Eric (a pseudonym) is a low-level heroin dealer. He buys the drug in batches from other suppliers higher up the illicit market’s proverbial food chain and then repackages that supply to sell to his customers. One day, not long ago, Eric came home to see a news story on a local television station about an overdose victim found and resuscitated in a public parking lot nearby. The newscast reported that fentanyl, a powerful synthetic opioid, was present in the drugs this individual had used and was likely the cause of their overdose. Eric was shocked to recognize the overdose victim on the news as one of his own clients. This is how Eric learned that the heroin he was currently selling was adulterated with fentanyl.

Our research team heard this story by way of one of Eric’s regular clients, a middle-aged white woman named Sandy (also a pseudonym). She spoke in detail about the conversation she had with Eric when he called her to warn her about what he saw on the news. He recommended she throw away the drugs she had just bought from him in light of this newly discovered fentanyl contamination. Allegedly, Eric even offered to replace the drugs they threw out with a new batch of (ostensibly fentanyl-free) heroin at a discounted price. “I know it sounds crazy to you guys,” Sandy said in our interview, “but some of them [dealers] actually care if they serve you that [fentanyl]. They don’t want you dead. They need you for that money. And they don’t want you dead. So, there are some dealers that actually care.”

The axiom that “epidemics are fundamentally social processes” (Maher, 2002) has been foundational to social medicine and, later, critical medical anthropology since this view was broadly promoted by physician-anthropologist Rudolph Virchow in the 19th century. “Medicine,” he wrote in his 1848 administrative report on an infectious disease outbreak in the desperately oppressed region of Upper Silesia, “has imperceptibly led us into the social field and placed us in a position of confronting the great issues of our time.” (Virchow, 2006). Following this insight, research into the social-epidemiological dynamics of opioid overdose throughout the world have paid close attention to structural and individual-behavioral drivers of negative health outcomes (Butt et al., 2017; Culbert et al., 2016; Fornili, 2018; Gilbert et al., 2013; Heimer, 2018; Perlman & Jordan, 2018). Further, the ability of social relationships and social network dynamics to shape health outcomes on a population-level has been well established by recent research on sexually-transmitted HIV (Brennan et al., 2012; Teixeira da Silva et al., 2019; Valenzuela-Jiménez, Manrique-Hernández & Idrovo, 2017). Research exploring the impact of social and relational factors on overdose, however, remains sparser.

In the past few years, a growing number of mixed-methods research studies have successfully contextualized different features of the U.S.’ current opioid overdose epidemic within the social environments that engendered them. Many have explored individual or micro-level factors, including risk-reduction strategies developed by people who use drugs to cope with the falling availability of prescription opioids in the drug market (Mars, Bourgois, Karandinos, Montero & Ciccarone, 2014) and the potential harms of unintentional fentanyl-exposure (Carroll, Marshall, Rich & Green, 2017; Rhodes et al., 2019). Recent studies have also investigated the structural, or macro-level, factors that shape the overdose risk environment, demonstrating how the risk of fentanyl exposure may vary according to the kind of heroin products—or products presented to consumers as heroin or as heroin-like (Ciccarone, Ondocsin & Mars, 2017)—that are locally available, the marketing methods used to sell those products, and the amount of product information made available to consumers through local market structures (Mars et al., 2015, 2016).

Little research on the current, fentanyl-fueled opioid overdose epidemic, however, has explored in-depth the role of socio-relational factors, such as the nature of the relationship between Eric and Sandy and the sense of mutual obligation they feel towards each other, in shaping the health outcomes of people who use drugs. Early substance use research in the U.S. has posed these vital questions about the social terrain of substance use and its related harms that remain highly relevant today (Bourgois, Prince & Moss, 2004; Singer, Valentín, Baer & Jia, 1992; Spradley, 1968). To the best of our knowledge, however, only two studies conducted during the current epidemic have explicitly analyzed the role of socio-relational factors in shaping health outcomes. The most recent of these, which was conducted among a predominantly African-American cohort of people who use drugs in urban North Carolina, produced two important findings: (1) relying on trusted or familiar dealers was as a commonly reported fentanyl-avoidance strategy used by individuals who preferred not to consume fentanyl-adulterated drugs and (2) participants reported encountering fentanyl-adulterated heroin—despite not seeking fentanyl—when they were unable to purchase drugs from dealers whom they knew and trusted (Rhodes et al., 2019). Additionally, an earlier study conducted among a predominantly white cohort of people who use drugs in Providence, Rhode Island, found that some participants described their overdose risk as directly increased or directly decreased as a result of actions allegedly taken by their dealers, suggesting that the nature of consumer-supplier relationships may differentially impact overdose risk in the illicit drug market (Carroll et al., 2017).

