One of the most important tools for addressing the opioid overdose epidemic in the US is a drug called naloxone. It’s a prescription medication that reverses the effects of an opioid overdose — kicking opioids off certain receptors in the brain and thereby bringing people back from the brink of death. Every state in the country has altered its laws in recent years to make naloxone easier to access.
But now two economists are arguing that ready access to naloxone can have dire unintended effects. In a working paper posted last week, Jennifer Doleac of the University of Virginia and Anita Mukherjee of the University of Wisconsin found that greater access to the drug increases opioid use, opioid-related crime, and, in some places, deaths from overdoses.
The authors maintain that despite those results, we shouldn’t limit access to this lifesaving drug. Still, many public health experts are worried that the analysis could be used to justify limits on access to naloxone. Some experts condemned the authors’ approach as reinforcing stigma by stereotyping people who use drugs as reckless and criminal. Others on social media went further and hurled vitriolic personal attacks at the authors.
The suggestion that naloxone might cause more harm than good is not new, but the paper is bringing the idea fresh mainstream attention. “The coldly logical response to [the study] would seem to be to discontinue naloxone use,” wrote the Washington Post columnist Megan McArdle (who did not ultimately endorse that “logical” conclusion, finding in it “something repulsive”).
Some of the criticism of the paper has been unfair. Several public health experts said it was irresponsible to publish such an inflammatory finding before the paper had been through peer review. But economists routinely make public unreviewed working papers, in part to improve them.
The paper displays methodological flaws — plus a failure to consider how people who use drugs think
Still, while Doleac and Mukherjee are accomplished researchers, I found the paper unpersuasive — drawing on my understanding of the drug addiction literature as well as my experience working with people who use drugs. The authors failed to consider other reasons naloxone access might increase emergency room visits and reported crimes, for instance, reasons besides encouraging extra-risky opioid use.
And their understanding of how people use naloxone contradicts what people who use drugs actually say about their own experiences.
In short, it would be unwarranted in the extreme to limit access to naloxone in any way.
The authors of the study were exploring a phenomenon that economists call “moral hazard.” The idea is that giving people a safety net, where danger exists, leads them to act in riskier ways. As an analogy, imagine installing a sprinkler system in an apartment building. The residents now consider themselves protected from the risks of a fire.
Someone concerned about moral hazard might predict that the residents might start lighting their apartments with candles or leaving the oven on to cook a turkey while they’re at work. If the residents’ pastimes get risky enough, that could offset the protective effect of the sprinkler system, leading to more fires than before, not fewer.
But don’t sprinklers save lives? Indeed, claims that moral hazard offsets the benefits of risk-reducing measures are controversial. In one study, for instance, a group of economists concluded that the availability of HIV treatment had caused people to have more sexual partners, leading to increases in HIV incidence. But more recent evidence that HIV treatment greatly reduces transmission calls those findings into question.
Doleac and Mukherjee argue naloxone serves as just this kind of safety net with unintended consequences. Is it?
First, it’s important to note that other work contradicts theirs. At least one study found no increase in self-reported drug use or addiction severity among people who use heroin after receiving overdose education and naloxone training. This was a small, interview-based study without a comparison group, but, unlike Doleac and Mukherjee’s analysis, it directly measured the effect of naloxone possession on drug use frequency and addiction severity.
How the controversial study was done
Doleac and Mukherjee generally look at the period from 2010 to 2015 (although when measuring emergency room admissions they go back to 2006). They focus on urban areas — since that’s where naloxone is most readily, and immediately, available.
For overdose deaths, they use data from the Centers for Disease Control and Prevention. A different data set covers ER visits in urban areas. The crime data largely focuses on 410 communities in 31 states, with a fuller analysis looking at 2,800 communities.
The authors don’t directly measure the effects of naloxone access on drug use behavior. Rather, they measure the effect of the implementation of naloxone access laws on outcomes they believe reflect drug use. This is tricky because, for instance, naloxone access laws can be passed in response to rising overdose deaths, making it seem like laws increase overdoses.
To account for this problem, the authors used statistical techniques commonly employed in economics to estimate causal relationships. One approach is to use changes in outcomes within states, rather than between states, to estimate the effects of laws.
The authors also looked closely at prelaw trends in emergency room visits, crime, and so on. That helps take into account the concern that states may pass naloxone laws when they see overdoses start to increase, which may also create the impression that the policy is causing overdoses.
The authors’ statistics are generally thorough, but there are several methodological problems. For one thing, there are different kinds of naloxone access laws, and several states have passed multiple naloxone access laws at separate times. There’s reason to think some types of laws, such as those allowing naloxone purchase without a prescription, do much more to increase access than others, such as those removing civil liability for prescribing naloxone. The authors don’t account for this, instead lumping all laws together.
What’s more, many of these laws were passed in 2015, the final year of analysis in the paper. That means the authors have extremely limited data on the all-important period following many of the laws’ implementation.
