A critical content analysis of media reporting on opioids: The social construction of an epidemic.

BACKGROUND
The 2000s have seen a proliferation of media reporting about opioid use in North America. Given the significant role that popular media plays in shaping the public's perceptions and understandings of the issues that it represents, analysing the content of this media coverage can help understand public discourse about opioid use.


METHODS
We conducted a critical content analysis of Canadian newsprint media reporting on opioids using a sociological lens. We performed a qualitative thematic analysis of these texts, coding 826 articles and applying a critical discourse analysis in our interpretation of the findings.


FINDINGS
Our analysis showed a slow transition from a conversation primarily about clinical pain care towards a discussion of criminality, especially the increasingly fluidity of boundaries between prescription opioid use and the illegal drug trade. Patients tend to be dichotomized as either innocently following physician prescriptions or drug-seeking, as an aspect of lives characterized by addiction and street crime. These depictions map onto characterizations of physicians as naively following pharmaceutical industry advice or becoming irrelevant once criminality is introduced.


DISCUSSION
The social construction of the opioid epidemic polarizes individuals as good or bad with little attention paid to underlying institutional interests both in the creation of the problem or in the solutions that are proposed. We show that as concerns about harms from opioids become more pronounced, the narrative shifts to home in on illicit street-use with a corresponding uptake of stigmatizing references to so-called addicts. Concurrently, most references to the pharmaceutical industry disappear from view. This framing of the problem defines the kinds of solutions that then seem natural. For example, increased criminalization is suggested for people who use drugs and stigmatizing those who suffer with chronic pain becomes a higher priority than implementing safer and more effective therapies for managing their pain.


Background
Since the turn of the millennium there has been a proliferation of news media reporting about the use of opioids in Canada. Given the significant role that popular media plays in shaping public perceptions and understandings of the issues and groups that it represents (Hodgetts and Chamberlain, 2014;Henderson et al., 2000) analysing the content of this media coverage can help understand public discourse about opioid use. Media is always political to the extent that its use reflects a particular perception of what the world is like and how it should be (Wienke, 2005). Accordingly, media analysis can shed light on the interactions between media representations and normative attitudes towards practices that are controversial, such as opioid use (Altheide and Schneider, 2013). Moreover, the information conveyed in media often shapes health policy (Gugsa et al., 2016;Hudacek et al., 2011). Given that opioid use is both a stigmatized practice Sharfstein, 2014a, 2014b) and a hot-button issue in the Canadian (Government of Canada, 2018; Weeks and Howlett, 2016) and international health policy contexts (Centre for Disease Control, 2018; National Institution on Drug Abuse, 2019), a media content analysis of this issue is timely and informative.
We report here on how Canadian popular news media have de- T research study, entitled COPE (chronic pain ethnography) (Webster et al., 2015). COPE uses a sociological lens through which to examine the organization of the care of patients with chronic pain across the province of Ontario. Having collected over 60 interviews with care providers -especially family physicians -and over 40 hours of ethnographic observations in primary care settings, we found that the prescription of opioids and their potential misuse were of major concern (Rice and Webster, 2017;Webster et al., 2019b). Moreover, it is clear from our study that policy decisions about prescription opioid use have fundamentally shaped the care that these patients receive (Webster et al., 2019a). This has been concurrent with an array of opioid-related harms such as epidemic levels of addiction, overdose and overdose mortality, all commonly referred to as the opioid epidemic. These issues have invoked urgent calls on policymakers to address issues of opioid use (Bains, 2016;Government of Canada, 2016). In examining this issue we have been guided by the following question: How are normative discourses about the nature, causes, and consequences of the opioid epidemic conveyed through print media, and how have these discourses developed over time?

Methods
We conducted a critical content analysis (Macnamara, 2005;Carvalho, 2008) of Canadian newsprint media reporting on opioids. Critical content analysis has been described as an approach that explores "power in social practices by understanding, uncovering and transforming conditions of inequity and locating sites of resistance and change. " (Short et al., 2016) Quantitative content analysis alone is limited by the fact that quantitative indicators are insufficient to capture the social meaning or impact of media messages (Newbold et al., 2002). A critical content analysis that takes a mixed qualitative and quantitative approach can achieve a deeper level of analysis by examining both the degree to which audiences are exposed to particular issues (quantitative), as well as identifying the themes, tone, and context of media coverage (qualitative) (Macnamara, 2005). Accordingly, we also undertook a qualitative thematic analysis of these media texts, coding the 826 articles according to the coding framework explained below (Schreier, 2012;Patton, 2014).
