Policy Analysis No. 864

Overdosing on Regulation: How Government Caused the Opioid Epidemic

By Jeffrey Miron, Greg Sollenberger, and Laura Nicolae
February 14, 2019

Opioid overdose deaths have risen dramatically in the United States over the past two decades. The standard explanation blames expanded prescribing and advertising of opioids beginning in the 1990s.

This “more prescribing, more deaths” explanation has spurred increased legal restrictions on opioid prescribing. Federal and state governments have enacted a variety of policies to curtail prescribing and doctor shopping, and the federal government has raided pain management facilities deemed to be overprescribing. Supporters believe these policies reduce the supply of prescription opioids and thereby decrease overdose deaths.

We find little support for this view. We instead suggest that the opioid epidemic has resulted from too many restrictions on prescribing, not too few. Rather than decreasing opioid overdose deaths, restrictions push users from prescription opioids toward diverted or illicit opioids, which increases the risk of overdose because consumers cannot easily assess drug potency or quality in underground markets. The implication of this “more restrictions, more deaths” explanation is that the United States should scale back restrictions on opioid prescribing, perhaps to the point of legalization.

Continue to full version


Opioid overdose deaths have risen dramatically in the United States over the past two decades (Figure 1). The standard explanation blames expanded prescribing and advertising of opioids beginning in the 1990s.

Figure 1
Unintentional opioid overdose deaths, 1999-2017

Sources: Centers for Disease Control, National Center for Health Statistics, “Multiple Cause of Death 1999-2017,” CDC WONDER online database.

This “more prescribing, more deaths” explanation has spurred increased legal restrictions on opioid prescribing in the United States. Most states have enacted Prescription Drug Monitoring Programs (PDMPs), which aim to curtail doctor shopping and over­prescribing, and many states have capped legal opioid prescription doses. The federal government now limits opioid production and raids pain management facilities deemed to be overprescribing. In October 2018, the federal government enacted legislation that increases monitoring of prescribers and grants funding for organizations and hospitals that attempt to reduce prescribing.1 Congress is also considering additional regulation, such as limiting initial opioid prescriptions to small doses and restricting prescription packaging sizes.2 Supporters believe these restrictions will reduce the supply of prescription opioids and thus decrease overdose deaths.

We suggest that the opioid overdose epidemic has resulted from too many restrictions on prescription opioids, not too few.3 The risk of overdose from the proper medical use of prescription opioids is low. Worse, restrictions on prescribing push users from prescription opioids toward diverted or illicit opioids, which increases the risk of overdose because consumers cannot easily assess drug potency or quality in underground markets. Since 2011, rapidly increasing deaths from heroin and synthetic opioids such as fentanyl have driven up the opioid overdose death rate despite reduced prescribing. Restrictions on prescribing also risk pain undertreatment, harming patient quality of life and driving some to suicide.4 The implication of this “more restrictions, more deaths” explanation is that the United States should scale back restrictions on opioid prescribing, perhaps to the point of legalization.

We acknowledge that the case for the “more restrictions, more deaths” explanation is not conclusive; for example, we cannot quantify how many opioid users transact in underground markets or assess the causal effect of specific policy restrictions. We suggest, however, that available evidence is far more consistent with the “more restrictions, more deaths” explanation than the standard view.

The paper proceeds as follows. We first outline the contrasting “more prescribing, more deaths” and “more restrictions, more deaths” explanations. We then review evidence that addresses these competing views of the opioid epidemic. In the final section, we discuss the policy implications of our findings, including the case for legalizing opioids.

More Prescribing, More Deaths

In 1999, the unintentional opioid overdose death rate in the United States was roughly two per 100,000 people; by 2017, it had increased to roughly 13 per 100,000.5 Through 2012, natural or semisynthetic opioids such as OxyContin and Vicodin accounted for more than half of these deaths. Since 2010, heroin and synthetic opioids such as fentanyl have accounted for a growing share, with nearly 80 percent attributed to these two drug categories in 2017.6 Figure 1 presents these data for 1999-2017.

The standard explanation argues that this dramatic rise in opioid overdose deaths resulted from an expansion of opioid prescribing that began in the 1990s. Doctors had previously prescribed opioids for short-term pain and for palliative care in terminally ill cancer patients, but generally not for chronic conditions (such as back pain, osteoarthritis, fibromyalgia, or headaches) due to fear of patient addiction or abuse.7

New research in the 1980s, however, suggested that long-term medical use of opioids posed little risk of addiction.8 This evidence, along with the concerns of some healthcare providers that physicians were undertreating pain, prompted medical boards, pain societies, and patient support groups to advocate opioid analgesic treatment of chronic noncancer pain. Pharmaceutical companies supported this change and argued that new slow-release opioids like OxyContin had particularly low risks of addiction.9

According to proponents of the “more prescriptions, more deaths” explanation, however, this early optimism about long-term opioid prescribing relied on limited and unpersuasive evidence. Proponents of this view argue that the expansion in opioid prescribing in the 1990s caused increased addiction, overdoses, and deaths. The implication of this view is that restrictions on prescribing can reduce these harms.

More Restrictions, More Deaths

The “more restrictions, more deaths” explanation for the opioid epidemic holds that users face greater risk of overdose when policy restricts legal access. The 1970 Controlled Substances Act (CSA) places all drugs into one of five schedules based on the Drug Enforcement Administration’s (DEA) assessment of each drug’s medical value relative to its potential for abuse.10 Schedule I drugs (e.g., heroin, marijuana, LSD) are not legally available under federal law.11 Schedule II-V drugs are available by prescription, subject to DEA restrictions and oversight. Unscheduled drugs, such as acetaminophen or ibuprofen, are available over the counter.

Opioids are exclusively available by prescription. Thus, while most opioids are legal to produce, distribute, and use within the CSA rules, they are not as freely available as standard legal goods. Doctors generally limit prescriptions due to medical norms and legal restrictions. Individuals whose demand for opioids exceeds these limits then seek opioids from diverted or illicit sources.

Diverted or illicit opioids are more dangerous than legally provided versions. Quality control is poor in underground markets because reliable suppliers cannot legally advertise their goods and because consumers cannot sue for damages due to faulty or mislabeled products.12 The underground drug trade incentivizes trafficking in high-purity products to facilitate evasion.13 Consumers cannot easily assess the purity of the products they consume, so they accidentally take high-dose drugs or versions laced with more potent opioids like fentanyl (30 times stronger than heroin).

