Our finding that articles about prescription opioids routinely preceded overdose deaths associated with the same medications supports previous research, but we cannot draw a causal conclusion from our study alone, as other researchers have pointed out 
. In conjunction with the work of others 
, our findings support previous hypotheses about role of the news media in perpetuating public beliefs about psychoactive substance use, and the potential role the media may have in drug use trends. We document that reporting on prescription opioids in the 2000s followed strikingly similar patterns as for drugs in previous decades, and we provide a quantitative argument supporting the often-repeated demand for journalistic standards for reporting on substance use issues.
The pattern of news media outlets reinforcing each other to arrive at a general held belief has been noted for stories about epidemics of other drugs of abuse 
, [27 [cited in 11]]
]. This phenomenon is termed intermedia convergence
by Reese and Danielian (1989), and has been recently noted to be accelerated in the current more rapid and internet-influenced news cycle 
. The early emergence of problems as reported by the news media about northern New England and Appalachia may in part explain the sustained focus on these regions (see 
), despite national level data suggesting a broader ranging problem 
. In a recent review of media portrayals of drug use, Talyor (2008) cites extensive sociological evidence to advance the theory that news reports have focused on “outsiders,” socially marginalized classes of people, often racial minorities, going as far back as the mid-1800s in the United States [see also: 19]
. In this paper we document that with prescription opioids in the current decade, the outsiders appear to be rural residents, with a particular focus on Appalachia and northern New England, as shown in .
We posit that there may be co-linear relationships between the medical use of a prescription opioid, medical consequence from the use and abuse of the product, the public attention devoted to the topic, and news media reports. Co-linearity does not prove causation, however, and our intention in conducting this research is to advance the dialogue about the potential interconnectedness of these seemingly independent phenomena. While news articles may contribute indirectly to illicit demand for short-acting prescription opioids, there are many complex factors that influence poisoning mortality and are poorly understood: improper marketing, lax prescribing standards, patient and health care worker errors, breaches of the pharmaceutical supply chain, the influence of social networks, individual physical comorbidities, intervention programs and fear of accessing emergency medical care.
Clinical experience leads us to believe that drug use and overdose have multiple influences at the individual, social and genetic levels. The same multi-factorial influences may make drug abuse stories attractive subjects for journalists, combined with the fundamental issues crossing sociodemographic lines. It has also been noted that “media portrayals of the consumption and effects of alcohol do not comprise a neutral, scientific reality but rather signify a network of social knowledge, value systems, and symbols,” with the same likely occurring for other substances 
. There may be public expectation for coverage, and while predictable, the media aren't necessarily acting beyond the pale of their professional responsibilities. Yet, there is little formal guidance to journalists for appropriate coverage of this sensitive topic. The editors of PLoS point out that “today's health reporters may have been covering crime last week and politics the week before,” 
and media reporting of medical news has long been criticized 
. At the intersection of crime, health and politics, news reporting about drug abuse reflects a complex interaction of social factors, with little guidance as to how these issues should be covered. A recent article by PLoS Medicine editors on news coverage of suicides in the military reminds us that “reporters have a choice about how they frame stories about mental illness” 
. While individual-level influence of news reporting cannot be deduced from surveillance-oriented studies, and require more intensive analyses 
, our results suggest the need for a thorough evaluation of this topic, with an eye towards ethical standards that promote responsible journalism 
. These standards could be analogous to guidelines for stories on suicide 
, where the role of the news media in copycat suicides has been well documented 
, although some researchers believe the link to be less than definitive 
. Furthermore, our study suggests that there may be opportunity to harness news reports as an early warning indicator of emerging drug problems, similar recent work in infectious disease surveillance 
In support of the long-standing need for journalistic standards, we identified certain articles that amounted to inadvertent endorsements of prescription drug abuse; while dangers associated with drug abuse were mentioned in later paragraphs of articles, the opening lines presented an enticing scenario. A 2000 Portland Press Herald
article opened with: “Euphoria envelopes your body in a warm, cozy hug. Problems dissolve. Limbs tingle. Life feels perfect. These are the sensations that drug addict Christopher Coughlin says he felt using OxyContin, a highly addictive opiate that is sweeping Maine, from the streets of South Portland to the rural communities of Washington County.” 