Our analysis drew on aspects of Foucauldian discourse analysis and followed the methods outlined by Anabela Carvalho (2008). In drawing on Foucault's use of discourse (Arribas-Ayllon and Walkerdine, 2017;Foucault, 1979), we analyzed ways of speaking with particular attention to practices that systematically organize what it is possible to say and do. We wanted to know where attention is drawn, and conversely from where attention is distracted, in public conversations about what has been widely coined the opioid crisis. We also sought to understand the history of the present in which the opioid crisis has unfolded nationally and internationally (Foucault, 1979) in order to emphasize how knowledge about this problem has been produced and introduced certain subjectivities found in characterizations such as the uninformed physician or the drug-seeking addict.
Our approach was informed by critical theory insofar as it focused on what Fairclough and others have termed the new capitalism (Fairclough, 2003) and in particular the "radical attacks on social welfare and the reduction of the protections against the effects of markets that welfare states provide for people" (Fairclough, 2003). As we have noted in our previous ethnographic work, many of those effected by both chronic pain and opioids are among the most vulnerable in our society (Webster et al., 2019a, b). Our current analysis therefore included identifying the collective definition of the opioid problem as articulated in public discourses, with a view to explicating the underlying social norms and interests driving these definitions (Brown, 1995;Goode and Ben-Yehuda, 1994;Kuper et al., 2013).
Our analysis is restricted to newsprint media, as this form of media is easily accessible for researchers through online databases (Gugsa et al., 2016). A third-year medical student (JER) consulted with a librarian to identify media sources and establish a timeline. Through this consultative process we decided to focus our search on three major Canadian national news sources as a matter of feasibility and because other smaller local newspaper outlets often carry stories originally run in the national papers (CBC News, The Globe and Mail, and the National Post). We began from the year 2000 since a preliminary search demonstrated that there was minimal newsprint reporting regarding opioids prior to this date. All data from 2000 to 2016 were sourced over a two-month period between July and August 2016. Data from September 2016 to December 2017 were sourced in October 2018. Searches were conducted using ProQuest, using the search terms opioid AND (prescription OR medication) OR (pain AND narcotic). After deleting duplicates and discarding irrelevant articles, this search identified 826 articles.
We analyzed these 826 articles using both qualitative and quantitative media content analysis methods (Hodgetts and Chamberlain, 2014;Carvalho, 2008). The quantitative component involved counting the number of references to particular topics, issues, or persons to identify prominent issues, themes, and trends in coverage (Hodgetts and Chamberlain, 2014). To determine which issues we would focus on, a random sampling of twenty articles was selected and independently read by two authors and a medical student (FW, KR, and JER). An initial coding framework was developed through a series of team meetings and organized around the themes that emerged from the twenty randomly-selected articles. Differences in interpretation were resolved by consensus.
A data extraction form (Appendix A) was developed collaboratively as a means of systematically recording data derived from the articles. Tracked topics include which individuals and/or groups were reported to be affected by the opioid epidemic (e.g. teenagers, drug addicts), the source or nature of the problem (e.g. inappropriate prescribing) and what solutions were proposed to the opioid epidemic (e.g. abstinence, additional law enforcement). JER coded the 425 articles that were published between 2000 and July 2001 and KR coded the 401 articles that were published between July 2016 and December 2017. NVIVO 10 data analysis software was used for storage and organizational purposes. Throughout the coding phase JER and KR were in regular communication with the principal investigator FW, and the coding framework was modified as needed through an iterative, collaborative process (Hodgetts and Chamberlain, 2014).
