Multiple Logics: How Staff in Relapse Prevention Interpellate People With Substance Use Problems

This study analyzes how staff in Swedish alcohol and other drug (AoD) treatment interpellate service users as people who can benefit from relapse prevention. Relapse prevention is a widely used intervention. Research is scarce, however, on how relapse prevention is practiced locally and how treatment staff perceive the relationship between AoD use as a problem and relapse prevention as a solution. Drawing on Actor-Network Theory and critical studies of AoD issues within this tradition, we elucidate how staff through specific interpellative logics enact service users, their individual characteristics, and living conditions. The data derive from interviews with 18 professionals working with assessment, counseling, case-management, therapy, and healthcare at AoD treatment agencies in the Stockholm region. The results show that the participants drew on four interpellative logics, and thereby enacted service users as four different object types. Region and network logics pinpointed that individuals have stable observable characteristics that determine their problems and eligibility for treatment (e.g., living conditions, diagnoses). Fluid and fire logics emphasized that their characteristics also vary depending on context and can be present and absent at the same time (e.g., harms, agency). This flexible interpellation of service users echoes the tendency among treatment staff to embrace sometimes irreconcilable understandings of AoD problems and to enact multiple realities of addiction. This suits a professional field where many factors are thought to cause and help resolve problems, but where the treatment supply is often limited to specific interventions. We conclude that it is easier to create a reasonable match between the service delivered and the potential service user if the characteristics of the latter are considered diverse and flickering. This exemplifies Carol Bacchi’s tenet that problem representations are adjusted to fit the solution at hand.


Introduction
This study analyzes how staff in Swedish alcohol and other drug (AoD) treatment make up service users as persons who can benefit from relapse prevention interventions. Carol Bacchi (2017) argues that political objects such as clients and patients are made up in practice, and that they have no inherent, natural, characteristics. Problem representations of this kind are therefore political and strategic, produced to suit the solutions that are at hand . To scrutinize how service users are made up by staff we will turn to Actor-Network Theory (ANT), and more precisely John Law's discussion on how interpellative logics enact objects, subjects and different realities (Law, 2000). We use data from an exploratory research project on the discourse and practice of relapse prevention in Sweden. Relapse prevention is listed as a recommended intervention in the Swedish National Board of Health and Welfare's (NBHW) guidelines (Socialstyrelsen, 2019), which makes it relevant to study how staff enact relapse prevention, how they approach craving and relapse as targets of intervention, and how they interpellate service users. The current study focuses on the latter aspect.
Relapse prevention has become a central intervention in the field of AoD treatment. Given that addiction is often conceptualized as a relapsing disorder, the intervention makes intuitive sense. Although there are now somewhat different models, Marlatt's formulation is still the dominant one (Brandon et al., 2007). Relapse prevention is by its inventor defined as a "self-control programme" for maintaining behavioral changes (Marlatt & George, 1984, p. 261) and a central aim is to help people identify, anticipate and deal effectively with "high-risk situations" through cognitive-behavioral principles. While individuals enrolled in relapse prevention may have their own distinct high-risk situations, relapse prevention offers an overall template for how to approach relapses within treatment. As such, it arguably entails a certain degree of "standardization" (Timmermans & Epstein, 2010), at least regarding the terminology that is used. However, despite its centrality in the field of addiction, there exists surprisingly little critical research on relapse prevention. An exception is Theodoropoulou's (2020) study that challenges the negligence of treatment to target structural and material conditions that affect addiction recovery processes (see also Fomiatti et al., 2019).
The perspectives of professionals on service users are important to scrutinize as they may reveal how policy goals and treatment philosophies are put into practice. Following Hasenfeld (2010), AoD treatment agencies can be defined as human service organizations that try to transform people with substance use problems into people without such problems. In order to define what type of transformation is deemed adequate, service providers at such agencies need to sort, classify and categorize the raw material of people (Hasenfeld, 2010, p. 11). In doing so, they single out individuals with substance use problems as potential targets of intervention. French philosopher Louis Althusser (2010) used the word interpellation to describe how the subject is materialized within the framework of ideology and discourse, and how it, when called upon by an authority "comes into being" (Butler, 1995, p. 6). The staff are engaged in interpellation when discussing problems, framing concerns, and enacting the realities of service users (Mol, 2010, p. 264).