Underground opioid markets are therefore more likely than legal markets to supply hyperpotent products, such as heroin or fentanyl, and synthetic “designer drugs” of uncertain potency and quality, such as the heroin substitute Krokodil.14 While potent opioids would likely exist in a legal market (e.g., high-proof spirits exist in the alcohol market), consumers are unlikely to mistake these for less potent versions. Thus, restrictions that push opioid consumption underground likely increase the risk of overdose.

Consumers of illicit or diverted products also face a higher risk of adverse drug interactions. Drugs obtained in underground markets do not come with warning labels, and users cannot discuss safe use with their physicians, making them more likely to combine opioids with alcohol or other medications that suppress respiration. Consumers in underground markets may also have a higher risk of overdose because they are less likely to consume drugs in familiar environments. Using drugs in familiar environments can reduce tolerance by inducing an anticipatory response.15

The “more restrictions, more deaths” explanation thus suggests that, beginning in the 1990s, doctors began prescribing opioids to an increasing number of patients. This increased the number of individuals who demanded opioids for longer than the duration of their prescriptions, whether for recreational use or because of ongoing pain or physical dependence. When their prescriptions ended, many of these patients turned to diverted or illicit opioids, which generated more overdoses due to the greater risks of underground use. According to this view, loosening restrictions on opioid prescribing would lower the opioid overdose rate.

A complementary hypothesis is that overdoses have occurred not only from patients cut off from a prescription supply but also from individuals who consumed diverted opioids for recreation or self-medication. Increasing restrictions on the legal supply of opioids during the 1990s and 2000s pushed these individuals further into the black market and spurred more uncertainty about the quality and potency of the diverted or illicit opioids they consumed.16

Evidence Against the Standard Explanation

The standard explanation for the opioid epidemic rests on three claims: that long-term opioid use generates addiction; that long-term opioid use or addiction generates overdoses; and that overdoses have risen in sync with opioid prescribing over the past 20 years. We address each of these claims.

Long-Term Use and Addiction

The claim that long-term medical use generates addiction is the opposite of the consensus that began to emerge in the 1980s, which held that long-term medical use rarely generates addiction. Proponents of the standard explanation argue that, in coming to this more benign view of opioids, physicians and pharmaceutical companies relied excessively on a 1980 letter to the editor of The New England Journal of Medicine, which stated:

Although there were 11,882 patients who received at least one narcotic preparation, there were only four cases of reasonably well documented addiction in patients who had no history of addiction. . . . We conclude that despite widespread use of narcotic drugs in hospitals, the development of addiction is rare in medical patients with no history of addiction.17

Advocates of the standard view argue that this letter provided insufficient grounds for the conclusion that long-term opioid use poses little risk of addiction, noting the letter’s brevity and limited peer review.18

Studies since the 1980s, however, have consistently found only a modest risk of addiction or dependence from the medical use of opioids, in the range of 0 to 5 percent.19 A 2018 study of more than 568,000 patients receiving opioids between 2008 and 2016 found a “misuse” rate of 0.6 percent.20 Reports of opioid abuse did not increase in the 1990s despite increased medical use.21 A substantial fraction of patients who exhibit addiction after medical use have a preexisting psychological disease or history of addiction.22 According to the medical handbook Clinical Drug Data, “addiction does not occur when these drugs are used for legitimate painful conditions.”23

Long-Term Use and Overdose

The claim that long-term opioid use or addiction generates more overdoses is not supported by the evidence: long-term opioid use has minimal life-threatening consequences under appropriate medical guidance.24 As long as “escalations in opioids are carefully titrated on the basis of appropriate control of symptoms … concerns that death will be hastened by opioids are unwarranted.”25 Patients receiving long-term stable doses of an opioid rarely suffer from respiratory depression because they quickly develop tolerance to the drug.26 Respiratory depression is more likely to occur as a result of consumption from the underground market, when doses are more likely to be taken without regard to the drug’s half-life or combined with other drugs.27

As a crude measure of opioid risk, consider that in 2017 American physicians wrote nearly 200 million prescriptions for opioid pain relievers.28 The Substance Abuse and Mental Health Services Administration (SAMHSA) estimates that in 2017, nearly 87 million noninstitutionalized adults in the United States had used prescription pain relievers in the past year.29 The number of unintentional non­heroin or synthetic opioid overdoses was about 9,000, or 0.01 percent of the population taking prescription opioids.30 For comparison, a study analyzing the nonopioid antipsychotic drug Clozapine found a sudden death rate of 0.71 percent for those treated with the drug in the sample.31 The overall mortality rate for prescription opioids is comparable to the fatality risk of one year of daily aspirin use.32

Trends in Prescribing and Overdoses

The claim that opioid prescribing and unintentional opioid overdose deaths have risen concurrently over the past two decades is also subject to important caveats. First, the increasing trend in prescription opioid overdose deaths over the past several decades, during which prescribing generally increased, is likely overstated. Second, trends in opioid prescribing and the overdose death rate have recently diverged as prescribing has decreased, while deaths caused by heroin and synthetic opioids have accelerated. This suggests that prescribing is not the main driver of opioid overdoses and supports the “more restrictions, more deaths” explanation.

Death statistics may overstate the actual prevalence of prescription opioid overdoses due to errors in cause-of-death determination. Medical examiners and coroners generally classify drug-related deaths based on the results of forensic toxicology screens. Higher levels of opioid prescribing from the 1990s to 2010 may have increased the number of opioid-positive toxicology screens because the share of people using prescription opioids increased. This made it more likely that the screens would detect high prescription opioid concentrations in a person’s bloodstream at the time of death, regardless of the actual cause. A high concentration of opioids at the time of death does not by itself imply that overdose was the cause of death, since the lethal concentration level depends on a person’s tolerance, rate of drug metabolism, severity of chronic pain, and other factors.33 Thus, a higher rate of opioid-positive toxicology screenings is not indicative of an increase in prescription opioid overdose deaths.

If determining the cause of death were an exact science, the higher frequency of opioid-positive screens due to increased opioid prescribing over the past several decades would not affect reported cause-of-death statistics. In practice, cause-of-death determinations are subject to significant error, and the increased rate of prescription opioid detection by forensic toxicology screens could mechanically increase the number of reported overdose deaths.34 Toxicology screens of drug-poisoning decedents frequently reveal multiple drugs or alcohol, making it difficult to ascertain the true cause of death.35 Medical examiners and coroners tend to classify deaths caused by a combination of several different drugs as opioid overdose deaths as long as opioids are present in concentrations considered to be above the lethal level.36 As a result, death certificates may overstate the actual number of prescription opioid overdoses.

