging people and providers to issues related to adherence, (dis)continuation and 'persistence' with PrEP use (Celum et al., 2021;Coy et al., 2019;Laborde et al., 2020;Pillay et al., 2020;Scott et al., 2019).Making PrEP work in the many settings in which it is being implemented is far from straightforward and arguably the complexities of implementing PrEP are multiplying.The concept of 'PrEP' has expanded beyond a daily or on-demand ('2-1-1') dosing regimen (World Health Organization, 2019), with a pipeline of current and future drug agents and different ways of taking or administering PrEP, including long-acting pills, injections, and implants (Beymer et al., 2019;Philbin & Perez-Brumer, 2022).These 'second-generation PrEP agents' are positioned as ways to expand choice in the ways that people protect themselves from HIV, but it is as yet unknown how they will be received, taken up, or made to work in different settings (Bekker et al., 2022;Marcus & Krakower, 2022;Philbin & Perez-Brumer, 2022).How PrEP is offered and used has also clearly been disrupted by COVID-19, with the pandemic unsettling people's lives, sexual practices, engagement with health services, and HIV prevention efforts, and encouraging many users to pause, stop or refashion the way they take PrEP (Callander et al., 2022;Hammoud et al., 2021;Matambanadzo et al., 2021;Murphy, 2021;Torres et al., 2021).
These new modalities and disruptions provide insight into what PrEP is and what it might become, changing the way it is accessed, the embodied experience of taking the drug, engagement with clinical services, and how it is understood to 'work' (Smith et al., 2023).While the possibilities for antiretroviral-based HIV prevention are expanding, we caution against a tendency to think of interventions, like 'PrEP', as if they were singular and stable with predictable and universal effect potential across different contexts and situations (Holt, 2021;Rosengarten & Michael, 2009).As interventions of promise become seemingly more 'evidence-based', they can be presented as if solid, stable and immutable (Latour, 2005) that is objects that are universal and unchanged across time and space.Interventions may also be seen as 'fixed' elements of standardised packages coordinated for scale-up (Lampland & Star, 2009), that is, PrEP is presented as a 'package' with roughly predictable elements that can be translated across contexts.However, this misses the complexities and contingencies through which interventions are transformed in practice, and the unexpected effects they make (Lazarus et al., 2021;Smith et al., 2023).As PrEP intervening expands in scaleacross contexts and situationsit also multiplies in meaning and effect.We argue, therefore, that rather than being materialised as a singular or stable thing; that PrEP is situated, fluid and mutable, even if presented in evidence-based approaches as if it were singular (Mol, 2002).
PrEP is multiple (Mol, 2002); not merely in the sense of the plurality of antiretroviral drugs used as PrEP, the variety of dosing strategies, and current and future modes of administration but in the sense that it is understood, practised and experienced in varied ways by different users, in different settings, and in changes over time (Holt, 2021).Take for example, daily oral PrEP, the most common way that PrEP is globally understood and used.After a decade of implementation, it is tempting to consider daily oral PrEP as a 'stable' object, but how is daily oral PrEP (re)shaped through the rollout of new dosing strategies and modalities of PrEP, and how is it differentially enacted across settings and amongst different populations?Treating a technology, such as PrEP, as emergent and unfolding in its situation imagines it at once as a 'singular-multiple' (Mol, 2002;Rosengarten & Michael, 2009); that is, different versions of prophylaxis do their work differently, but nonetheless they hold fidelity to a unifying category of 'pre-exposure prophylaxis' for HIV infection.Through the next sections, we consider how PrEP is made multiple in different ways, including (1) in time, (2) through its varied implementation practices, and (3) taking account of both time and implementation, in local ecologies.

PrEP in time
Over time, PrEP has been made into multiple sociomaterial 'things', including as a pill regimen, a successful or absent HIV prevention strategy, an object of research, a policy instrument, and a cultural icon.Early PrEP trial scientists conceptualised PrEP as both singular and multiple (Rosengarten & Michael, 2009); as a singular product when confident about what PrEP was and what it could achieve, and as an unfolding process when anticipating what was unknown about how PrEP might be implemented in practice.These unknowns about the rollout of PrEP were reflected in ambiguities about who PrEP was perceived to be suitable for and populations to be prioritised.Early published research, clinical guidelines, and policies configured a variety of (sometimes contradictory) ideal PrEP users (Holt, 2015).While the early introduction of PrEP was met with enthusiasm from some about its potential to prevent HIV, there was uncertainty about whether PrEP would disrupt, complement, or replace existing HIV prevention strategies (particularly condoms), or even work in practice (Auerbach & Hoppe, 2015;Young et al., 2021).