The purpose of this study is to elaborate and build upon these previously generated hypotheses. We aim to explore the social and relational factors that shape the current opioid overdose epidemic within a population of people who use drugs in Providence, Rhode Island. Specifically, we aim to describe the interpersonal relationships between people who use drugs and the individuals who act as their drug suppliers (whether regularly or irregularly) in order to consider how these social ties and the culturally-reinforced mutual obligations between them shape overdose risk. Put another way, this study asks how we can make sense of people like Eric, the heroin dealer who warned his clients about fentanyl contamination in his drugs, and the potential public health impact of the relationship he has built with Sandy by considering these supplier-consumer relationships through an ethnographic lens. Further, how might national-, state-, and community-level responses to today’s opioid-overdose epidemic be improved if we are able to trace the role that he and others like him have in shaping patterns of overdose among participants in the illicit drug market?

The illicit opioid market in the U.S. has undergone several transformations in recent decades. The history of current trends ostensibly begins in the early 1990s, when Colombian-sourced heroin—remarkably cheaper and purer than heroin originating elsewhere—began appearing in the U.S. drug market (Ciccarone, Unick & Kraus, 2009). Closely on the heels of the growing market share held by Colombian-sourced heroin came another significant trend: a massive growth in opioid prescribing in the U.S. health care system, including an increase in the retail sales of oxycodone (brand names include Percocet, Percodan, and OxyContin) of nearly 600% between 1997 and 2005 (Manchikanti, 2007). Many people who used opioids at that time did not use diverted prescription opioids exclusively. A large study of substance use trends in New York City in 2008 found that about 1 in 5 individuals who had recently used diverted prescription opioids also reported snorting heroin and about 1 in 5 had reported injecting heroin in the past year (Davis & Johnson, 2008). Nevertheless, rates of fatal overdose involving prescription opioids tripled over the same time period (Compton, Jones & Baldwin, 2016).

In 2010, the FDA approved a new, abuse-deterrent formulation of OxyContin, which “[was] intended to prevent the opioid medication from being cut, broken, chewed, crushed, or dissolved to release more medication” (U.S. Food & Drug Adminisration, 2010). In essence, this meant that commercially manufactured OxyContin, then an extremely popular product in the U.S. illicit/diverted drug market, could no longer be readily prepared or processed for snorting or injecting. Heroin—cheaper, purer, and more prevalent than it had ever been thanks to the glut of Colombian-sourced product—quickly became the substitute for many who could no longer access OxyContin. By 2012, ethnographic research conducted in several U.S. urban centers identified a sizeable cohort of (often, but not always) younger people who inject opioids who had initiated drug injection with diverted prescription opioids (like OxyContin) but then transitioned to heroin injection when the availability of early formulation OxyContin dropped and its street price increased (Cicero, Ellis & Surratt, 2012; Mars et al., 2014). Retrospective studies of nation-wide health data have directly linked the OxyContin reformulation to higher rates of hepatitis c (Powell, Alpert & Pacula, 2019) and opioid overdose (Evan & Lieber, 2018), indicating that this transition from OxyContin to heroin was widespread. Between 2010 and 2012, heroin overdose deaths doubled in 28 U.S. states (Rudd, Aleshire, Zibbell & Gladden, 2016).

In 2013, the State Health Laboratory and the State Medical Examiner of Rhode Island—where the study presented in this paper was conducted—reported a series of unusual deaths to the U.S. Centers for Disease Control and Prevention (CDC). Ten decedents who had experienced a fatal overdose tested positive for the synthetic opioid acetyl fentanyl in postmortem toxicology screening (U.S. Centers for Disease Control & Prevention, 2013). This was the first time that a fentanyl analog not commercially available (i.e. not prescription fentanyl) was identified in a cluster of fatal overdoses. Similar fentanyl-associated fatalities continued to appear in Rhode Island into the next year (Mercado-Crespo, Sumner, Spelke, Sugerman & Stanley, 2014). Fentanyl-related fatalities began to appear in neighboring Massachusetts in 2014 (Somerville et al., 2017), and, by 2016, were observed across the eastern and central regions of the U.S. (O’Donnell, 2017; Peterson et al., 2016). The prevalence of fentanyl in the illicit market—by now well established as a supply-side response to shifts in the heroin market rather than a result of consumer demand (Mars, Rosenblum & Ciccarone, 2018)—has since continued to spread across the U.S., fundamentally changing the shape of the illicit drug market and the opioid overdose epidemic. By 2017, 47,600 opioid overdose deaths were identified in the U.S., more than 28,000 of which involved synthetic opioids like fentanyl (U.S. Centers for Disease Control & Prevention, 2019a, 2019b).