In addition, it is sometimes difficult for medical examiners or coroners to distinguish between deaths caused by prescription and illicit opioids. Death certificates often misclassify heroin-overdose deaths as morphine related because medical examiners rarely identify deaths as heroin related without the presence of a metabolite that is unique to heroin but rapidly metabolizes into morphine.37 In 2016, the CDC reported that the growing practice of mixing illicit fentanyl with counterfeit opioid pills has likely increased the misclassification of fentanyl deaths as prescription overdose deaths.38 As a result, overdose statistics may overstate the risks of prescription opioids and obscure the increasing mortality of illicit opioid use, inflating the increasing trend in overdose deaths from prescription opioids.

The claim that the sharp increase in opioid overdose deaths between 1999 and 2010 was caused by increased prescribing during this period is also inconsistent with evidence that prescription opioid addiction rates did not increase. Survey data find that the non­medical use of pain relievers remained stable or declined over the 2002-2010 period (Figure 2a).39 Similarly, recreational use of OxyContin, Vicodin, and narcotics other than heroin among high school seniors decreased slightly (Figure 2b).40 The decline in nonmedical use of pain relievers at the same time that opioid prescribing was increasing suggests that the increase in opioid prescribing did not cause a significant increase in opioid addiction and that the reported number of overdose deaths from prescription opioids may be overstated.

Figure 2a
Past month nonmedical use of pain relievers by age group, 2002-2017

Sources: Substance Abuse and Mental Health Services Administration (SAMHSA), “National Survey on Drug Use and Health,” 2014, 2015, and 2017.

Figure 2b
Annual prevalence of drug use for 12th graders

Sources: Lloyd D. Johnston et al., Monitoring the Future National Survey Results on Drug Use: 1975-2017, Volume 1: Secondary School Students (Ann Arbor: Institute for Social Research, University of Michigan, 2017).

Finally, the claim that the recent increase in opioid overdose deaths reflects the dangers of prescribing is contradicted by the recent surge in overdoses caused by heroin and synthetic opioids such as fentanyl. Figure 3 displays these data for the 1999-2015 sample. The correlation between opioid prescribing and unintentional opioid overdose deaths is positive through 2010 but weakens significantly and becomes negative afterward. Whereas opioid prescribing began declining in 2011, unintentional opioid overdoses continued to climb at a faster rate. Although deaths from prescription opioids declined, deaths from heroin and synthetic opioids such as fentanyl increased. In 2017, heroin and synthetic opioids accounted for more than three-fourths of all opioid overdose deaths.41

Figure 3
Unintentional opioid overdose deaths vs. legal opioid consumption, 1999-2015

Sources: Centers for Disease Control, National Center for Health Statistics, “Multiple Cause of Death 1999-2017,” CDC WONDER online database; “United States: Opioid Consumption in Morphine Equivalence mg per person,” Pain and Policy Studies Group, 2015, pdf.

Evidence for the Alternative Explanation

The “more restrictions, more deaths” view posits that opioid overdoses result mainly from restrictions on opioid access, which push consumers to higher-potency products and hamper their ability to determine the potency or quality of the drugs they consume. This view is supported by evidence that restrictions on opioid prescribing over the past decade may have pushed opioid users to the underground market, increasing the harms associated with illicit drug use. At a minimum, increasing regulation of opioid prescribing has failed to decrease opioid overdose mortality over the past several years, weakening the case for additional regulations. We suggest that deregulation of opioid prescribing may decrease the harms of illicit drug use and promote other benefits to public health and safety.

Federal law has limited opioid access for over a century. The Harrison Narcotics Tax Act of 1914 first regulated and taxed the production, importation, and distribution of opiates, laying the groundwork for a regulatory regime that gradually morphed into prohibition.42 In 1951, the Durham-Humphrey Amendment to the Food, Drug, and Cosmetic Act created a mandatory distinction between drugs that could be purchased over the counter and those that required a prescription.43 The 1970 Controlled Substances Act placed all federally regulated drugs in one of five schedules, and in 1986, the Anti-Drug Abuse Act established criminal penalties for possession of controlled substances and mandatory minimum sentences for offenses involving heroin, fentanyl, and other drugs.44

More recently, restrictions on the legal supply of opioids have limited access to opioid-dependence treatment and may have pushed users to underground markets. For example, the federal government restricts prescribing of maintenance treatment of opioid dependence with drugs such as methadone and buprenorphine, a Schedule III opioid partial agonist often used to treat dependence. While the 2000 Drug Addiction Treatment Act partially liberalized controls on maintenance treatment by allowing qualifying physicians to prescribe and dispense buprenorphine, access to this treatment remains highly restricted. The Act limited the number of patients a physician can treat at one time and imposed substantial regulation on participating physicians, such as training requirements, DEA oversight and onsite inspections, and sometimes an additional fee.45 Only 5 percent of physicians are licensed to prescribe buprenorphine, and few licensed prescribers treat the maximum permitted number of patients.46 Surveys of physicians indicate that the main impediments to buprenorphine prescribing include a lack of knowledge about how to acquire a DEA license and fear of buprenorphine diversion.47 The scarcity of buprenorphine treatment may have pushed opioid users to underground markets.

In 2001, methadone oversight shifted from the Food and Drug Administration (FDA) to SAMHSA, which required that methadone treatment programs for opioid dependence undergo a lengthy peer review accreditation process.48 The number of facilities dispensing methadone in opioid treatment programs dropped by about 20 percent after 2001 and did not rebound for four years.49 Furthermore, the Federal Opioid Treatment Standards published in 2015 mandated in-clinic opioid maintenance treatment for most patients and reserved maintenance treatment exclusively for patients who have been addicted to opioids for at least one year, further reducing access to treatment.50

Fearing drug diversion, some states have enacted moratoria on establishing methadone clinics.51 West Virginia, for example, placed a moratorium on methadone clinics in 2007, and Georgia placed a one-year moratorium on granting new licenses to opioid treatment clinics in 2016.52 The annual opioid overdose death rate in West Virginia has generally continued to increase since 2007, and the death rate has continued to increase in Georgia since 2016.