Professionals' interpellation of service users has consequences for how the latter are met in treatment (Karasaki et al., 2013), what duties are assigned to them, and what goals of intervention they are supposed to adhere to . This, in turn, is likely to shape how they understand themselves (Fomiatti et al., 2017). Interpellation processes, then, are far from innocent or of narrow academic interest only. To date, most studies exploring professional views on AoD problems have relied on quantitative approaches and a large share of studies comes from the US (Barnett, Hall, et al., 2018). Given that substance users have been found to be quite differently made up, or interpellated, in different national settings (see Lancaster et al., 2015 for a comparison of Australia and the UK), such studies need to be carried out across various contexts (Seear, 2020). Sweden constitutes an interesting case for exploring these issues further.
In Sweden, social understandings of AoD use have traditionally had a prominent role and the focus of policy has been on protecting the "vulnerable" from harm (Moore et al., 2015, p. 424). However, as in many other Western societies medically oriented descriptions of AoD problems have been more or less prominent during different times, with a clear upsurge during recent years (Edman & Olsson, 2014). The responsibility for AoD problems in Sweden is divided between municipal social services and regional healthcare. This means that the two systems overlap, and that substance users can have as their main service provider either one or both of them. Social services cater to "clients," often with a wide array of problems in addition to substance use, and healthcare focuses on medical aspects and serves "patients" with stable as well as unstable living conditions. This split responsibility between different systems suggests that we may see even more complex understandings of the phenomenon in Sweden compared to what has been shown in other countries (e.g., Fraser, 2016). While an early, quantitative, inquiry shows that medical, social and moral views on addiction are supported simultaneously among service providers in Swedish AoD treatment (Palm, 2004), it does not get into the specifics of how such views are combined in practice. Our study contributes to this field of research by providing an up-to-date analysis of how service users "come into being" through service providers' interpellation.

Critical Research on Addiction
The phenomenon of addiction has interested scholars for a long time and the literature on the topic is enormous. One of the most influential explanatory models is the brain disease model of addiction (BDMA). Within this framework, addiction is conceptualized as "a chronic, relapsing illness, characterized by compulsive drug seeking and use" (Leshner, 1997, p. 45) with parallels drawn to hypertension, asthma and diabetes (McLellan et al., 2000). The core of the condition is taken to be a loss of control over consumption. The addict wants in this view to abstain from substance use, but is incapable due to a "compulsion" (see Volkow et al., 2016). This "acquired disease of the brain" is thought to circumvent people's ability to make volitional choices through a variety of biological mechanisms (Volkow et al., 2016, p. 263). The world's largest funding body of alcohol and drugs research-the US National Institute of Drug Abuse (NIDA)-has promoted the BDMA during the past 25 years and thus strengthened its impact on policy and practice (see Hall et al., 2015). Although the validity of the view of addiction as a chronic relapsing disorder has been challenged when it comes to general population samples with addiction problems (e.g., Cunningham & McCambridge, 2012;de Bruijn et al., 2006), the notion of addiction as a chronic relapsing disorder continues to be central, particularly in clinical settings.
The prominence of the BDMA has increased over the years, following broader trends toward medicalization, but there has also been a parallel growth of research that questions it (see e.g., Hall et al., 2015;Heather, 2017;Heather et al., 2018;Lewis, 2017). For example, the Addiction Theory Network describes its aim as "opposing the dominant influence of the BDMA and collaborating to develop alternative ways of understanding and responding to addiction" (Heather et al., 2018, p. 294). Such critical scholars have maintained that clinical research has not really matched the far reaching promises offered by the BDMA (Hall et al., 2015) and that the model may not necessarily reduce stigma as claimed by its proponents, but rather reinforce it (Heather, 2017, see also Fraser et al., 2017. Another central critique levelled at the notion of addiction-thus not limited to the BDMA-is that the phenomenon is not as easily demarcated as a medical or psychiatric problem as often thought. Keane (2012, p. 366), for example, discusses the transition from DSM-4 to and shows that contrary to the system's intention to single out the disease of addiction, the criteria in DSM produce addiction as a "hybrid medical-ethical entity." According to her, the clustering of heterogeneous symptoms is amplified in the DSM-5, compared to in DSM-4 (Keane, 2012). A key question emerging from this is whether a broad diagnosis such as substance use disorder should be seen as one single entity or rather as multiple ones. Pienaar and colleagues (2017) clearly reject the former conception, viewing it instead as "an emergent, labile phenomenon, multiple in the definitions it is given . . . " (p. 499).
While researchers have debated whether addiction should be defined as a disease or not, and whether it exists as a demarcated entity or not, empirical studies illustrate a much more complex understanding among service providers, policymakers and service users. Service users, for example, seem to hold a rather ambivalent stance toward the BDMA (Meurk et al., 2016). What this illustrates empirically is just how elusive it is to try to capture a coherent, non-contradictory view of addiction among providers and users of services Fraser, 2016;Karasaki et al., 2013). This literature states conclusively that addiction comes in many forms and resembles what Law and Singleton (2005) refer to as "messy objects." To deal with this complexity, many of these studies have drawn upon insights provided by ANT, an approach we follow too.