In the age of fentanyl, navigating product uncertainty in the illicit opioid market in Rhode Island—as elsewhere—presents many challenges. The gradual saturation of fentanyl into the New England drug market brought with it a sudden proliferation of visually and chemically varied “heroin” products that, according to consumers, presented with an assortment of atypical colors, textures, smells, and phenomenological effects (Carroll et al., 2017; Ciccarone et al., 2017). In Rhode Island, specifically, individuals who use opioids have reported adopting a variety of pseudo-strategies for detecting and avoiding fentanyl-contaminated drugs (checking for taste, smell, color when cooked in solution, etc.) as well as harm reduction strategies for reducing overdose risk (such as using a smaller quantity of drugs or consuming drugs more slowly) when the purity and potency of the drug is unknown, but these strategies are not always reliable (Carroll et al., 2017; Rouhani, Park, Morales, Green & Sherman, 2019). While it may have once been the case that adding inert or neutral cut to illicit opioid products (and thereby decreasing their potency) was neither universal nor systematic in the 1990s (Coomber, 1999), broad consensus has emerged in the age of fentanyl that products are frequently adulterated—either with neutral cutting agents, or with powerful synthetic opioids, or both—at unknown stages in the supply chain (Ciccarone et al., 2017; Mars, Ondocsin & Ciccarone, 2018). In Rhode Island, as elsewhere, the high prevalence of social supply (drugs brought and sold among close contacts with little or no regard to profit motive) in the local market obscures the presence and origin of fentanyl as “cut” even further, as both consumers and low level suppliers (including social suppliers) are an additional step removed from the adulteration process and the supply chains in which that adulteration takes place.

It is well established that product source, product purity, and the relative openness or closedness of a local drug market varies geographically between—and sometimes within—different parts of the U.S. (Mars et al., 2018; Mars, Bourgois, Karandinos, Montero & Ciccarone, 2016; Rosenblum et al., 2014). This variability, in turn, shapes the local risk environment for overdose (Mars et al., 2015) as well as the kinds of trust that may be established between supplier and consumer within a particular market (Carroll et al., 2017; Mars et al., 2018; Rhodes et al., 2019). Though trust in one’s supplier has been reported as a reason for disinterest in drug checking technology—such as fentanyl test strips—(Bardwell, Boyd, Arredondo, McNeil & Kerr, 2019), high levels of reported trust in local suppliers has not dampened enthusiasm for drug checking tools in Rhode Island (Goldman et al., 2019; Krieger et al., 2018). These strips only became widely available to individuals who use drugs in Rhode Island in 2018 (Miller, 2018), more than a year after this present study was completed. Thus, participants in this study were limited in their ability to identify fentanyl in the local drug supply through the use of fentanyl test strips, reliant on their own physical senses and the information that passed between consumers and, occasionally, their suppliers, to make consumer choices—choices that were, at best, only partially informed.

A total of 92 individuals completed a survey and an interview for this study. Of those, 33 (35.9%) identified as female and 41 (44.6%) identified with a racial or ethnic group other than white. The majority of participants (n = 78, 84.8%) reported regular (at least weekly) non-medical use of an opioid of some kind. A small minority of participants (n = 9, 9.8%) reported using opioids less than weekly. The majority (n = 55, 59.8%) reported regular (at least weekly) use of heroin, specifically. The remaining minority reported regular non-medical use of prescription opioids (n = 12, 13.0%), or use of both heroin and prescription opioids on a regular basis (n = 11, 12.0%). The remaining 14 participants either reported intermittent use or were missing this data on the frequency of opioid consumption on their survey form. No significant differences in preference for heroin or prescription opioids were found between male-and female-identifying participants or between white and non-white participants.