A key objective was to examine changes in how opioids were framed over time, as a means of exposing how normative discourses have shifted. Failure to incorporate a temporal element has been previously identified as a limitation of many content analyses (Carvalho, 2008). We planned to randomly select 10 articles from each year between 2000 and 2017 for more in-depth analysis. However, prior to 2006 fewer than 10 sources were published each year. To address this, we grouped articles from the early 2000s into two phases: year 2000-2001 (4 articles) and 2004-2005 (4 articles). We found no articles from 2002 to 2003. From 2006 forward there were 10 or more sources published each year and so each year from 2006 to 2017 was examined as a stand-alone group. This sampling strategy produced a total of 128 articles for more in-depth qualitative thematic analysis. KR, a medical anthropologist with longstanding background in thematic analysis, read each group of articles in detail in order to confirm the themes identified before moving on to the next batch. A sub-selection of additional articles was also read by the principal author, FW, a sociologist who has worked extensively conducting ethnographies in clinical settings, and by AS, a primary care physician with training in linguistics who has a leadership role in the development and delivery of opioid policy strategies.

Findings
We have organized our findings around the overall themes we identified in the selected media reports. We note, first of all, the dramatic increase in media reporting over time (see Fig. 1). Our headings are therefore descriptive of the accounts we found and do not represent our own view of the issues. Following the broad principles of discourse analysis (Arribas-Ayllon and Walkerdine, 2017) we paid attention not only to what was said but how it was articulated. Words of course do not just describe, they also create. Our analysis highlights some of the widely held social values, biases and beliefs underlying media reporting on opioids.

Nature of the problem
Many articles used dramatic and sensational language to describe opioid use in Canada; these include references to "crisis"; a "hidden killer; and a "skyrocketing" problem. Moreover, there was a lot of overlap between defining the scope of the problem (Table 1), and efforts to identify which individuals and groups are to blame for the problems arising from opioid use. For example, roughly 10 per cent of all articles identified unsafe prescribing as the root of the problem, thereby implicating physicians as a blameworthy cohort. Similarly, discussions of the legal and political context were bound up with blaming governing bodies (e.g. provincial and federal governments). This perspective is evident from the following data extracts: Medical experts said Ontario is ill-equipped to address the opioid crisis, which ranks as a leading cause of accidental deaths in the province. They are calling for pre-emptive measures to manage the outbreak, including wider availability of naloxone ("Fentanyl fatalities surge to new heights", Howlett & Woo, The Globe and Mail, Feb 22, 2016).
Hedy Fry, the Liberal health critic, said [Rona] Ambrose, the Federal Minister of Health, signalled the expansion of opioid supply by generic manufacturers was a concern two years ago, but has done nothing about it. "During that time there have been increasing addiction numbers in Canada, particularly among First Nations" ("Ottawa moves to rein in Oxycontin; 'Silent Killer'; About turn on permitting generic versions", Ivison, National Post, May 15, 2015).
A secondary theme was that of criminality, which encompassed problems of drug trafficking (85 articles, 10.3% of the total), the behaviour of those labelled as addicts (for instance, seeking out fentanyl on the street; 54 articles, 6.5%), the crushability of prescription opioids (which can then be taken in manner that is not recommended by clinicians, but which produces a more immediate and intense psychoactive effect; 28 articles, 3.4%) and unscrupulous physicians and pharmacists who knowingly profit from the illegal trade in prescription  *Articles for 2016-2017 are listed separately given the significant increase in number of articles during this time period.
Affected Individuals/Groups Note: These tabled percentages reflect the percentage (rounded to one decimal place) of total articles within the data set; many articles do not identify specific individuals or groups. opioids or who barter opioids for sexual favours from vulnerable patients.

Who is thought to be affected?