As pointed out by Suzanne Fraser and others (Fraser, 2016;Fraser et al., 2014, the field of AoD typically encompasses two opposing but coexisting realities. On the one hand, addiction is defined as a demarcated condition (e.g., a disease of the brain) that transforms people into a different state of being, usually characterized by a lack of human agency. On the other hand, people (including addicted persons) are assumed to be self-aware and self-monitoring, and the problems they either experience or do not experience are context-dependent and related to lifestyle. This multiplicity obviously has a bearing on how substance users are interpellated. Research shows that clinical staff oscillate between discourses of disease, recovery and morals in their understanding of service users , and perceive them simultaneously as disordered and in control (Fomiatti et al., 2017), producing inconsistent views on service users' agency (Karasaki et al., 2013).
Below we analyze the multifaceted interpellation of service users that can be expected when treatment staff discuss a problem allegedly characterized by the individual's loss of control (addiction) and a solution that centers on self-control (relapse prevention). Our study shows how distinctly different interpellative logics intersect, enact multiple realities, and contribute to legitimize the supply of interventions that is at hand.

Analytical Approach
As Mol (2010) emphasizes, ANT does not offer a coherent theoretical framework for explaining phenomena, but rather a "repertoire" that facilitates empirical studies. While not claiming to do a full-fledged ANT-analysis, we argue that concepts from this theoretical tradition "sensitize" (Mol, 2010) us to complexity and elusiveness (Law & Urry, 2004). This study adheres to the ANT-principle to make fine-grained analyses of local phenomena, rather than to make sweeping, universal claims as to how something "is." Law (2009, p. 151) argues that everything related to a specific context, for example a market of goods with its buyers, sellers and rules of conduct, assemble and enact a practice that in turn make up a certain reality. Thus, in ANT it is crucial to describe and analyze in detail such practices in which different actors, networks and material as well as immaterial objects are intertwined and together provide boundaries to other practices. As with the broader phenomenon of addiction outlined above, treatments for such problems can be perceived as practices. They materialize in spaces (social service agencies, healthcare clinics, AA-meetings, etc.), and encompass actors (substance users, service providers, policymakers, etc.), networks (diagnostic systems, treatment manuals, evidence-based methods, etc.), and objects (substances, urine samples, detoxification wards, etc.).
Within this theoretical framework, however, objects can include everything from abstract ideas and identities to concrete technologies and bodies (Law & Mol, 2001). In discussing the potential of science to know objects that appear "messy" and complex, Law and Singleton (2005) argue that: "objects come in forms that cannot be known within the most obvious versions of common sense, and that in thinking about this it is useful to work on different models for imagining objects." They further conclude that: "realities, messy or otherwise, are enacted into being . . . such enactments take place in the practices of getting to know those realities" (Law & Singleton, 2005, pp. 334-335, original emphasis).
Following this reasoning, service users, those who manifest substance use problems, can be conceived theoretically as objects that are enacted in practice through service providers "getting to know" them, and interpellating them in research interviews. According to Law, such performances of interpellation distribute subject-positions and object-positions, and are "modes of ordering" reality, that is, "arrangements that recursively perform themselves through different materials-speech, subjectivities, organizations, technical artefacts; and that therefore, since they perform themselves alongside one another, also interact with one another" (Law, 2000, p. 23). For Law (2000), interpellation is thus not a singular process but rather multiple, where different interpellations may clash with each other, potentially leading to "conflicting subject-positions" (p. 24).