Over the last decade, PrEP has been remade and is reshaping the landscape of HIV prevention, moving from a technology of unknown promise, imagined for particular risky bodies (Holt, 2015;Race, 2016), to a celebrated 'game-changing' technology in global HIV prevention strategy (Thomann, 2018;World Health Organization, 2019).However, actualising the promise of this technology remains complex.In the current era of PrEP, a key challenge is how to move from a sense of generalised 'evidence-based' prospect into an understanding of how this intervention is made to work in local sites and implementation practices.This is also a key concern for implementation science.However, in implementation science what is often left unexamined is how interventions are transformed through the process of implementation.That is, what PrEP is changes in time.While PrEP was initially enacted as an object of unknown promise, as described above, it is now more commonly performed as an evidenced intervention and a foregone conclusion with universal effect potential (Bekker et al., 2022;Marcus & Krakower, 2022;Smith et al., 2022b).That is, the implementation question has shifted from 'how do we encourage people to consider using PrEP' to 'why are these groups of people not using PrEP?'.This is a different enactment of PrEP, and a major shift in narrative: the initial rollout of PrEP configured it as an unproven strategy that was difficult to persuade people to use, but this has been cast aside and PrEP reconfigured as always having been a straightforward and effective intervention.This may make it difficult for doubts and complexity about PrEP to be voiced in its varied implementations (Philbin & Perez-Brumer, 2022).
Time also entangles with PrEP in other important ways.The field is simultaneously invested in encouraging key populations to start using PrEP, but also with users 'persisting' with it and documenting reasons for discontinuation (Celum et al., 2021;Coy et al., 2019;Laborde et al., 2020;Pillay et al., 2020;Scott et al., 2019).HIV prevention requires sustaining the conditions that support people to use PrEP amidst the variabilities of people's lives.The dosing of PrEP also involves different temporal elements.PrEP dosing is potentially mundane and quotidian if it can be enfolded into domestic routines, or it may require a high degree of planning and experimentation with bodily routines and life circumstances to manage side effects and achieve optimal protection (Smith et al., 2023).These aspects of timing are particularly evident with multiple oral dosing options, in which PrEP pills are taken according to anticipated sexual risk.Long-acting forms of PrEP will extend the timing of PrEP and HIV prevention in new ways, prompting different relationships with clinic visits and with anticipated sexual risk.These expanding, differing forms of PrEP present a problem for the field: how to attend to this complexity (timing, pills taken, sexual risk events) in methods that aim to measure 'prevention-effective adherence' (Bavinton et al., 2021;Haberer et al., 2015)?While the field attempts to treat it as a stable object fit for evaluation, PrEP's enactments are multiplying over time.

PrEP varies in local implementation
Experiences of PrEP and its effects vary by the way PrEP is implemented, including in different places.Implementation scientists regularly consider the challenge of 'context' in making interventions like PrEP successfully work (Cáceres et al., 2015;Geng et al., 2022), but there is a tendency to see both interventions and contexts as stable (even if complex) rather than coconstitutive and thus evolving.Instead, through the lens of an evidence-making intervention approach, we can attend to how PrEP is itself shaped through implementation, and that the practices of implementation shape the setting.For example, in Australia PrEP was first implemented as a daily dosing strategy (Grulich et al., 2021;Murphy, 2021).The initial emphasis on daily dosing meant that some clinicians in Australia were hesitant about recommending on-demand PrEP even when international and national guidelines endorsed this method (Smith et al., 2021).In contrast, there has been greater fluidity in PrEP dosing strategies in the Netherlands, France, and Quebec, Canada, where the option of either daily or on-demand PrEP was explicitly encouraged (Molina et al., 2017;van Hoek et al., 2021).Similarly, the US experience of PrEP has contended with a framing of users as 'Truvada whores' and related stigma (Calabrese et al., 2019;Spieldenner, 2016), while Australia has largely avoided a preoccupation with this discourse (Haire et al., 2021).In these various examples, we can see how what PrEP can become depends on practices that include or relate to a hinterland of other knowledge practices (Law, 2009); different PrEPs are being done in different places.