We tracked which groups or individuals were reported as being affected by or at risk from opioids (see Table 2). The three groups most frequently referenced were: "young people" (102 articles; 12.3%), followed by "legitimate users" (82 articles; 9.9%), "women" (63 articles; 7.6%), and "addicts" (49 articles; 5.9%). We emphasize that these categories emerged from the data, and do not reflect our own moral judgements or terminology. "Legitimate users" were described as people who had become dependent or had overdosed after being prescribed opioids to treat persistent pain. These individuals were often framed as having been misled to believe that opioids are not addictive and pose little risk to health, as the following excerpt illustrates: She [young woman prescribed Oxycontin following a knee injury] still has a hard time comprehending how a simple prescription to help her cope with an injury nearly destroyed her life. "I had no idea they could be as addicting and powerful as they were," she said in an interview. "Nobody's really immune to this. It could be anybody" ("Alternative opioids surge in wake of oxycodone crackdown", Weeks, Carly; Howlett, Karen, The Globe and Mail, Aug 19, 2015) Many articles evoked a moral distinction between legitimate users of the opioid epidemic, and stigmatized groups such as poor, streetaffected drug-users. This distinction was both produced by journalists, and also emerged from the journalists' selection and interpretation of first-person narratives of interviewees featured in these articles. For instance: Leonard Briere is no junkie. The Manitoba man severely damaged his right hand and wrist in a November 2013 gun accident and joined the six million Canadians who depend on prescribed opioid pain killers to get through the day. But Briere can't get the drugs he says he needs to return to work as a Winnipeg-area construction superintendent ("The other side of opioids; Painkiller crackdown hurts genuine sufferers", Ian Macleod, National Post, May 9, 2016). "I'm a business owner," he [Jim Burke] added. "I'm fully active in the community. And I feel like I'm getting lumped into the same category as drug addicts." ("Government asked to boost generic OxyContin supplies," CBC News, May 19, 2014).

Solutions
Solutions offered for how the crisis should be addressed show highly divisive perspectives. On the one hand, we found a range of solutions and strategies that take a non-punitive approach and are sympathetic towards opioid users. These include developing or scaling up access to addiction treatments such as methadone, buprenorphine and detoxification services, availability of antidotes (such as naloxone), improving patient education, and, in a very few cases, considering decriminalization of drugs. These approaches stand in sharp contrast to proposed solutions which take a more punitive approach. These include mobilizing law enforcement to monitor and police opioid users and imposing stricter regulatory consequences for physicians who prescribe inappropriately. The conceptual distance between these approaches is evident in the following extracts: The reasons for misuse of drugs can be complex and there is no magical solution -especially not the simplistic "tough on crime" approach. As R. Gil Kerlikowske, former head of the U.S. Office of National Drug Control Policy, has said: "We cannot arrest our way out of the drug problem. What is required is a series of measures, such as better education about the real benefits and risks of drugs like painkillers (for patients and medical practitioners alike), sounder prescribing practices, investment in non-pharmaceutical pain-control methods, better access to addiction treatment and harm-reduction measures. What is also needed are pragmatic approaches to dealing with one of the frightening symptoms of the public health crisis that is opioid abuse: overdoses ("Opioid overdose tool is harm-reduction at its best", The Globe and Mail, André Picard, Apr 14, 2014) The dramatic increase in the abuse of opioids such as Oxycontin has led the Ontario College of Physicians and Surgeons to propose some dramatic initiatives to curb addiction. Key among the recommendations is more information-sharing among doctors, pharmacists and police. Along with establishing a drug monitoring system, the college is also calling on the Ontario government to change the Personal Health Information and Protection Act to compel doctors to contact police if they believe a patient is breaking the law ("Ontario doctors tackle opioid abuse", CBC News, Sept 9 2010).
Of the minority of articles that are not closely aligned with either of the approaches outlined above, 63 articles (7.6%) advanced the position that improved non-opioid treatments for chronic pain would help alleviate the problems caused by prescription opioids. This solution was only raised in 6 articles prior to August 2016, in contrast to 57 articles between September 2016 and December 2017. This recommendation is grounded in the notion that patients request opioids and physicians prescribe them because of the lack of alternative forms of effective pain relief, thereby linking the opioid epidemic to a chronic pain epidemic in Canada (Fashler et al., 2016;Schopflocher et al., 2011). A number of articles also propose modifying the commercial products themselves to make it more difficult to alter their route of delivery. Finally, a minority of articles (16, 2%) voiced the pessimistic opinion that the problem is now so out of control that there is little hope for solving it. As suggested by an editorial writer in The National Post, the opioid epidemic in Canada has become "a classic case of War on Drugs whack-a-mole, and F. Webster, et al. Social Science & Medicine 244 (2020) 112642 there is no comprehensive solution in sight" (Jan 30, 2015). This is in reference to the continued availability of newer opioids on the street regardless of physicians' approach to prescribing, and the fact that reduced access to a given opioid (in this case, OxyContin®) merely results in an increase in overdoses from other opioids (see Table 3). Tracing the development and emergence of these themes through time exposes an evolving narrative (see Fig. 2). As the image shows, we see a slow transition from a conversation primarily about clinical pain care towards a discussion of criminality, in particular the increasingly fluid boundaries between prescription opioid use and the illegal drug trade. In this shifting conversation, patients tend to be presented as either innocently following physician prescriptions at one end of the spectrum, or "drug-seeking" street users at the other. Their depictions map onto how physicians are perceived; as naively and misguidedly following pharmaceutical industry advice or in fact irrelevant once criminality is introduced -except for rare instances when they are implicated in this criminality.