Restrictions on opioid maintenance treatment contradict decades of literature showing that maintenance treatment mitigates heroin and opioid dependence, reduces drug overdose deaths, and generally decreases the mortality rate of opioid use.53 Medically assisted treatments using substances such as buprenorphine and methadone are associated with reductions in opioid-related mortality and illicit drug use.54 Access to methadone treatment is associated with fewer deaths. The U.S. Code of Federal Regulations has also acknowledged that the use of methadone “has been shown to be an effective part of a total treatment effort in the management and rehabilitation of selected narcotic addicts.”55

Federal and state policies have also increasingly regulated prescription opioids, contributing to a decline in opioid prescribing starting in 2011 (Figure 3). This may have exacerbated heroin mortality and the undertreatment of pain. For example, in 2007, Congress granted the FDA authority to require drug manufacturers to conduct postmarketing safety studies and develop Risk Evaluation and Mitigation Strategies (REMS) for a prescription drug’s approval, increasing the regulatory costs of drug distribution and potentially limiting patient access to important therapies.56 In 2012, the FDA approved a REMS program for extended-release and long-action opioids that included new product labeling and required manufacturers to offer opioid training programs for prescribers on a voluntary basis.57 Since 2013, the FDA has gradually introduced additional postmarketing requirements for opioid products.58 States have also increas­ingly regulated legal opioid access; by the end of 2017, 26 states had imposed mandatory limits on prescribing for acute pain.59

Federal and state crackdowns on “pill mills,” networks of doctors and pain clinics that prescribe high quantities of opioids, have further reduced the availability of prescription opioids.60 In 2011, for example, Florida banned pain management clinics from dispensing drugs and required extensive medical examinations before and after prescribing opioids for chronic pain.61 In 2017, the Department of Justice created the Opioid Fraud and Abuse Detection Unit to increase monitoring of physicians and pharmacies deemed to be dispensing “disproportionately large amounts of opioids.”62 National media coverage of these crackdowns has increased physician fears of disciplinary action or prosecution, reducing opioid prescribing.63 Surveys indicate that these regulations have decreased physicians’ willingness to prescribe opioids, potentially causing them to undertreat pain.64

In the face of public pressure, in 2010 Purdue Pharmaceuticals introduced a reformulated, abuse-deterrent version of OxyContin. This reformulation made the drug less appealing to opioid abusers and caused many to substitute to heroin, an easily available and cheaper substitute. The reformulation of OxyContin led to an increase in heroin-overdose rates.65

In 2015, Endo Pharmaceuticals introduced an abuse-deterrent reformulation of the extended-release hydromorphone, Opana ER. The reformulation caused users to inject the drug and was linked to at least one HIV outbreak of roughly 190 people, the largest outbreak in Indiana history.66

State Prescription Drug Monitoring Programs (PDMPs) have also been implemented to reduce opioid access by preventing doctor shopping and reducing “excess” prescribing.67 These programs require doctors who prescribe opioids (and other controlled substances) to enter this information in a database that allows or requires other prescribers to check a patient’s history before writing prescriptions.68

Studies of PDMP effectiveness generally find that the programs modestly reduce prescribing and prescription opioid deaths and find an ambiguous or positive association between PDMPs and increased deaths from nonprescription opioids such as her­oin.69 However, recent work suggests that the reported effect of PDMPs on opioid-related harms is highly sensitive to the dates chosen to represent the start of PDMP implementation, which are not consistent across studies.70 The existing literature also largely relies on data sources for PDMP implementation dates that do not report detailed information regarding how the dates were determined.71 We leave this as an area for future work.

In October 2018, Congress passed the SUPPORT Act (Substance Use Disorder Prevention that Promotes Opioid Recovery and Treatment for Patients and Communities), which increases federal assistance for state PDMPs, expands access to opioid use disorder treatment, expands efforts to identify overprescribing, and grants funding to hospitals that limit the use of prescription opioids.72 The SUPPORT Act’s expansion of access to opioid use disorder treatment is a positive step toward decreasing opioid overdoses. However, the Act’s incentives for hospitals to limit prescribing may push users to consuming illicit opioids, increasing mortality and other risks to public health, such as higher HIV infection rates. Increased federal monitoring of prescribers could also contribute to the undertreatment of pain by exacerbating fear of regulatory sanctions.

While government and public pressure to reduce opioid prescribing may have reduced prescription overdoses (Figure 3), available evidence suggests that the decline in opioid prescribing caused by regulation of the prescription opioid supply has fueled the acceleration in heroin and fentanyl deaths since 2011. While abuse of prescription opioids declined beginning in 2010, the rate of heroin abuse sharply increased between 2008 and 2014.73 In 2013, the share of heroin users who had abused or were dependent on opioid analgesics was more than double the share in 2002.74

In 2015, the DEA reported that the declining availability of prescription opioids compared to heroin and the reformulation of OxyContin had contributed to the accelerating rate of prescription opioid abusers switching to heroin since 2010.75 Many young heroin users state that they transitioned from using (usually diverted) prescription opioids to heroin when prescription opioids became difficult to acquire due to decreased physician willingness to prescribe and increased police monitoring of pill markets.76

Proponents of opioid-prescribing regulations argue that while decreased prescribing may harm people who switch to more dangerous drugs like heroin, it will also reduce the creation of new addicts by limiting exposure to opioids in the first place. However, so long as heroin is illegal, the overdose risk of increased heroin use far outweighs that of prescription opioids. In 2017, roughly 10 times more people had used nonheroin opioids than heroin in the past year, yet the numbers of overdoses from heroin and nonheroin opioids were approximately equal.77

Furthermore, it is likely that a substantially smaller share of prescription opioid users would eventually transition to using heroin if prescription opioids were legal. Prescription users who switch to heroin are primarily driven to do so by heroin’s greater availability or lower price.78 Greater access to prescription opioids would decrease the incentive to switch to heroin.

Furthermore, concerns about creating new addicts should not dissuade doctors from prescribing opioids as medically indicated. As previously discussed, proper medical use of opioids carries little risk of addiction or overdose. Most people who abuse opioids after being exposed to them through a physician’s prescription already have a history of psychoactive drug use, and nearly three-quarters of people who misuse prescription pain relievers obtain them from sources other than their doctors, such as friends or relatives.79 Interviews of heroin users indicate that of those who first initiated regular opioid use with prescription opioids, most used diverted opioids from friends or family.80 Proper medical treatment of pain is unlikely to cause opioid dependence or heroin use, and undertreating pain harms patients’ quality of life and has led to a number of suicides.81

Evidence from other countries also suggests that increased legal access to opioids reduces deaths and improves health outcomes, such as improved treatment for opioid dependence and lower HIV infection rates. In 1995, France allowed physicians to prescribe buprenorphine for maintenance treatment without patient caps or special licensing requirements, leading to a fivefold reduction in heroin deaths and an estimated 3,900 lives saved.82 In the 1970s, Hong Kong expanded access to medically assisted treatments for opioid dependence, leading to a rapid decrease in HIV infection rates.83