Our focus on interpellation as an agent-driven performance with discursive and material effects aligns with previous critical studies asserting that objects such as "drugs," "addiction" and "addicts" lack essential characteristics, and that they are instead performed locally giving rise to multiple and transitory realities (e.g., Fraser et al., 2014;Seear, 2020). According to this, multifaceted interpellations are not reducible to different perspectives on the same phenomenon, but rather indicative of different realities (Law & Urry, 2004, p. 397). In this study, we merge the discussion of performance through interpellation (Law, 2000) with the discussion of how objects can be multiple (Law & Singleton, 2005;Mol & Law, 1994). We use the concept "interpellative logic" (Law, 2000, p. 23) to scrutinize how service providers through interpellation enact service users in different ways; as objects that metaphorically build on regions, networks, fluids, and fire (Law & Mol, 2001;Law & Singleton, 2005;Mol, 1999). In the analysis, we will refer to these four different "models for imagining objects" (Law & Singleton, 2005, p. 335) as representing a "region logic," a "network logic," a "fluid logic" and a "fire logic." The first interpellative logic clusters objects with same or similar characteristics, and provides clear boundaries between dissimilar objects (Mol & Law, 1994, p. 643). For example, differentiating individuals based on place of residence or gender would be an example of this region logic. The second type of interpellative logic encompasses complex relations between actors and other objects in networks. In this case, relations such as shared beliefs and knowledges make it possible to demarcate and compare objects (Mol & Law, 1994, p. 648). This network logic relies on, for example, measurement tools and calculations (such as diagnostic systems). The third logic enacts objects as continuous, not well defined and with "varying shades and colours" (Mol & Law, 1994, p. 660). According to this fluid logic, some objects are simply unstable, having characteristics that flow in and out of sight depending on context, which creates a reality of "open-endedness" (Mol, 1999, p. 83). In her analysis, Mol shows that the reality of anemia differs between sites (Mol, 2010, p. 260). Anemia is a medical condition that can be enacted through the two first logics (it is common in region X, and it can be diagnosed by technology Y), but also as a very gradient object that lacks a clear cutoff point (Mol & Law, 1994, p. 659). Lastly, we will delineate an interpellative logic that is even more elusive, that makes and remakes objects as flickering and unstable, characterized also by the realities that their presence makes absent and othered (Law & Singleton, 2005, p. 342). While this fire logic appears counterintuitive, Law and Singleton (2005) explain how treatment of alcohol liver disease is a "messy" practice that builds on a "series of absent presences" (p. 345). On the one hand, diagnosing and curing the disease is a technical affair that targets the individual as a patient within healthcare. On the other hand, the ways clinical staff approach patients, and the treatment-related decisions they make, can sometimes, but not always, consider the everyday life of drinkers and their prospects to live a sober life. This illustrates how some of the diseased patient's characteristics, for example life situation outside the clinic, can be said to flicker; to be absent-present backdrops. This fourth interpellative logic also draws attention to the interrelatedness of the different logics. If extrapolating the discussion by Law and Urry on scientific methods, to our focus on performances of interpellation, we can assert: "[T]he realities that they produce also overlap and interact with one another" (Law & Urry, 2004, p. 397).

Data and Methods
Our ANT-inspired approach assumes that addiction realities are enacted in local practice. This kind of study thus requires rich and detailed empirical data. To this end, we use in-depth, qualitative interviews with staff in Swedish AoD treatment. While interpellation processes may not only operate through words (Law, 2000, p. 15), analyzing such verbal data should be a reasonable methodological approach. As "stories" can be considered performative (Law, 2000;Law & Singleton, 2000), interviews, together with a focus on interpellative logics, make it possible to elucidate and understand complexity and multiplicity.
The project was approved by the Stockholm Regional Ethics Board (dnr 2018/1064-31/5). We conducted an internet search for AoD treatment agencies that claim to offer relapse prevention. In line with the NBHW's guidelines, it became evident that most agencies that inform the public about their services also identify relapse prevention as a key intervention in solving substance use problems (and other behavioral problems such as gambling). Due to this homogeneity in the supply of treatment on the national level, for convenience we limited the sample to a few agencies near Stockholm (where the research project was based). Through first contacting and informing gatekeepers about the study, the final sample of treatment providers comprised: one municipal social service agency, one outpatient treatment center organizationally linked to the first agency, one local healthcare clinic in the same area, and two specialized clinics for treatment that cater to the whole Stockholm area. The interviews with participants (n ¼ 18) were conducted individually, face-to-face, except for one group interview (two participants). Interviews lasted between 45 and 75 minutes and covered perspectives on service users, on craving and relapse, and on treatments, services and approaches. The sample included 72% females and 28% males from the following staff types: social worker (n ¼ 9), nurse (n ¼ 3), psychologist (n ¼ 4) and addiction specialist (n ¼ 2). The participants were at mid-to late career level in their respective professional positions. Interviews were transcribed verbatim by assistants.