Coverage from the early 2000s emphasizes the inadequacy of current treatment options for chronic pain and the need for more compassionate policies around methadone use and the availability of safeinjection sites for street-affected users. These early articles convey the notion that there is much that is unknown about both chronic pain and opioids, but that it is nevertheless clear that these drugs are necessary, low-risk, and effective for relief of chronic pain. In addition, spokespersons in this early time period were often physicians who in fact advocated for the use of opioids for chronic pain. An example of this is the article by Mike Nichols, wherein a cancer patient named Alice says that "If I wasn't so afraid of death I would have taken my life a long time ago" as a response to her inadequately managed pain (Macleans, Coping with Pain, January 4, 2001).
By the mid-2000s we encounter in our archive the first reported cases of people -usually teens -dying through the use of prescription opioids. Accidental prescription opioid-related deaths continue to feature in media coverage up to the present day. Early on (circa 2005-2006) these tragedies are framed as a consequence of naiveté on the part of both prescribing physicians and users, some of whom are accessing these drugs illegally. For example, one article in 2004 led with "Two teenagers -one with a sore throat, the other headachesdied within hours of being prescribed a powerful pain-relief patch designed for treating adults with chronic pain" (Tom Blackwell, National Post, November 6, 2004).These sources continue the narrative that opioids are effective, needed, and that chronic pain sufferers are entitled to the relief that these drugs provide. It is claimed that existing policies and guidelines are excellent, but some physicians and members of the public are poorly informed.
In 2006 we see the emergence of a theme that continues and escalates in medical coverage right up to the present: the increasing fluidity of boundaries between prescription opioid use and the illegal drug trade. This is, by far, the single most prominent theme to emerge from the mid 2000s. It is clearly conveyed in these articles that prescription drugs (especially oxycodone) are set to replace drugs such as heroin and crack as the drugs of choice for street-affected drug users, and that  police drug raids frequently entail seizures of prescription opioids alongside illegal drugs (e.g. cocaine). In this narrative, people who use drugs are positioned as criminals and the state (police) are brought in to maintain public safety. In 2010 we witness a departure from the notion that opioids are effective and necessary for pain relief. In contrast to coverage from the early 2000s, by 2010 these drugs are no longer described as safe, but rather as highly addictive and risky. Readers are now presented with the perspective that increases in prescriptions are worrisome developments. This impression becomes amplified over the next couple of years. Physicians now begin to appear in news media accounts with a different message, as expert interviewees raising alarm bells around the toxicity of opioids ("One-third of opioid-related deaths in Ontario were among people actively treated with a prescription opioid in 2016", Toronto Star, August 29, 2018). By 2012, media coverage clearly conveys a sense of danger around prescription opioids. The tone suggests that Canada is facing an addiction problem that is out of control and is difficult to manage. In particular, there are increasing concerns that the increased availability of prescription opioids has turned unknowing patients legitimately seeking treatment into addicts who will source opioids illegally in desperation and that they are able to do so because the supply of prescription opioids is now connected to sophisticated networks of organized crime.
While early articles focused primarily on Atlantic Canada as the locus of prescription opioid harms (either the provinces of Prince Edward Island, Newfoundland, and Nova Scotia specifically, or the Atlantic region as a whole), by 2013 we see an acknowledgement that opioid harms are a nation-wide problem. By 2016, the geographic focus has shifted westward to prairie Canada and British Columbia, where both the illegal street trade in fentanyl and fentanyl-related deaths are occurring on a disproportionate scale. Canada did not start tracking opioid deaths nationally until 2016, but deaths involving fentanyl have risen dramatically from 47% of all deaths in the beginning of 2016 to 75% by the end of 2018 (Government of Canada, 2019). Over this same time period, the overdose mortality rate in British Columbia has been greater than 20 per 100,000 compared to a rate of less than 10 per 100,000 in the Atlantic provinces (Fashler et al., 2016;Schopflocher et al., 2011).