Expansion of legal access to drugs in other countries has also been associated with declines in overdoses and deaths. In 2001, when Portugal decriminalized all drugs, it had the highest overdose rate in Western Europe. Drug-related deaths and HIV diagnoses attributed to injecting declined substantially, and Portugal now has the lowest overdose rate in Europe.84 In 2009, the Czech government decriminalized a variety of drugs, including heroin. By 2015, the number of drug-induced deaths in the Czech Republic had declined by about 20 percent from its peak in 2010.85 In the United Kingdom, Germany, Switzerland, and Canada, physicians can prescribe heroin for the treatment of severe dependence on heroin and other opioids.86 Heroin-assisted treatment has been associated with consistently positive therapeutic outcomes and reduced illicit heroin use in these countries.87

Safe injection rooms, which allow users of opioids and other drugs to access clean needles in a supervised and controlled setting, have become prevalent in cities across Europe and Canada.88 The sites reduce the use of contaminated needles and the pressure to consume drugs in a solitary or unfamiliar setting. Opponents fear such sites will increase drug use, but little evidence supports this fear.89 In fact, safe injection sites are associated with “lower overdose mortality … 67% fewer ambulance calls for treating overdoses, and a decrease in HIV infections.”90

Summary and Policy Implications

The standard view of the opioid epidemic argues that increased prescribing caused the recent increase in opioid overdose deaths. Medical use of opioids, however, is not a major cause of opioid addiction or overdose. Instead, available evidence suggests that the array of recent state and federal restrictions on legal access to opioids likely contributed to increasing overdoses by pushing users to diverted or illicit sources. Over the past few years, the opioid epidemic has accelerated due to overdoses caused by heroin and synthetic drugs such as fentanyl, despite reduced prescribing. Further restrictions on prescribing are unlikely to decrease overdose deaths.

A simple first step in decreasing the risks associated with the consumption of opioids from diverted or illicit sources is to increase legal access. For example, the federal government could end or decrease the regulation of methadone, buprenorphine, and even morphine- or heroin-maintenance treatment for opioid dependence and remove rules that limit prescribing or increase the costs of opioid production. Federal and state governments could also end raids on pill mills. These reforms could increase access to opioid dependence treatment, prevent the undertreatment of pain, and reduce the harms associated with underground consumption.

The United States could consider making all opioids “more legal” by shifting opioids to less regulated schedules or even over-the-counter status. In the extreme case, opioids would be legally available for purchase without a prescription. While modest reforms to regulation can decrease the prevalence of underground opioid consumption, outright legalization would eliminate the underground market entirely. Individuals who choose to purchase and consume opioids would be able to do so in a safer setting, reducing the dangers of use. We suggest this would counteract the recent increase in opioid overdose deaths.

Beyond any implications for overdose deaths, restrictions on legal access to opioids should be assessed in light of all their costs and benefits. Even if increased opioid prescribing heightens the frequency of opioid dependence, prescribing also improves the quality of life of patients who suffer from severe or chronic pain. Decreased prescribing in recent years, for example, has apparently driven at least 23 patients to suicide.91 We have focused here on overdose deaths in particular, but we emphasize that a complete analysis of restrictions on prescribing almost certainly suggests that the harms of regulation outweigh the risk of increasing opioid dependence through greater legal access.


1. SUPPORT for Patients and Communities Act, H.R. 6, 115th Cong. (2017-2018).

2. Margot Sanger-Katz and Thomas Kaplan, “Congress Is Writing Lots of Opioid Bills. But Which Ones Will Actually Help?,” New York Times, June 20, 2018.

3. Previous authors have raised many of the points we raise in this paper. We attempt to synthesize these analyses and provide additional evidence. See also Jeffrey A. Singer, “The Drug Prohibition Is to Blame for the Opioid Crisis,” Cato Institute, Commentary, December 4, 2018; Jacob Sullum, “Opioid-Related Deaths Keep Rising as Pain Pill Prescriptions Fall,” Reason, November 29, 2018; J. J. Rich, “The Opioid Fix That Wasn’t,” Reason, October 26, 2018; and Mark Edmund Rose, “Are Prescription Opioids Driving the Opioid Crisis? Assumptions vs. Facts,” Pain Medicine 19, no. 4 (April 2018): 793-807.

4. Stefan G. Kertesz, Adam J. Gordon, and Sally L. Satel, “Opioid Prescription Control: When the Corrective Goes Too Far,” Health Affairs (blog), January 19, 2018.

5. CDC.gov, National Center for Health Statistics, “Multiple Cause of Death 1999-2017,” CDC WONDER online database. Data are from the Multiple Cause of Death Files, 1999-2017, as compiled from data provided by the 57 vital statistics jurisdictions through the Vital Statistics Cooperative Program.

6. CDC.gov, National Center for Health Statistics, “Multiple Cause of Death 1999-2017.”

7. Roger Chou et al., “Clinical Guidelines for the Use of Chronic Opioid Therapy in Chronic Noncancer Pain,” Journal of Pain 10, no. 2 (2009): 113-30.

8. See, for instance, Russell K. Portenoy and Kathleen M. Foley, “Chronic Use of Opioid Analgesics in Nonmalignant Pain: Report of 38 Cases,” Pain 25, no. 2 (1986): 171-86.

9. Art Van Zee, “The Promotion and Marketing of OxyContin: Commercial Triumph, Public Health Tragedy,” American Journal of Public Health 99, no. 2 (2009): 221-27.

10. DEA.gov, Drug Information, “Drug Scheduling.”

11. An exception to this restriction is made for persons registered with or authorized by the DEA to conduct medical research, chemical analysis, or instructional activities. See DEA.gov, Diversion Control Division, “Title 21 of the Code of Federal Regulations, Part 1301-Registration of Manufacturers, Distributors, and Dispensers of Controlled Substances.”

12. Jeffrey A. Miron and Jeffrey Zwiebel, “The Economic Case against Drug Prohibition,” Journal of Economic Perspectives 9, no. 4 (1995): 175-92.

13. Walter Block, “Drug Prohibition: A Legal and Economic Analysis,” Journal of Business Ethics 12 (1993): 689-700.

14. Global Information Network about Drugs (website), Drug Facts, “Designer Drugs.”

15. Shepard Siegel, “The Heroin Overdose Mystery,” Current Directions in Psychological Science 25, no. 6 (2016): 375-79.

16. Some evidence suggests that the second hypothesis is more realistic. Survey data indicate that nearly three-quarters of people who engaged in nonmedical use of prescription pain relievers in the past year obtained them from sources other than their doctors, such as friends or relatives. Interviews of heroin users also indicate that among those who first initiated regular opioid use with prescription opioids, most began with prescriptions diverted from friends or family. See Rachel N. Lipari and Arthur Hughes, “How People Obtain the Prescription Pain Relievers They Misuse,” CBHSQ Report (SAMHSA), January 12, 2017; S. G. Mars et al., “ ‘Every “Never” I Ever Said Came True’: Transitions from Opioid Pills to Heroin Injecting,” International Journal of Drug Policy 25, no. 2 (March 2014): 257-66; and S. E. Lankenau et al., “Initiation into Prescription Opioid Misuse amongst Young Injection Drug Users,” International Journal of Drug Policy 23, no. 1 (January 2012): 37-44.