The data were first analyzed inductively which rendered content-based codes such as: service user categories, addiction problems, diagnoses, craving experiences, causes of relapse, treatment processes, relapse prevention in practice and so forth. These codes were then roughly compiled into the more generic categories of problem definitions and solutions. In the present analysis, we concentrated on how the participants described the problems of service users (i.e., the first generic category) and in particular on how they understood and drew boundaries between different individual characteristics, preconditions for lifestyle change and social circumstances. The data were then searched for instances that illustrated the four interpellative logics, outlined above. This deductive coding of data proceeded until the participants' various styles of interpellation of service users appeared to have been covered and made meaningful within this theoretical framework. As will be seen below, references to sociodemographics, such as age and living conditions were coded as illustrating a region logic, and discussions on diagnosable conditions such as addiction and ADHD as a network logic. Descriptions of service users as complex individuals whose experiences and actions are context-dependent (e.g., that addiction can be more or less severe) were coded as illustrating a fluid logic. Finally, statements where a certain aspect of the service user (e.g., agency) was both present and absent in the discussion, were coded as illustrating a fire logic. Extracts that exemplified the four interpellative logics particularly well, and that gave a fair representation of the data as a whole, were chosen for further analysis and presentation. Some participants referred mainly to one or two logics, but the general tendency among participants was to oscillate between logics and adjust the interpellation of service users according to the point that was being made during that specific stage of the interview.

Interpellation of Service Users According to Region and Network Logics: Different, but Addicted
We will first discuss instances where staff described clear-cut differences and similarities between service users. These stories simplified the complexity of dealing with people and referred to "essences" (Fraser, 2016). The participants were prone to differentiate between individuals according to substance use, age and social conditions, but they also distinguished between users with or without motivation for lifestyle change. The service users were in these instances interpellated according to a region logic. There were clear dividing lines between persons who were either qualified or unqualified as service recipients. Typically, this reasoning concentrated on differences between for example socially integrated excessive alcohol users and marginalized opiate users. While some participants claimed to deal with heterogeneous service users, and others with a more homogenous group, they all stressed that region differences exist (though they used other words) and that these play a key role in relapse prevention and the rehabilitation process. This is illustrated by Barbara (nurse at a specialized clinic) who describes that some patients are denied admittance, and that those who are admitted are relatively affluent and share an "unhappiness with their alcohol consumption": We usually explain to people why we started this agency, and that we have rather specific criteria for admittance. You are not allowed to have had contact with traditional addiction services during the last two years. You may not have an ongoing contact with social services. You may smoke and use snuff tobacco, and use alcohol, but no other drugs.
According to other representatives of this organization, restricting the services to individuals with less severe criminal and social problems than typical social service clients also means that persons with psychiatric problems were rejected. The latter were perceived unable to benefit from relapse prevention based on cognitive techniques and homework. While narrow inclusion criteria were typical for this particular organization only, the general message from the participants was that service users can be and are distinguished based on socio-demographics, substance use patterns and basic attitudes. This rigid interpellation of service users was, however, qualified and made more complex when the participants talked about individual characteristics that are not easily identifiable.
In such instances, the participants interpellated the potential service users according to a network logic with a focus on demarcated conditions that played down diversity. Thus, whereas the region logic centered on differences, the second logic emphasized similarities among otherwise different individuals. As concluded in previous research, treatment staff sometimes reference the BDMA strategically to reduce substance use-related stigma and convince service users about the need for treatment (Barnett et al., 2020). Similarly, the participants in our study appeared to rely on fixed diagnostic tools and psychometric scales when emphasizing that people with diverse problems and situations can also be perceived as alike. A case in point was those who highlighted that addiction has similar qualities and developmental stages regardless of substance and type of addictive behavior. While not iterating the BDMA's focus on a diseased brain, Douglas and Betty (social workers) explain that addiction is an emotional disease characterized by avoidance of negative feelings. As an example, they declare that persons who work intensely to save up money for extended periods of leisure and travel may have "two active addictions, at first you escape through work, and the next six months you escape through alcohol". This view on addiction as a condition that people either have or do not have was common in the data. That addicted persons lose control was according to the network logic described as a measurable characteristic that worked as a common denominator among heterogeneous service users. Kimberley (psychologist) describes how persons may both differ and be similar: I believe the largest patient group is persons with alcohol addiction. Then there are quite a few with drug addiction, particularly cannabis, amphetamine, cocaine, and a smaller share with opioid addiction. And this includes those who have been addicted for a very long time and those who are quite new to it. Young and old. And those too, who live a more stable and ordinary life so to speak, and those who have really suffered from their addiction problems. So there is quite some variation at our agency, regarding how burdened our patients are.
This quote highlights the alleged commonality of addiction among the patients that cuts through other individual differences. It also underlines that persons with addiction problems can be interpellated according to both a region logic (emphasizing differences) and a network logic (emphasizing similarities). Juxtaposing similarities and differences, Paul (addiction specialist) describes how a sample of volunteering research participants from the normal population were screened for biomarkers of excessive alcohol use: Their CDT-values [carbohydrate-deficient transferrin, a marker of chronic alcohol use] were as high as in our classic [clinical] group . . . all addiction criteria were equally fulfilled, but they were not harmed as much.