Attending to how PrEP is made differently in some settings is vital for understanding why it may or may not work 'successfully'.For example, as Lazarus and colleagues (2021) have noted in an ethnographic study of PrEP in Southern India, global policy and data on PrEP's 'effectiveness' can fail to account for the divergent ways in which PrEP entangles with bodies, life histories, social worlds, regimes of care, and engagements in health promotion.In their study, some women reported unexpected and dramatic side effects from taking PrEP, including dizziness and vomiting, while other described PrEP curing them of bleeding, discharge, and pain, or PrEP leading to increased appetite, energy, and vitality (Lazarus et al., 2021).While these varied claims about PrEP's effects sit uneasily with biomedical framings of PrEP's side effects, anthropologists have long pointed out that pharmaceuticals are embodied and produce different experiences across culture and place (Hardon & Sanabria, 2017;Lock & Nguyen, 2018;Whyte et al., 2002).As Lazarus and colleagues (2021) observed, the implementation of PrEP enrolled women into a different relationship with their bodies, including forms of healthcare and monitoring they could not otherwise afford.This produced very different local experiences of what PrEP was and the effects it had (Lock & Nguyen, 2018).Attending to local entanglements and contingencies has important implications for understanding how PrEP works (or not) in local communities, with effects which extend beyond simply taking a (daily) pill.Attending to PrEP as a thing in-the-making and as a contingent matter of local practice provides a way of understanding and optimising its implementations.
Implementation of PrEP differs greatly between populations and communities in different settings.In multiple settings many individuals have commenced PrEP but then discontinued after a year (Celum et al., 2021;Coy et al., 2019;Pillay et al., 2020;Scott et al., 2019).Higher rates of discontinuation have been documented in some populations in the USA, including transgender women, Black people, people who inject drugs, and people who lack comprehensive health insurance (Coy et al., 2019;Scott et al., 2019).Sex workers and gay and bisexual men in South Africa have identified side effects and feeling stigmatised for using PrEP as common reasons for stopping use (Pillay et al., 2020).In contrast, cities such as London and Sydney have successfully enrolled large numbers of gay and bisexual men to use PrEP, resulting in declines in HIV infections (Grulich et al., 2021;Nwokolo et al., 2017).These 'successful' settings are reflective of the way economically and socially privileged communities were engaged by clinicians, health promoters, and policymakers to take up PrEP, but these efforts were also met by communities who drove (and often led) demand for access to PrEP, actively adapting local community and sexual cultures to make PrEP work (Holt, 2021;Murphy, 2021;Paparini et al., 2018).These are not simply 'ingredients' that are manipulated to make a universal concept of PrEP 'work' in local conditions, but rather reflect how PrEP as an intervention is made to work in practice.The success of PrEP requires resourcing and sustained efforts, but also local adaptation of the concept into people's everyday lives and communities (Brown & Di Feliciantonio, 2022;Martinez-Lacabe, 2019;Orne & Gall, 2019;Smith et al., 2022a).It is here that we can notice how the 'context' in which PrEP is being implemented is not held fixed and stable, but is rather fluid and being constantly remade in relation to what PrEP is becoming in its situation.We need an implementation science which seeks to notice this emergence; attending to how not only interventions but also contexts and systems adapt and transform in time and in practices.

Local ecologies of PrEP
As discussed above, PrEP changes in time and varies across settings, and these transformations are always in the making.However, the conditions that make PrEP work are not a monolithic or static 'context'.The concept of 'ecologies' is useful for understanding the processual and recursive transformations of interventions like PrEP, in which 'environments' or 'social context' are continually emergent as complex and adaptable assemblages (Rhodes et al., 2021).Rather than approaching 'individuals' as being shaped by their 'contexts', a focus on ecology enables a focus on the ways in which humans and nonhumans are afforded similar attention in analysing situated effects of interventions (Rhodes et al., 2021).For example, the COVID-19 pandemic has challenged the work and progress of HIV prevention across the globe, albeit in unexpected and difficult to measure ways (Hammoud et al., 2021;Matambanadzo et al., 2021).However, framing a singular and stable backdrop of 'COVID-19' as a new context omits the ways in which this new pandemic continually unfolds and is entangled in everyday life.