17. Jane Porter and Hershel Jick, “Addiction Rare in Patients Treated with Narcotics,” New England Journal of Medicine 302, no. 2 (January 10, 1980): 123.

18. Harrison Jacobs, “This One-Paragraph Letter May Have Launched the Opioid Epidemic,” Business Insider, May 26, 2016.

19. See, for example, David A. Fishbain et al., “What Percentage of Chronic Nonmalignant Pain Patients Exposed to Chronic Opioid Analgesic Therapy Develop Abuse/Addiction and/or Aberrant Drug-Related Behaviors? A Structured Evidence-Based Review,” Pain Medicine 9, no. 4 (2007): 444-59.

20. G. A. Brat et al., “Postsurgical Prescriptions for Opioid Naïve Patients and Associations with Overdose and Misuse: Retrospective Cohort Study,” BMJ 360 (January 17, 2018).

21. David E. Joranson et al., “Trends in Medical Use and Abuse of Opioid Analgesics,” JAMA 283, no. 13 (2000): 1710-714.

22. See David A. Fishbain et al., “What Percentage of Chronic Nonmalignant Pain Patients Exposed to Chronic Opioid Analgesic Therapy Develop Abuse/Addiction and/or Aberrant Drug-Related Behaviors? A Structured Evidence-Based Review,” Pain Medicine 9, no. 4 (2007): 444-59.

23. Kelly M. Smith et al., Clinical Drug Data, 11th ed. (New York: McGraw-Hill, 2010), p. 52. Studies that report high rates of problematic drug behavior following medical use of opioids often rely on broad definitions of problematic drug use, such as a single incident of a dose violation or a lost prescription. These behaviors may be indicative of pain undertreatment, physical dependence, or recreational use of opioids, which are distinct from addiction. Such studies likely overestimate the prevalence of opioid use disorders. See Jette Høsted and Per Sjørgen, “Addiction to Opioids in Chronic Pain Patients: A Literature Review,” European Journal of Pain 11, no. 5 (July 2007): 490-51.

24. Howard S. Smith, Opioid Therapy in the 21st Century (New York: Oxford University Press, 2013) p. 90.

25. Russell K. Portenoy and Nessa Coyle, “Controversies in the Long-Term Management of Analgesic Therapy in Patients with Advanced Cancer,” Journal of Pain and Symptom Management 5, no. 5 (1990): 307-19.

26. Portenoy and Coyle, “Controversies in the Long-Term Management of Analgesic Therapy in Patients with Advanced Cancer.” See also Michael Zenz, Michael Strumpf, and Michael Tryba, “Long-Term Oral Opioid Therapy in Patients with Chronic Nonmalignant Pain,” Journal of Pain and Symptom Management 7, no. 2 (1992): 69-77.

27. Smith, Opioid Therapy in the 21st Century, p. 90. We do not suggest that opioid consumption is safe only under medical guidance; if opioids were deregulated or fully legalized, consumption from the legal market would likely be safer than underground use. However, evidence on this question is scarce due to the long history of prohibition.

28. CDC.gov, Opioid Overdose, “Prescription Opioids.”

29. SAMHSA, “Results from the 2017 National Survey on Drug Use and Health: Detailed Tables, Table 1.53A,” https://www.samhsa.gov/data/sites/default/files/cbhsq-reports/NSDUHDetailedTabs2017/NSDUHDetailedTabs2017.htm.

30. CDC.gov, National Center for Health Statistics, “Multiple Cause of Death, 1999-2016.”

31. Ilan Modai et al., “Sudden Death in Patients Receiving Clozapine Treatment: A Preliminary Investigation,” Journal of Clinical Psychopharmacology 20, no. 3 (2000): 325-27.

32. See Joshua T. Cohen and Peter J. Neumann, “What’s More Dangerous, Your Aspirin or Your Car? Thinking Rationally about Drug Risks (and Benefits),” Health Affairs 26, no. 3 (2007): 636-46. The authors find that daily aspirin use has a fatality risk of 10.4 per 100,000 person years, which translates to a 0.000104 fatality risk per year.

33. Lynn R. Webster and Nabarun Dasgupta, “Obtaining Adequate Data to Determine Causes of Opioid-Related Overdose Deaths,” Pain Medicine 12, Supp. 2 (2011): S86-92.

34. See, for example, Lauri McGivern et al., “Death Certification Errors and the Effect on Mortality Statistics,” Public Health Reports 132, no. 6 (2017): 669-75; and Ann E. Sehdev and Grover M. Hutchins, “Problems with Proper Completion and Accuracy of the Cause-of-Death Statement,” Archives of Internal Medicine 161, no. 2 (2001): 277-84.

35. See, for example, Haylea A. Hannah et al., “Using Local Toxicology Data for Drug Overdose Mortality Surveillance,” Online Journal of Public Health Informatics 9, no. 1 (2017): e143; and Lynn R. Webster et al., “An Analysis of the Root Causes for Opioid-Related Overdose Deaths in the United States,” Pain Medicine 12, Supp. 2 (2011): S26-35.

36. According to Dr. Steven Karch, a forensic pathologist, medical examiners may wrongly classify deaths as opioid overdoses on the basis of high opioid concentrations detected by toxicological screens. He notes that “there are plenty of people walking around with levels of opioids in their bodies that would be declared toxic if they were dead … in a medical examiner’s office,” emphasizing that toxicology reports can lead medical examiners to mischaracterize deaths as “overdoses.” See Radley Balko, “The New Panic over Prescription Painkillers,” Huffington Post, February 8, 2012.

37. DEA.gov, “National Heroin Threat Assessment Summary,” DEA Intelligence Report, April 2015.

38. CDC.gov, “Overdose Deaths involving Opioids, Cocaine, and Psychostimulants-United States, 2016-2016”; and DEA.gov, “National Heroin Threat Assessment Summary—Updated,” DEA Intelligence Report, June 2016.