The participants also saw diagnoses such as ADHD, PTSD and depression as very common among addicted people, thus exemplifying that they often had more diagnoses than addiction. They stressed that there are similarities between patients with several diagnoses, but also the importance of identifying what behavior belongs to what diagnosis. When speaking about what problems her patients typically struggle with, Kimberley describes how it is beneficial but often difficult to know if their acts stem from addiction or from other neuropsychiatric conditions: Primarily they have difficulties maintaining sobriety and they lack strategies for understanding why relapses occur. To be able to see early signs. They also have to deal with neuropsychiatric symptoms. The focus is here on identifying and labelling differences between conditions. Kimberley stresses that this endeavor is challenging since different diagnoses catalyze each other and blur the boundaries between them. While this draws on a network logic, presupposing that service users either have or do not have certain diagnoses, it also opens up for an interpellation where individual characteristics and social conditions are intertwined. In the next section we will discuss the participants' take on multiplicity and coincidence, rather than on singularity and order.

Interpellation of Service Users According to Fluid and Fire Logics: Multiple Selves in Relapse Prevention
The data also encompassed a more dynamic interpellation of service users which puts human conditions in context and includes the individual's experience of being interpellated. In the following quote, Karen (psychologist) describes what labels she prefers when interacting with patients.
I believe the patients themselves usually say "substance abuse," but we try not to use that term since it is not part of the diagnostic system. We say "risky use" about alcohol. And then, I usually say "use" when talking about drugs, because I think it is less stigmatizing in some way. And then of course, "addiction" if the person says that he/she is addicted. This illustrates how interpellation can be both influenced by technical aspects such as fidelity to diagnostic systems and by pragmatic ideas about what language is more productive for successful treatment outcomes. Karen rationalizes her abandonment of the term "substance abuse" with reference to changes in diagnostic discourse, while the other terms appear to have been chosen based on more patient-centered concerns. This interpellation of service users draws on a fluid logic. Their characteristics (and if they indicate certain diagnoses) are not assessed and accepted at face value, but incorporated into clinical practice and utilized in accordance with what best serves the purpose Fomiatti et al., 2017;Karasaki et al., 2013).
Moreover, the participants' discussions about prospects of and predicaments in reaching stable sobriety and wellbeing indicated that the service users were interpellated as individuals in complex situations that encompassed several roles and obligations. Dorothy (nurse), for instance, says that she sometimes normalizes the difficulties patients refer to by emphasizing that non-addicted persons also encounter problems. With this, she plays down the special status of addiction, and emphasizes that there is no clear dividing line between addicted and non-addicted persons.
We can have problems too, even though we don't have addiction problems so to speak. One should not forget about who one is in all this. Often people with addiction can only see themselves as having this addiction, and forget about seeing themselves as the persons they are in general, the person one is. That maybe, in fact, one has a job.
The predicaments of life in general are here intertwined with the predicaments of addiction, which yields a more fluid interpellation of service users than the one offered by region or network logics. It acknowledges that addicted persons do not live in a vacuum, but are embedded in social relations and material conditions. Dorothy continues this thought when she speaks about the need in relapse prevention to "always pay some attention to the addiction" when handling high-risk situations in everyday life, such as after work parties. The main point being that different aspects of people's lives are meshed together and that it is therefore difficult for service providers to standardize assessments and interventions. In a similar way, Charlotte (social worker) describes that the problems and needs of individuals in opioid substitution treatment, who use large amounts of alcohol on top of their prescription, are very complex. On the one hand, they avoid problems related to illicit intravenous substance use, and on the other hand they continue to use substances and suffer from harms: Actually, one does not change behavior, but methadone still saves people from criminality, prostitution, disease and so on. I'm not saying that all on methadone do it, but it is quite problematic, and now we accept it without cutting their methadone off. We tolerate much more. They are people with significant other problems, for whom one focuses on reaching a state of "good enough." But I don't think it is that simple. Because it still leads to alcohol-related injuries and early death for these people.
According to this, there are different views on how certain patients' addiction ought to be targeted in service provision. Charlotte questions the interpellation of these service users as impossible to rehabilitate, and highlights instead that their continued alcohol use should be taken seriously and addressed in treatment. Such interpellations approached addiction as merely one out of several aspects of people's lives that are intertwined. The addiction diagnosis, as with other characteristics, was sometimes emphasized and sometimes de-emphasized according to this fluid logic.
We will now turn to the last logic, fire. As has been touched upon above, the participants tended to blend in technical terms (such as the neuropsychiatric symptoms, CDT-values, etc.) in rather mundane accounts of problems. This indicates that service users were perceived as having traits that were simultaneously present and absent. Faye (social worker), for instance, argues that the addicted brain sometimes makes people unable to choose their actions and that relapse prevention can help them understand this and develop cognitive techniques to deal with it.