One problem: you’re supposed to go to the ER after being administered naloxone
They also make a fundamental error in their reasoning. The authors find that naloxone access laws lead to more opioid-related emergency department visits, the premise being that naloxone access laws increase opioid overdoses. But there’s a far more likely explanation: People are generally instructed to seek medical care for overdose after receiving naloxone.
Overdose is a general term to describe experiencing the toxic effects of drugs. People can overdose, and often do, without either dying or seeking medical attention. If people who would otherwise overdose without medical attention are instead using naloxone and going to emergency rooms, that’s a good thing.
Doleac and Mukherjee also find that naloxone access laws lead to more opioid-related crimes, arguing that people commit crimes to fund their riskier drug use. But arrests and reported crimes are often not a good measure of underlying criminal behavior. (Consider the much-cited fact that white and black people use and sell drugs at similar rates, yet drug arrests for black people are several times higher.)
A more likely reason for increases in “opioid-related” arrests is that possession of naloxone can be treated as a clear marker for people who use opioids, making them more visible targets for law enforcement. Drug policy organizations are aware of this potential problem and work to educate people who use drugs and law enforcement about the right to have naloxone.
The authors find that naloxone access laws appear to increase opioid-related theft but not overall theft, which casts doubt on their claims — and lends support to the competing theory that a greater proportion of thefts are being categorized as opioid-related, perhaps because people have naloxone on them.
“Naloxone parties” are an urban myth, yet the authors cite them to buttress their theory
But let’s step back from the data for a minute and consider the authors’ theory of how people who use opioids are acting. People who are physically dependent on opioids experience terrible withdrawal symptoms if they don’t use opioids: Imagine the worst flu symptoms you’ve ever had combined with anxiety and depression. An inescapable part of how naloxone works is that it puts people into instant withdrawal. It is hard to imagine people who use opioids planning to incorporate the horrors of withdrawal into their drug-using routine.
Of even more concern is that the authors cite in support of their arguments the existence of “naloxone parties,” at which people supposedly use copious quantities of opioids, expecting to overdose, and then rescue each other with naloxone. But these parties are an urban myth.
Getting high just to end the night in severe pain, anxiety, nausea, and vomiting is not most people’s idea of a party. Part of the confusion over naloxone parties comes from the fact that naloxone can be present, and should be present, at parties where people use opioids, in case someone overdoses unintentionally.
There’s a related problem with the idea that people who use opioids may take more risks because they know first responders have naloxone. As a drug policy advocate writing for the Fix puts it, “There are two things most drug users avoid at all costs: withdrawal and police. Overdosing and having first responders show up to administer Narcan summons both.” (Narcan is the most popular brand of naloxone.)
The fact that the proposed mechanism of the paper is not consistent with the experiences of people who use drugs and work with people who use drugs is important.
The authors find that on average, naloxone access laws are not associated with changes in opioid overdose deaths. That might be a bit disappointing, but it’s not altogether surprising. Successfully addressing the opioid epidemic requires a long-term, multifaceted response including criminal justice reform and treatment expansion.
Naloxone may not be causing a steep drop in overall deaths, but it is extending the lives of individuals, giving them more time with their loved ones and another chance to seek treatment.
The authors find that naloxone has positive effects when treatment centers are prevalent
As it happens, the authors’ handling of the treatment issue is somewhat contradictory. On the one hand, they find that where naloxone is readily available, there are fewer Google searches for treatment options.
On the other, they find that in areas with the most treatment programs, naloxone laws may reduce opioid mortality. That is a remarkable and hopeful finding, but it’s buried under the paper’s more controversial conclusions. And it’s not consistent with the authors’ moral hazard hypothesis, which predicts that naloxone makes treatment less attractive since it allows people to avoid death without treatment.
The authors only find that naloxone access laws increase opioid overdose deaths in small to medium cities and in the Midwest, but they don’t have convincing explanations for why that might be. That suggests the finding could be spurious, the result of a chance, or the product of some other unaccounted-for factor.
The authors claim that opioid mortality might have increased in the Midwest in part because naloxone makes people feel emboldened to use fentanyl, a synthetic opioid up to 100 times more potent than morphine. But this, too, is inconsistent with the experience of many people who use drugs. Fentanyl is increasingly added into heroin, but people who use drugs generally don’t know when it’s present; most opioid users are not seeking out fentanyl.
There are other problems with the authors’ approach — for instance, the fact that there is evidence of measurement error in opioid death data. It may be the case that states that pass naloxone access laws also make efforts to more accurately classify opioid-related deaths.
But despite my concerns, I appreciate the authors’ attention to this issue. We can’t stop researching controversial questions because policymakers may misconstrue the findings.
But researchers should do our best to inform studies with the lived experiences of the people whom our research is most likely to affect. I hope the authors of this hotly debated study will consider this point as they revise and refine their analysis.
Alex Gertner is an MD/PhD candidate in the department of health policy management at the University of North Carolina Chapel Hill. Find him on Twitter @setmoreoff.
The Big Idea is Vox’s home for smart discussion of the most important issues and ideas in politics, science, and culture — typically by outside contributors. If you have an idea for a piece, pitch us at thebigidea@vox.com.