39. SAMHSA, “Behavioral Health Trends in the United States: Results from the 2014 National Survey on Drug Use and Health,” https://www.samhsa.gov/data/sites/default/files/NSDUH-FRR1-2014/NSDUH-FRR1-2014.pdf; and SAMHSA, “Behavioral Health Trends in the United States: Results from the 2015 National Survey on Drug Use and Health,” 2016, https://www.samhsa.gov/data/sites/default/files/NSDUH-DetTabs-2016/NSDUH-DetTabs-2016.pdf. In 2015, SAMHSA changed the wording of its NSDUH survey to measure past-year “misuse” of prescription pain relievers, rather than “nonmedical use.” As a result, NSDUH estimates from 2015 or later may not be directly comparable to estimates from before 2015. The new questionnaire asks respondents whether they used a prescription drug “in any way a doctor did not direct you to use it,” which likely results in higher estimates than the number of estimated “nonmedical” users, because it may include respondents who use prescription pain relievers to self-medicate for pain. In this case, the change in wording would imply an even more substantial decline of nonmedical use of prescription pain relievers in 2015 and 2016. See SAMHSA, “Results from the 2016 National Survey on Drug Use and Health: Detailed Tables,” 2017, p. 8, https://www.samhsa.gov/data/sites/default/files/NSDUH-DetTabs-2016/NSDUH-DetTabs-2016.pdf.

40. Lloyd D. Johnston et al., Monitoring the Future National Survey Results on Drug Use: 1975-2017: Overview 2017, “Key Findings on Adolescent Drug Use,” Table 2, “Trends in Annual Prevalence of the Use of Various Drugs in Grades 8, 10, and 12” (Ann Arbor: Institute for Social Research, University of Michigan, 2018), http://monitoringthefuture.org/data/17data/17drtbl2.pdf.

41. It is not clear that the increase in prescription opioid overdose deaths from 1990 to 2010 and the increase in heroin and fentanyl deaths since 2010 have increased drug-related harm as a whole. In proportional terms, the increasing trend in drug overdoses has remained essentially constant since 1968, although the composition of overdoses caused by drug type fluctuated substantially. See Jalal et al., “Changing Dynamics of the Drug Overdose Epidemic in the United States from 1979 through 2016,” Science 361, no. 6408 (September 21, 2018): 1184. The recent rise in fentanyl deaths may also reflect increased testing for fentanyl in toxicological screenings as fentanyl use has become more prevalent. The data may, in part, increasingly reveal an existing fentanyl problem rather than a rapid emergence of fentanyl overdoses in the past several years.

42. National Alliance of Advocates for Buprenorphine Treatment, “Harrison Narcotics Tax Act,” 1914, https://www.naabt.org/documents/Harrison_Narcotics_Tax_Act_1914.pdf.

43. Prior to this law, a recommendation as to whether a drug should be taken only under the supervision of a physician was given only by the drug’s manufacturer. See John P. Swann, “FDA and the Practice of Pharmacy: Prescription Drug Regulation before the Durham-Humphrey Amendment of 1951,” Pharmacy in History 32, no. 2 (1994): 55-70.

44. DEA.gov, “The Controlled Substances Act”; and The Anti-Drug Abuse Act of 1986, H.R. 5484, 99th Cong. (1986).

45. SAMHSA, “Buprenorphine Training for Physicians,” 2018, https://www.samhsa.gov/medication-assisted-treatment/training-resources/buprenorphine-physician-training. Initially, providers who obtained a waiver under the Drug Addiction Treatment Act (DATA) to prescribe buprenorphine could apply to increase their cap from 30 to 100 patients after a year of experience and a lengthy application process, although a majority of providers have not done so. As of 2018, 9,777, or 19.5 percent, of DATA-waived physicians were granted this increase. Under the Comprehensive Addiction and Recovery Act of 2016, physicians can apply to increase their maximum cap to 275 patients after two years of experience. As of 2018, 4,161, or 8.1 percent, of DATA-waived physicians received this clearance; 72.4 percent of DATA-waived physicians are still allowed to treat only 30 patients at a time. See SAMHSA, “Physician and Program Data,” https://www.samhsa.gov/programs-campaigns/medication-assisted-treatment/training-materials-resources/physician-program-data.

46. Judy George, “Why Do So Few Docs Have Buprenorphine Waivers?,” MedPage Today, February 18, 2016; and Andrew S. Huhn and Kelly E. Dunn, “Why Aren’t Physicians Prescribing More Buprenorphine?,” Journal of Substance Abuse Treatment 78 (July 2017): 1-7.

47. Huhn and Dunn, “Why Aren’t Physicians Prescribing More Buprenorphine?”

48. Luc R. Pelletier and Jeffrey A. Hoffman, “New Federal Regulations for Improving Quality in Opioid Treatment Programs,” Journal for Healthcare Quality 23, no. 6 (2001): 29-34.

49. SAMHSA, National Survey of Substance Abuse Treatment Services, https://wwwdasis.samhsa.gov/dasis2/nssats.htm.

50. SAMHSA, “Federal Guidelines for Opioid Treatment Programs,” January 2015, https://store.samhsa.gov/system/files/pep15-fedguideotp.pdf.

51. Christine Vestal, “In Opioid Epidemic, Prejudice Persists against Methadone,” Stateline, PewTrusts.org., November 11, 2016.

52. Christine Vestal, “Still Not Enough Treatment in the Heart of the Opioid Crisis,” Stateline, PewTrusts.org., September 26, 2016; and “Despite Overdose Epidemic, Georgia Caps the Number of Opioid Treatment Clinics,” Morning Edition, NPR, June 15, 2016.

53. See, for example, Martin Connock et al., “Methadone and Buprenorphine for the Management of Opioid Dependence: A Systematic Review and Economic Evaluation,” Health Technology Assessment 11, no. 9 (March 2007): 1-171.

54. SAMHSA, “Buprenorphine Waiver Management,” https://www.samhsa.gov/programs-campaigns/medication-assisted-treatment/training-materials-resources/buprenorphine-waiver; and Marc R. LaRochelle et al., “Medication for Opioid Use Disorder after Nonfatal Opioid Overdose and Association with Mortality: A Cohort Study,” Annals of Internal Medicine 169, no. 3 (June 19, 2018): 137-45.

55. 21 U.S.C. § 291.501.

56. Stephen Barlas, “Pharmacy Groups Want to Change the FDA’s REMS Authority,” Pharmacy and Therapeutics 37, no. 1 (2012): 39-40.

57. FDA.gov, Information by Drug Class, “Timeline of Selected FDA Activities and Significant Events Addressing Opioid Misuse and Abuse.”

58. FDA.gov, Information by Drug Class, “New Safety Measures Announced for Extended-Release and Long-Acting Opioids.”

59. C. S. Davis et al., “Laws Limiting the Prescribing or Dispensing of Opioids for Acute Pain in the United States: A National Systematic Legal Review,” Drug and Alcohol Dependence 194 (January 1, 2019): 166-72.