It is really the addicted brain that makes you relapse. Not always a choice. Of course, it can be a matter of choice too, but this is an addiction that one cannot always control. That is why I believe it is so important to work actively with these parts. Make them and help them see "what is it that makes me relapse on some occasions, and how can I prevent myself from relapsing again?" In Faye's reasoning, there appears to be no conflict in simultaneously stating that addicted persons both lack and have agency. This interpellation of service users relies upon the fire logic that lets the BDMA explain the behavior that interventions target, and it lets the volition of individuals explain why relapse prevention can be effective. The limits of service users' free will are however elucidated by Linda (psychologist), when she mentions how relapses are defined in a non-abstinence based treatment model, called controlled drinking: It really depends on how the individual sees it. ( . . . ) Some people try after a period of sobriety . . . because it is suggested so to speak that you need to reset the reward system of the brain, and what happens is a form of rehabilitation. And in this model, they try to assess whether or not some people can actually handle drinking. I don't work with it myself, but I believe they talk about three [alcohol] units a week. ( . . . ) And in such cases, relapse is really defined with reference to the loss of control and the negative consequences that occur. In this quote, the self-determination of the service user works as a backdrop ("how the individual sees it") which justifies moderation instead of abstention. When the brain has healed sufficiently, it might well be that the person can "handle drinking." But the staff's reliance on individual rationality presupposes that alcohol consumption does not exceed a certain level. The struggle to maintain selfdetermination by keeping consumption levels in check, and the idea that agency is fleeting, is also discussed by Mark (addiction specialist): "After about three units, depending on body weight and so forth, the individual may reach the loss of control stage." In another instance, Mark describes everyday situations in which drinking is expected as potentially relapse provoking "exposures (to alcohol)." According to this fire logic, the service users' ability to act rationally is not constant, but affected by both biochemistry and social circumstances. They are autonomous and subjugated at the same time. When one characteristic leads, the other lurks in the background. James (social worker) elucidates this when he speaks about the lack of low threshold housing for marginalized substance users. He advocates a residential situation where these individuals can stay as long as they do not "drill big holes in the walls and scare neighbors to death." The stereotypical uncontrolled dope fiend is relevant here to the interpellation in its absence. It describes what typical service users are not, but what they can be.

Discussion
Drawing on treatment staff's discussions on relapse prevention, this study adds to prior critical research on how service providers understand substance use problems and solutions. The findings corroborate, but also deepen, the conclusion that an inherent complexity prevails in such understandings. They suggest that Swedish service providers not only position themselves in relation to different discourses (Barnett, Dilkes-Frayne, et al., 2018a) and views on addiction problems (Palm, 2004), but that they also enact different realities (Fraser et al., 2014). Our methodological approach made it possible to scrutinize in detail how complexity and multiplicity are performed. People with substance use problems were interpellated according to several logics, with the end result that they were enacted as several different objects simultaneously. The analyses revealed a broad distinction between rigid and flexible interpellations. The rigidity was evident in the region and network logics (e.g., in discussing the specificities of certain living conditions and diagnoses), whereas the flexibility pertained to the fluid and fire logics (e.g., in discussing how substance use harms and agency are relative and context-dependent). The interpellation of service users according to region and network logics are closely aligned with Hasenfeldt's (2010) tenet that sorting and classifying are key activities in human service organizations. Interpellations based on the fluid and fire logics, however, are not. In fact, it would be impossible to neatly sort and classify service users interpellated according to these logics; they would escape any simple taxonomic system.
In line with previous studies, the data illuminated a dynamic between different, apparently noncongruent, realities of addiction (Fomiatti et al., 2017(Fomiatti et al., , 2019Fraser, 2016;Karasaki et al., 2013). These multiple realities interacted with each other and overlapped (Law & Urry, 2004, p. 397), as exemplified in interpellations of service users as having both fixed and fluid characteristics and problems. Service users' problems were enacted as both specific and not specific to the "disorder" without this being described as contradictory. The question, then, is how to understand this multiplicity.