Furthermore, from 2013 onward there is growing acknowledgement in the popular media that existing addiction and detoxification services are inadequate. This dovetails with greater but still minimal discussion about marginalization (e.g. socioeconomic, racialized) in determining both whether a patient is able to access adequate pain relief, and how likely they are to be a victim of opioid harms. In sharp contrast to coverage a decade earlier, after 2014 we see much longer, featurelength articles, including first-person narratives of street-affected opioid users. These people are generally presented in a sympathetic light, variously as victims of inappropriate prescribing, slaves to powerful addictions that they are unable to overcome given the inadequacy of existing services, and as key experts and allies in fighting the epidemic given their first-person expertise and unique access to individuals at risk of overdose (e.g. as users themselves, they are able to recognize an overdose and can administer naloxone to others in their networks).
Finally, by 2015-2017 the narrative has shifted one final time to focus almost exclusively on overdose deaths caused by fentanyl, corresponding with a sharp rise in the proportion of deaths involving fentanyl (Quinones, 1975). These articles suggest that the opioid crisis in Canada now involves complex and sophisticated international networks of illegal drug manufacturing and distribution, alongside the appearance of synthetic chemicals that were never commercially produced or available by prescription (e.g. W-18). Physicians' prescribing practices -and indeed the entire health care system and pharmaceutical industry -have largely disappeared from view, with the notable exception of the Sackler family, whose investment in, benefit from and advocacy of opioid manufacturing did become news. However, the main focus remains on individual physicians or patients as being problematic --rather than how pharma's influence is institutionalized and embedded in medicine.

Discussion
The amount of human suffering and death due to opioid use is tragic and overwhelming. Sadly, ill effects from prescription drugs are not a new problem. For example, Thalidomide, a drug prescribed to pregnant women, resulted in birth defects in many newborns (Botting, 2002). Likewise, epidemics of prescribed opioids have been described as far as back as the l800s in the United States (Quinones, 1975). Besides prescribed substances, harm and death through alcohol remains a problem of vast consequence both nationally and internationally (Ramstedt, 2004;WHO Report, 2018) yet is less often reported as an epidemic or crisis.This leads us to reflect on how this contemporary crisis is socially constructed; not to refute the damage and loss caused by opioids but in order to understand the particular social values, biases and concerns that are reflected in how it is discussed in the media (McHoul and Grace, 1993).
First we note that many of the problems with drugs or substances are reported as separate rather than inter-related phenomena. For example, in their examination of opioid deaths in Ontario, Canada Gomes et al. note that 26 per cent of decedents had a diagnosis of alcohol use disorder, 28 per cent had an active benzodiazepine prescription at the time of death and 42 per cent had an emergency department visit for mental health disorder within three years of their death due to opioids (Gomes et al., 2018). Similarly, the association between alcohol related deaths and low SES is widely known (Wilkinson and Marmot, 2003). However, these associations are rarely commented upon in news media reporting of opioids. In other words, very little if any discursive light is shone on the associations between opioid use, mental illness, alcohol use, and poverty. By keeping these dimensions separate, the common systemic or institutional forces at play in such phenomena are kept relatively invisible.
The line between socially sanctioned substances and illegal drugs is also more clearly drawn in media accounts than it might be in reality. Alcohol is sold through government sanctioned stores throughout most of the world, for example. Similarly, opioids are largely physician prescribed so the medical community is implicated in this problem. Those who prescribe -physicians -are not viewed as "drug pushers" but as naïve and well-intentioned practitioners who have been influenced by dishonest pharmaceutical companies  in the same way that historically physicians were deployed by tobacco companies in advertising campaigns (Gardner and Brandt, 2006). An interesting twist on this is that some advocacy groups are currently calling for a "safe" drug supply of opioids and stimulants and use the distribution systems for alcohol and other substances as a model to draw from (Safe Supply Fact Sheet, 2019).