60. CDC.gov, Office for State, Tribal, Local, and Territorial Support, “Menu of Pain Management Clinic Regulation.”

61. National Institute of Justice, Drug and Crime Research, “Florida Legislation Helps Reduce the Number of ‘Pill Mills,’ ” February 8, 2018.

62. Department of Justice, Office of Public Affairs, “Attorney General Sessions Announces Opioid Fraud and Abuse Detection Unit,” press release, August 2, 2017.

63. Aaron M. Gilson and David E. Joranson, “Controlled Substances and Pain Management,” Journal of Pain and Symptom Management 21, no. 3, 227-37.

64. Gilson and Joranson, “Controlled Substances and Pain Management.”

65. William N. Evans, Ethan Lieber, and Patrick Power, “How the Reformulation of OxyContin Ignited the Heroin Epidemic,” NBER Working Paper no. 24475, February 14, 2018.

66. Jeffrey A. Singer, “Abuse-Deterrent Opioids and the Law of Unintended Consequences,” Cato Institute Policy Analysis no. 832, February 6, 2018.

67. Randy A. Sansone and Lori A. Sansone, “Doctor Shopping: A Phenomenon of Many Themes,” Innovations in Clinical Neuroscience 9, no. 11-12 (2012): 42-46.

68. CDC.gov, Opioid Overdose, “What States Need to Know about Prescription Drug Monitoring Programs.”

69. Anne Schuchat et al., “Prescription Drug Monitoring Programs and Opioid Death Rates-Reply,” JAMA 318, no. 20 (November 28, 2017): 2045; and Y. H. Nam et al., “State Prescription Drug Monitoring Programs and Fatal Drug Overdoses,” American Journal of Managed Care 23, no. 5 (May 2017): 297-303.

70. Jill Horwitz et al., “The Problem of Data Quality in Analyses of Opioid Regulation: The Case of Prescription Drug Monitoring Programs,” NBER Working Paper no. 24947, August 2018.

71. Horwitz et al., “The Problem of Data Quality in Analyses of Opioid Regulation.”

72. SUPPORT for Patients and Communities Act, H.R. 6, 115th Cong. (2017-2018).

73. Cicero et al., “Shifting Patterns of Prescription Opioid and Heroin Abuse in the United States,” Addictive Behaviors 65 (February 2017): 242-44.

74. Pradip K. Muhuri, Joseph C. Gfroerer, M. Christine Davies, “Associations of Nonmedical Pain Reliever Use and Initiation of Heroin Use in the United States,” CBHSQ Data Review (SAMHSA), August 2013; and Jones et al., “Vital Signs: Demographic and Substance Use Trends among Heroin Users-United States, 2002-2013,” Morbidity and Mortality Weekly Report 64, no. 26 (July 2015): 719-25.

75. DEA.gov, “National Heroin Threat Assessment Summary,” DEA Intelligence Report, April 2015.

76. Mateu-Gelabert et al., “Injection and Sexual HIV/HCV Risk Behaviors Associated with Nonmedical Use of Prescription Opioids among Young Adults in New York City,” Journal of Substance Abuse Treatment 48, no. 1 (January 2015): 13-20; and Mars et al., “Every ‘Never’ I Ever Said Came True.”

77. In 2017, about 900,000 people aged 12 or older used heroin, compared to roughly 10 million people who misused nonheroin opioids in the past year. Roughly 15,000 people died of heroin overdoses and of nonheroin opioid overdoses in 2017. See SAMHSA, “Results from the 2017 National Survey on Drug Use and Health: Detailed Tables,” September 7, 2018, https://www.samhsa.gov/data/sites/default/files/cbhsq-reports/NSDUHDetailedTabs2017/NSDUHDetailedTabs2017.htm#tab1-27A; and CDC.gov, National Center for Health Statistics, “Multiple Cause of Death 1999-2016.”

78. DEA.gov, “National Heroin Threat Assessment Summary—Updated, June 2016.”

79. Rose, “Are Prescription Opioids Driving the Opioid Crisis? Assumptions vs. Facts”; Cicero, et al. “Psychoactive Substance Use Prior to the Development of Iatrogenic Opioid Use”; and Lipari and Hughes, “How People Obtain the Prescription Pain Relievers They Misuse.”

80. Mars et al., “Every ‘Never’ I Ever Said Came True”; and Lankenau et al., “Initiation into Prescription Opioid Misuse amongst Young Injection Drug Users.”

81. Kertesz et al., “Opioid Prescription Control: When the Corrective Goes Too Far.”

82. Maria Patriezia Carrieri et al., “Buprenorphine Use: The International Experience,” Clinical Infectious Diseases 43, Supp. 4 (2006): S197-S215.

83. Hong Kong Department of Health, iContinuing Education on HIV/AIDS, “The Success of Methadone Treatment Programme in Protecting Hong Kong from an HIV Epidemic among Drug Users,” October 21, 2010.

84. European Monitoring Centre for Drugs and Drug Addiction (website), “Portugal: Country Drug Report 2018”; and Jeffrey A. Singer, “No Let Up on the Bad News about Overdose Deaths,” Cato at Liberty (blog), November 29, 2018.

85. European Monitoring Centre for Drugs and Drug Addiction (website), “Czech Republic: Country Drug Report, 2017.”

86. Steve Rolles, “Heroin-Assisted Treatment in Switzerland: Successfully Regulating the Supply and Use of a High-Risk Injectable Drug,” Transform, May 2016, and Canadian Press, “Canada Now Allows Prescription Heroin in Severe Opioid Addiction,” CBC, September 8, 2016.

87. Beau Kilmer et al.,“Considering Heroin-Assisted Treatment and Supervised Drug Consumption Sites in the United States,” RAND Corporation Research Report, 2018.

88. Erin Schumaker, “To Fight the Opioid Crisis, Health Experts Recommend Safe Places to Shoot Up,” Huffington Post, October 31, 2017.

89. European Monitoring Centre for Drugs and Drug Addiction (website), “Drug Consumption Rooms: An Overview of Provision and Evidence,” 2018.

90. Jennifer Ng, Christy Sutherland, and Michael R. Kolber, “Does Evidence Support Supervised Injection Sites?,” Canadian Family Physician 63, no. 11 (2017): 886.

91. Kertesz et al., “Opioid Prescription Control: When the Corrective Goes Too Far.”

Jeffrey Miron is director of economic studies at the Cato Institute and director of graduate and undergraduate studies in the Department of Economics at Harvard University. Greg Sollenberger is a fellow of the Casualty Actuarial Society and a practicing actuary. Laura Nicolae is a student at Harvard College and research assistant in the Department of Economics at Harvard University.