In accordance with Carol Bacchi's (2017) approach to move backward analytically from the solution (relapse prevention) to the problem (addiction), we may get some clues as to why service users were interpellated in this way. In our interpretation, the use of multiple interpellative logics seemed necessary in order to make sense of relapse prevention. Linking the condition (addiction as a demarcated entity) to the maintenance of a self-determined lifestyle (Fomiatti et al., 2019) requires that the service user is the "site" of both volition and non-volition (Karasaki et al., 2013), being both in and out of control (Fomiatti et al., 2017). In fact, this paradox may be necessary if a relevant match between the solution and the problem is to be obtained. If the addicted person would be interpellated as rational and deliberate only, relapse prevention would not be meaningful-there would be no "chronic relapsing disorder" (Leshner, 1997) to prevent. Similarly, if this person would be interpellated as a slave to the substance only, relapse prevention would be meaningless as the individual would not qualify as a "self-aware and self-monitoring subject" (Fraser et al., 2014, p. 35). The solution, then, in the form of a cognitive-behavioral intervention such as relapse prevention, seems to dictate how the service users are interpellated, and in turn, what kind of object(s) that they can and must be.
In our data, this was most evident when participants drew upon the fire logic. The fire logic allowed treatment staff to leave aside the question of why people develop addiction problems and instead focus on helping them cope with these problems. The service users were, following the fire logic, interpellated as having multiple selves with oppositional interests and desires. They were enacted as objects that both possess and lack volition which served to rationalize relapse prevention. Relying solely on the region, network or fluid logics would obscure the flickering light of agency that the intervention targets.
While participants' seemingly incommensurable interpellative logics could be written off as simply self-contradictory, we believe that they say something important about the elusiveness of addiction. If we move outside demarcated disciplinary silos, there is no scientific consensus on what addiction really is (Hall et al., 2017;West, 2006). Everything from moral models (Heather, 2017), to brain disease models (e.g., Leshner, 1997;Volkow et al., 2016) to notions of "rational addiction" (e.g., Becker & Murphy, 1988;Elster & Skog, 1999) to constructivist ideas about how social problems are made (e.g., Reinarman, 2005;Vrecko, 2010) have been advanced. The scholarly community has not unconditionally accepted any one of them (Seear, 2020). On a broader level, discussions about addiction also raise philosophical questions about the role of volition in human behavior. That service providers grapple with what addiction is should come as no surprise.
Besides adding to a growing body of ANT-inspired research on AoD issues that elucidate the complexity of interpellation in practice, this study makes a specific contribution in relation to relapse prevention. There is surprisingly little critical research on the topic. Trying to reduce this gap, we have explored how service users are interpellated in the context of relapse prevention in Sweden, where the responsibility for AoD treatment is split between social services and healthcare. The sample covered participants with divergent academic training and occupations. While representatives of different disciplines could be expected to share beliefs and opinions that reflect their respective occupational roles, this was generally not the case here. In fact, corresponding with Palm's early study (2004), no single participant interpellated service users in a way that consistently drew on a specific understanding of addiction, be this social (Moore et al., 2015) or medical (Edman & Olsson, 2014).
Thus, in our study, different occupational roles hardly produced their own realities; rather, the multiple realities of addiction were enacted at the level of the individual, whether one was a nurse, a psychologist, an addiction specialist or a social worker. At least in part, this convergence on multiplicity may be due to the centrality of both social services and healthcare in the Swedish treatment system, and the collaboration between the two. In no instances were we informed that there were any substantial conceptual or theoretical barriers between different professions. Relapse prevention is a key feature of Swedish AoD treatment (tangible and relevant for both providers and users of services), and this may work as a glue that binds multiple addiction realities together. The participants unanimously drew on different interpellative logics, and most notably on the fire logic, to make up service users as persons who can benefit from the intervention.

Conclusion
What we can conclude from our interviews with the service providers is that there was no single prevailing understanding of addiction and addicted people. Rather than one discourse, one interpellative logic or one reality, there were multitudes of them that intersected in complex ways. Participants drew upon this multitude when making up service users, with the end result that they were enacted as many different objects at once. We have argued that the flexible interpellation of service users found in this study may be a product of the intervention under study . Its strong focus on cognitive-behavioral techniques, together with its far-reaching ambition to foster lifestyle change, warrants an interpellation of service users that is rigid and flexible at the same time. To paraphrase a well-known expression: One size fits all, if they all come in all possible sizes. There is, however, a downside to this optimistic view on what relapse prevention can achieve. Our results suggest that it presupposes the enactment of service users as having some degree of self-control over their behavior. This raises the question whether the intervention may reinforce exclusion processes operating on people with substance use problems. The rather individualistic focus of relapse prevention (Theodoropoulou, 2020) may better suit service users in more advantageous situations than those who are more marginalized. At the same time, rigid interpellations of service users as suffering from a pathological condition outside of their own control (e.g., a brain disease) may not align with the realities of those with substance use problems who self-identify as rational, capable and relatively well-off. While these paradoxical issues are outside the scope of this study, they should be seriously considered by actors involved in relapse prevention.