A number of articles do feature physicians behaving very badly. However, in focusing on crimes like trading opioids for sex, these articles frame the problem as moral failings by individual physicians, rather than holding physicians accountable as a group of actors socially located within the institution of health care services. Much of physicians' work is prioritized around the distribution of pharmaceuticals as much as, if not more so, than the distribution of healthcare (Webster et al., 2019a;Whitehead et al., 2013). By positioning their role in the current crisis as either naïve or as outliers, physicians as a group maintain their status as highly respected members of Canadian society.
Related to this, people who are members of certain higher status groups who become addicted, such as those who are white and middle class, are often portrayed as innocent victims. This careful delineation between street users and innocents further highlights the stigma associated with addiction; which in turn is about stigma associated with lower socio-economic status, poverty, lack of adequate housing, education, etc. There is a tension embedded in this narrative in that the situation becomes a crisis only when it moves into the domain of crime.
F. Webster, et al. Social Science & Medicine 244 (2020) 112642 This is also concurrent with it affecting poor people/"drug addicts" for whom the public has less sympathy compared to innocent people whose opioid exposure was iatrogenic, or infants whose exposure was outside of their control.
In their important work on exposing the re-racialization of addiction inherent in what they term the "white opioid epidemic" (Mendoza et al., 2018) note that "narratives of white opioid users [have] disrupted notions of the addict as 'other', producing alternative logics of blame that focus on prescribers and the encroachment of dealers from outside of white neighbourhoods". Mendoza et al. contrast the recent "surge in opioid overdose deaths among suburban and rural Whites across the United States" with the "war on drugs" that took place in the 1980s-90s when drug-related deaths were framed as largely affecting the urban Black and Latinx populations. They claim that community groups and policymakers have been much more sympathetic to the white and middle-class victims of this new epidemic and more likely to invoke "narratives of individuals trapped in a vicious cycle of emotional or physical pain and trauma, self-medication with opiates and unwitting dependency" (Mendoza et al., 2018).
Impacts on Indigenous communities did not emerge as a major theme in this content analysis, despite there being good evidence of disproportionate harm. For example, an article in the Canadian Medical Association Journal (CMAJ) in 2018 reported that "In British Columbia, the mortality rate for Indigenous people who use drugs is 5 times higher than for other drug users. Despite representing just 2.6% of the total population, Indigenous Peoples account for 10% of overdose deaths. Indigenous women are 8 times more likely to have a nonfatal overdose and 5 times more likely to have a fatal overdose than non-Indigenous women. The severity of this crisis is likely understated owing to poor disaggregation of data on Indigenous Peoples in many settings." (Reconciliation and Canada's overdose crisis: responding to the needs of Indigenous Peoples, 2018) This last point is a peculiarly Canadian problem which has not systematically collected race data that can be linked to health outcomes, as is so readily done in the US.
In examining media narratives about opioid use over time, it does become clear that it is only viewed as a crisis once it moves from a context that feels safe -the clinic, the pharmacy, and the relationship between patient and physician -into the domain of street crime and illegal drug manufacturing. Although deaths from prescription opioids have been occurring since the early 2000s, news coverage becomes more dramatic and pessimistic once the proposed solutions involve policing and fighting international drug cartels, rather than regulating physicians' prescribing practices.