Strikingly, Breecher (1972) noted nearly identical language in a Newsweek
article from 1962, suggesting a longstanding cyclical pattern in the way news about drug abuse is presented: “You're in outer space. You're Superman. You're floating in air, seeing double, riding next to God. It's Kicksville. Are these the fantasies of narcotics addicts on a pop? No. More disturbingly, these hopped-up reactions are those of teenagers hooked on goofballs, model airplane glue, and cough medicine. Across the nation, police last week reported case after case of this alarming trend.” 
The similarity of language goes beyond semantics; it suggests a consistent mindset and approach to reporting about substance use that has not changed substantially in at least four decades, if not longer 
Our findings are consistent with findings from previous cycles of drug abuse concern. At the height of reporting on cocaine during the summer of 1986, Merriam (1989) found that 5% of space in national media were devoted to drug issues, starting at less than 1% in the start of that year 
. By using electronic news sources, we were able to expand the analytic space to include local and regional print news media. Reflecting the greater volume and diversification of news media outlets in the 2000s, we found that only 0.08% of news volume involved prescription opioids. While we cannot make a direct comparison between previous work and our analysis, new media will have to be considered in future analyses. Print news media articles serve as a measureable proxy for other forms of media that may also mention prescription opioids. In today's rich multimedia environment competing and synergistic influences of multiple sources of information need to be assessed systematically, and future research may well focus on blogs and other electronic news sources. An analysis of internet comments about news reports of adolescent suicides suggested the strong influence of the style of reporting and attitudes about mental health 
. Hawton and colleagues (1999) noted a short-term increase in patients presenting for intentional drug overdoses after the broadcast of a popular television drama depicting drug self-poisoning 
. Basic health psychology research is needed to guide further efforts in this area.
We compared print news media articles to unintentional and undetermined intent poisoning mortality because vital statistics are the only publicly available data source with information on substance use disorders that has national coverage and can be analyzed on a monthly basis. Data from vital statistics reports are known to have limitations 
, with particular fears of underreporting. Medical examiners and coroners in the United States do not have standardized assessment and attribution procedures for suspected drug poisoning deaths. In addition, acute poisonings due to patients' medication dosing errors with potent opioids would be classified the same as a fatal opioid poisonings from drug abuse under ICD-10 coding. It is unknown what proportion of the deaths observed were from each type of exposure (and what the overlap may be), but deaths from medical errors involving specific prescription opioid pain relievers have resulted in nationwide alerts from the Food and Drug Administration 
. In order to address this problem we conducted sensitivity analyses using different combinations of ICD-10 codes representing wider and narrower definitions, such as including mental health codes as has been suggested by others 
. While the absolute number of poisoning fatalities fluctuated with the different definitions, the proportional change appeared to be nearly constant over time, resulting in similar results in lag regression analysis. We do not present these results here for the sake of brevity.
Another possible explanation for part of the observed correlation is diagnostic suspicion bias, that medical examiners more carefully screened for opioid poisonings in deaths of uncertain cause, after they themselves were exposed the media reports about overdose, or concerns expressed in the medical literature. This type of response has not been noted for medical examiners and drug overdoses in the literature, but remains plausible. As our research team has described elsewhere 
, the proportion of unintentional poisoning deaths in which a narcotic was not specified decreased slightly in metropolitan areas between 1999 and 2003, and increased slightly in non-metropolitan areas. However it has been noted that voluntary reporting to the Food and Drug Administration's spontaneous adverse event reporting system increased following publicized problems with adverse events for pharmaceutical medications and devices 
. To what extent this introduces bias in our analysis is unclear.
While previous generations of researchers have examined the question of the news media and substance use reporting, much of the findings have been sequestered to the sociology literature and have focused on qualitative analysis of media content. Few studies have attempted to link media data with information on drug use trends in a quantitative manner with temporal and geographic specificity. Substance use disorders have become increasingly medicalized in the last two decades (i.e., the “chronic relapsing brain disease” model), and pain has been more systematically assessed and aggressively treated with outpatient pharmacotherapies. Simultaneously there has been a profusion of the types of media involved in reporting news. However, biomedical science research has paid limited attention to the social context in which drug use problems and pain are perceived on a community level and what role mass media play, especially electronic sources. Returning to Blendon and Young (1998), “In the future, if health professionals want to change the direction of Americans' beliefs on particular drug policies they will have to devote significant resources to gaining media attention for their views.” We agree with them, but feel that journalism standards would also be beneficial. Perhaps the next “drug epidemic” will benefit from a more balanced and public health approach to reporting, but it requires immediate interdisciplinary action.