Lockdown measures and changing sexual cultures in response to COVID-19 have produced conditions in which primarily daily PrEP users have re-evaluated their use of PrEP (Gaspar et al., 2022;Murphy, 2021), pausing or stopping use or experimenting with different dosing methods (Hammoud et al., 2021;Matambanadzo et al., 2021).There can be unexpected and unanticipated effects through the extending chains of relations constituted by COVID-19.For example, Matambanadzo et al. ( 2021) documented how PrEP use amongst female sex workers in Zimbabwe increased in 2020.They noted that the ability to procure clients had reduced due to disruptions in social life, leading to increased competition between sex workers for fewer clients, and reduced ability to negotiate condom use.Consequently, female sex workers were more interested in PrEP use, and service delivery adapted to support these workers (Matambanadzo et al., 2021).This example demonstrates how an extending chain of relations produces a different, unpredictable ecology in which using PrEP is made relevant and vital.
PrEP involves not just the taking of drugs, but also involves embodying ideas about community belonging, personal identity, shifting ideas about what constitutes safe sex, and anticipations about risk and future sexual practices (García-Iglesias, 2022;Girard et al., 2019;Haire et al., 2021;Hughes et al., 2018;Martinez-Lacabe, 2019;Møller & Ledin, 2021;Race, 2016;Smith et al., 2022b).If PrEP use constitutes an identity or sense of belonging, what does that mean when the necessity of taking pills is disrupted, and how might PrEP use be made differently afterwards?Elements of the 'assemblage' of taking PrEP may remain in place (Holt, 2021), and so people may continue to identify as PrEP users, even when pills remain 'on the shelf' and are not being consumed.COVID-19 has in some settings sufficiently disrupted PrEP use in ways that echo problems with older forms of HIV prevention, like condoms.For example, people may see themselves and tell others they are PrEP users (like how people identified as condom users), while not consistently using the intervention.Given COVID-19 has produced different types of 'PrEP use' in different locations and populations, how might other (re)emerging health threats further disrupt and alter what PrEP use is and how it is locally practised, with various and perhaps unanticipated effects?

Conclusions
We suggest that an 'evidence-making intervention' approach offers useful insights for attending to new and ongoing developments with PrEP (Holt, 2021;Holt et al., 2019;Rhodes & Lancaster, 2019;Smith et al., 2023).An 'evidence-making intervention' approach further extends the concerns of implementation science, which focuses on the complexity of interventions in translation (May & Finch, 2009;May et al., 2016;Peters et al., 2013), by seeing both intervention and context as becoming together, rather than conceptualising public health interventions, like PrEP, as stable evidence-based objects with universal effects that are implemented 'into' different fixed sites.An evidence-making intervention approach treats 'evidence, interventions and their effects as emergent, contingent, and multiple' (Rhodes & Lancaster, 2019), and pays attention to the processes and practices of implementation, including how interventions and their contexts are constantly re-made through networks of relations.We are suggesting an approach at this moment of HIV prevention which attends to how PrEP changes contexts, as contexts situate PrEP.In this way, attending to ecologiesthe environments comprising pharmaceuticals, people, places, and conditionsis key for situating how interventions are enacted.Implementation is not just an extension of a 'universal object', but rather serves to generate an intervention and context with its own social life and evidence.
Echoing that social scientistsand indeed, all who are invested in HIV preventionshould attend to more than 'getting drugs into bodies' (Auerbach & Hoppe, 2015), we must consider how interventions like PrEP are continually made through, and re-make, social, cultural, economic, and political conditions (Bernays et al., 2021).An implementation science is needed which focuses on how intervention 'success' is made in the implementation event, itself an effect of local ecology.As we attend to 'new generations' of PrEP, we must continue to ask: what is PrEP, and how is it being put to use and made to work in its implementations?

Disclosure statement
No potential conflict of interest was reported by the author(s).