A subset of 51 individuals (55.4% of the full study population) discussed their relationships with drug suppliers in their interview and, thus, were included in the qualitative data analysis presented here. Among that subset, the proportion of male-identifying participants who are non-white is nearly identical to that in the full study population (n = 13, 37.1% and n = 21, 35.6%, respectively). Among female-identified participants, however, individuals who discussed consumer–supplier relationships were almost exclusively white (n = 13, 81.3%); race data is missing for 2 female-identifying participants, and the remaining female participant identified her racial group as “other,” indicating that she preferred to be designated as “Italian.” (see Table 1). Though many participants reported having social and/or commercial connections to multiple dealers, all but one participant in this subset (n = 50, 98%) reported using a single, “primary” dealer from whom they prefer to procure opioids the majority (if not all) of the time.

The findings of this study suggest that, for many people who use drugs in Providence, Rhode Island, maintaining long-term relationships with trusted dealers is a key strategy for reducing the risk of substance use-related harm. Though not universal, a sizable number of participants reported typical behaviors from their dealers that align with the goals of consumer protection (i.e. refusing to sell fentanyl or openly communicating with clients about the presence or absence of fentanyl in heroin being sold), quality assurance (i.e. self-designed methods of “testing” heroin for fentanyl prior to selling it, seeking feedback and checking in post purchase), emergency first response (i.e. procuring naloxone and facilitating overdose reversal), and other forms of social and logistical support. In other words, some people who use opioids maintain generally positive relationships with their dealers, and those relationships appear to be protective against overdose as well as conducive to safer substance use behaviors.

The findings of this study also reveal that access to these potentially protective consumer-supplier relationships is not universal. Though the sampling method and the sample size of this cohort precludes any meaningful correlation analysis, several trends in the data bear explicit mention. First, male-identifying participants were much more likely than female-identifying participants to report a close relationship with their primary dealer with roots in a pre-existing friendship. Whatever the cause, it is possible that women are likely to face additional barriers to trust and social intimacy in these relationships—not least of which because they are more likely to need to build those relationships from scratch with each new supplier they meet. Second, though few people in this cohort reported having no meaningful relationships with suppliers, typically relying on the ability to buy from strangers or poorly-known acquaintances when buying drugs, the concentration of these reports among male-identifying participants of color (especially in a state like Rhode Island whose population is predominantly of white race) suggest that these individuals may be vulnerable to social isolation and, subsequently, greater risk of opioid-related harm than their white counterparts.

Further, this study predominantly included individuals who were already well connected to and regularly receiving services from a syringe services program or other community support organizations that distribute safer injection supplies and provide services with a harm reduction approach. Evidence of syringe service program’s protective effect against the spread of HIV and other blood-borne diseases is very-well documented (Abdul-Quader et al., 2013; Cooper et al., 2012). The individuals included in this study, many of whom appear to face a lower risk of overdose thanks, in part, to their supplier, have likely also significantly reduced their risk of overdose through receipt of harm reduction services. It is conceivable that reliable receipt of such services is an indicator of the social support networks that many of the participants in this study already enjoy. In other words, it is possible that individuals who have fostered deeper relationships with harm reduction staff may be more likely to have also fostered deeper relationships with their suppliers—using the same risk mitigation strategy in multiple domains of their personal lives. If this is, indeed, the case, then structural barriers to safer injection supplies and social barriers to trusted consumer-supplier relationships would likely have synergistic effects—amplifying both the risk of infectious disease and the risk of overdose among some populations while jointly reducing those risks in others. Future studies should investigate possible correlations between structural (macro-level) risk factors, socio-relational (meso‑level) risk factors, social capital, and incidence of overdose and other substance use-related harms.

Our findings are congruent with those found in the North Carolina study (Rhodes et al., 2019). Participants in that study also reported using trusted dealers as a personal fentanyl-avoidance and overdose-prevention strategy. In addition, participants in the North Carolina study indicated that their dealer stopped selling a particular batch after “a lot of people OD’d,” just as many participants in our study reported doing (Rhodes et al., 2019). Importantly, the study presented here also lends support to two conclusions put forward by the authors of the North Carolina study. First, both studies indicate that distinctions between drug “sellers” and drug “consumers” are often muddy. Many people in both studies reported buying and selling from friends, from individuals who also use, or from individuals whom they often use with. This fact throws into sharp relief the contradictions inherent in many states’ newly adopted “drug-induced homicide” or “death by distribution” laws, which typically allow for homicide charges to be brought against a “seller” believed to have sold drugs that resulted in an overdose (Blanchard, 2019; Mulvaney, 2017). Second, the findings of both studies imply that removing access to trusted dealers may put clients (who rely on those dealers for their fentanyl avoidance and overdose prevention effects) at immediate risk of overdose. Indeed, for many individuals in this study, the inability to access a trusted supplier was reported as the specific event that precipitated their most recent overdose.