As part of this analysis we turn our attention to noticing what is absent from many of the stories told -at least until very recently: specifically, why other forms of treatment and pain relief are not available, thus in part creating a need that is then filled by a harmful class of drugs. What is most strikingly absent is focus on the role of the pharmaceutical industry in Western health care as being a key part of the systemic root of the problem. The dominant narrative to emerge from these articles is that the the problem is bad or naive doctors who inappropriately prescribe, ill-informed patients who take opioids for pain, and street-affected "addicts". This directs attention away from focusing on the interests of the pharmaceutical industry that sometimes conflict with good health care and public health. Ultimately, this insistent emphasis on individual physicians or individual patients as the problem leaves us only with individual-level solutions, such as remediation for physicians accused of prescribing inappropriately or education for patients. As we have noted elsewhere (Webster et al., 2019a), the very naming of the opioid crisis casts it as a medical problem rather than an issue of public health. This chimes with the "unified neoliberalist agenda" (Brooks Dollar, 2018) as articulated by Dollar. In her illuminating comparison of criminalizing responses to crack cocaine, methamphetamine, and opioids in the United States, Dollar outlines how "pharmaceutical companies were well-aware that their prescribed and over-the-counter cold and allergy products were being used in manufacturing methamphetamine in the 1980s, but not wanting to compromise their profit, drug companies aggressively lobbied against legal regulation of their products." She posits that although bills were introduced to Congress calling for regulation, they did not succeed "owing largely to the multi-million-dollar contributions to politicians by pharmaceutical companies." (Brooks Dollar, 2018). Dollar draws a more explicit association between industry, politics, and the media than we can offer here. However, we do note that the neoliberal focus on individual responsibility permeates much of the media reporting in our analysis. Furthermore, some of the solutions proposed in media accounts, such as criminalization, policing, surveillance, and even patient education, in fact strongly reproduce the notion that people who are harmed by opioids are often blameworthy and even criminal. Such depictions move us even further away from understanding the opioid epidemic as being connected to issues of social justice, social inequities, or the biomedical and pharmaceutical-industry dominance of health care. Some of the reports do in fact hint at something more nuanced, such as the need for more family support, effective non-opioid treatments for chronic pain care, and better addiction treatment centres as well as harm reduction and de-criminalization. Ironically, media reporting, rather than the discursive activities of the medical or pharmaceutical industries, might have been more responsible for pushing the issue of harms from opioids to the foreground. Yet none of the reporting goes so far as to identify the biomedical and social embedding of opioids as the loci of the crisis, rather than the substances themselves.
These findings also align with our previous ethnographic work. In our ongoing and multi-year ethnographic study we have suggested that the language of the opioid crisis masks rising social inequities in the field of chronic pain (Webster et al., 2019a). It also locates the problem discursively in a class of drugs rather than in the historical processes and institutional priorities in which it is embedded. This in turn gives rise to intensifying concerns about patients who are feared to be drugseeking and reflects a situation in which blame for institutionalized lack of support and treatment are shifted to the most vulnerable actors in the social organization of care.

Limitations
Our search terms may not have been general enough to capture all opioids as it included concepts such as prescription, medication, and pain which keeps a focus on medical opioids. Future research could expand on the search we present here in order to determine if this leads to a new or different analysis.
Media texts are polysemic; as with all texts print media are subject to interpretation and thus cannot be analyzed objectively (Macnamara, 2005). Therefore, we cannot be certain that our interpretation is indicative of the meanings others will assign to the same texts. Nevertheless, while most texts can have multiple meanings, most have a preferred meaning which is difficult to ignore (Hall, 1980). We have aimed to identify and ground our analysis in these preferred meanings. Furthermore, rigorous media analysis seeks to link media texts to the social practices and political context from within which these texts arise (Hodgetts and Chamberlain, 2014). The larger ethnographic study within which this content analysis is embedded seeks to explore precisely this context. This in-depth knowledge of relevant content informed all stages of this study, and thus strengthens the validity of our analysis (Macnamara, 2005).

Conclusion
We strongly concur that opioid addiction and misuse are serious public health issues, with grave, often fatal, consequences for Canadians. In critiquing media discourse, we do not suggest that anxieties about opioids are unfounded. The problems of addiction and overdose are significant and real, and proposed solutions such as F. Webster, et al. Social Science & Medicine 244 (2020) 112642 improved access to antidotes and addiction treatments are urgently needed. We were interested, however, in what stories are told about these problems and specifically what may be left out by analyzing what is reported in the media, in order to advocate for an approach that attends both to the immediate needs of those affected by opioid harms, and to the broader structural circumstances of the current epidemic. In our study, we have shown that as concerns about harms from opioids are increasingly reported, the narrative moves away from stories about naïve physicians and so-called innocent patients towards an increasing focus on street-use and criminalization with a corresponding increase in stigmatizing references to those labelled as addicts. With this shift, references to broader social circumstances such as social inequity and racism disappear from view. Importantly, this specific framing of the problem defines the kinds of solutions that then seem natural. For example, people who use drugs or who suffer from poverty, mental health and addictions become criminalized and blaming and stigmatizing those who suffer with chronic pain becomes a higher priority than implementing safer and effective therapies for managing their pain.