The policy implications of these findings are significant. Put bluntly, arresting a dealer may directly contribute to overdose within their client population. Overall, the impact that drug policy, public health interventions, and/or law enforcement responses to substance use may have on the protective strategies that people who use drugs have created for themselves remains poorly understood. Without such understanding, good-faith attempts to disrupt macro-level drivers of the opioid-overdose epidemic (police sweeps, dealer take-downs, sudden pain clinic closures, etc.) may in fact only result in creating more harm among those who are already at risk (Carroll, Rich & Green, 2018). In the context of such disruptions, at a minimum, action should be taken to coordinate with public health interventions to reduce the risk of unintended consequences (Carroll et al., 2018).

Further, the Rhode Island legislature joined numerous other U.S. states in passing its own drug-induced homicide law in 2018. This new law allows for a life-sentence to be handed down to individuals found guilty of distributing illicit substances when those substances were implicated in a fatal overdose (Shihipar & Peterson, 2018). As of January 2020, only five cases in total have been brought under this new law, yet coverage of these cases have been widespread in local and national media (Associated Press, 2019; O’Laughlin, 2019; U.S. Attorney’s Office, Western District of Rhode Island, 2019). The full impacts of these prosecutions on individuals who use drugs in Rhode Island and the relationships upon which they rely to navigate an uncertain drug market remain unknown, though many have suggested that further criminalization through these laws are likely to have little impact on substance use other than hindering 9–1–1 calls during an overdose (Peterson et al., 2019). Nevertheless, mixed-methods research indicates that substance use and illicit drug distribution cannot be effectively deterred through increased threat sanctions and arrests (Bailey, 1983; Friedman et al., 2006, 2011). Research on criminal deterrence has also concluded that efforts to deter through increased criminalization simply restrict the characteristics of illegal behaviors, altering how (not whether) individuals produce or distribute illicit substances, occasionally resulting in increased risk to consumers of substance use-related harms (Barratt, Chanteloup, Lenton & Marsh, 2005; Dickinson, 2017; Friedman et al., 2006). Drug induced homicide laws should, therefore, be thought of not as deterrence strategies but as selective pressures that change the shape of the drug market. For individuals who rely on trusted suppliers for survival in an increasingly deadly drug market, this market pressures produced by this law—ostensibly enacted in their name—may simply serve to disrupt the one lifelines they currently have.

These findings should be interpreted with certain study limitations in mind. Data collection was carried out in a single urban center—Providence, Rhode Island—at a time when fentanyl was still a relatively new feature of the local drug market. This data may not be representative of other regions with different populations or different historical changes in the drug supply. The individuals who participated in this study were predominantly recruited through direct service points and may not be representative of other people at risk of overdose who are from different (i.e. more affluent) socioeconomic backgrounds or who do not actively seek harm reduction services of any kind. Further, no demographic information was collected from participants about their primary or secondary dealers. Thus, this study is unable to assess generational differences among dealers (especially different social norms between older and younger—or more experienced and less experienced—dealers). Nor can this study elaborate how risk environments faced by people who use drugs may differ according to whether or not their primary dealer also uses drugs.

Finally, female-identified participants included in this qualitative study were almost exclusively white. Based on a review of the history of data collection activities, this discrepancy appears to have resulted due to an unanticipated confluence of sampling strategies, selection of recruitment locations, and timing of hypothesis generation. Regardless of the cause, female-identified participants of color are notably under-represented in this analysis. Future studies should strive to correct this imbalance by ensuring that the experiences and perspectives are adequately sought out across demographic strata—especially women of color and sexual and/or gender minorities of color.

In summary, these findings suggest that socio-relational factors, especially interpersonal relationships between individuals who use drugs and their suppliers, significantly impact the synergistic relationships across multiple substance use-related harms in today’s opioid-overdose epidemic. Evidence-based prevention strategies that are based on an awareness of—or even designed to harness—the positive and protective relationships that people who use drugs have already constructed for themselves are likely merited. Policy responses to the opioid-overdose epidemic should be organized around proven harm reduction and overdose prevention strategies, but, as this study indicates, there may be merit in considering the impact of those approaches on networks of people who use drugs, not simply on individuals.