The idea of harm reduction begins with a realistic premise: people sometimes continue risky behavior despite prohibition, stigma, or advice to stop. Harm reduction policies attempt the opposite of a prohibition or abstinence approach. Harm reduction policies are based on the belief that lives can be saved by helping people move from more dangerous choices to less dangerous ones. Fentanyl test strips, vaping, and even ordinary automobile driving all demonstrate the promise of this approach and its limits. Harm reduction works best when the relative risks are clear, people can act on the information or substitute offered, and the intervention complements rather than replaces prevention and treatment. The lives saved, and the other harms prevented, extend both to those who’ve undertaken the risky behaviors and to nonparticipant bystanders.
Get Liberalism.org in your inbox.
When a behavior is stigmatized enough, we stop asking people to reduce harm and start asking them to eliminate it entirely. Hence, rules aimed at eliminating certain harmful behaviors remain popular among policymakers precisely because that’s their dream. But it isn’t what we get. In the real world, people can’t or won’t completely change their behavior. So the operative question becomes what would make people safer today, given what they will actually do, rather than what legislators wish they would do.
The logic is that whether and how are separate questions. Whether someone drives, smokes, or uses drugs is one choice. How dangerously they do it is another. Prohibition and abstinence-based rules try to resolve both at once. In an ideal world, prohibition eliminates the first choice, so the second never comes up. Harm reduction works on the second question alone. It takes the behavior as given, at least for now, and asks how to make it less dangerous. That doesn’t mean we have to stop caring whether people drive less, use drugs less, or smoke less. We might love for people to quit. Harm reduction means policy doesn’t require cessation to start saving lives.
We already accept this logic for driving. We know people are going to drive and accidents will happen. So we reduce the cost of those accidents by mandating seat belts and air bags, rather than simply telling people not to drive.
Despite our best efforts, some people will use drugs. How can we help those people stay alive? How can we help people who use drugs avoid overdoses and bloodborne illnesses? The prohibition/abstinence approach would encourage people to avoid overdoses and bloodborne illnesses by simply not using drugs. But that may not be possible for some people. The harm reduction approach suggests that naloxone, sterile syringes, and fentanyl test strips are a way to reduce the harms from drugs. Similarly, if people cannot quit nicotine, then vaping is a way to reduce the harm from cigarettes. Adding fentanyl test strips, naloxone, and clean needles doesn’t make drug use harmless; they make drug use less harmful. Vaping isn’t perfectly safe, but it is safer than smoking.
Harm reduction is the opposite of an “all or nothing” approach. Some may prefer to think categorically, but risks exist on a continuum rather than in the binary categories of safe and dangerous, and that is the central principle of harm reduction. Many policies can therefore improve health without eliminating the behavior producing the risk, and a marginal reduction in harm is better than no reduction.
Understanding this continuum means that we must reframe what counts as success. A policy evaluation that only asks whether drug use or smoking stops entirely will call most policies a failure. Even policies making people measurably safer will fail. Harm reduction instead takes the quality of people’s lives and health, as measured before and after the intervention, as the relevant standard.
Note that harm reduction does not require moral approval of the underlying activity. We don’t have to approve of people’s drug use to realize that we want them to stay alive long enough to quit eventually. Dead people don’t enter rehab. You can disapprove of smoking, doing drugs, or even driving, yet still want those activities to be safer.
The harm reduction approach embodies a respect for individual agency. It treats people as capable of weighing and curbing the risks of their own behaviors, even the relatively risky ones. It doesn’t assume that continued drug use or continued smoking reflects a failure to understand the danger. The practical consequence of a harm reduction approach is that help is not made conditional on compliance. A person does not need to commit to quitting before receiving naloxone, a clean syringe, or a fentanyl test strip. Furthermore, services are offered rather than imposed, meaning that someone can decline them without forfeiting future help.
This respect for people’s agency isn’t unconditional. We draw a specific line. We respect your agency as it applies to harms that befall you, since you’re making the choice. We don’t respect your agency for harms that spill onto others. A drinker who accepts the risk of a hangover is making a different kind of choice from a drinker who gets behind the wheel. Someone testing their own drugs for fentanyl is in a different situation from the neighbor who finds a used needle in a stairwell. Harm reduction can afford to defer to individual judgment precisely because so much risky behavior really is self-regarding. But not all of it is. Any framework built on respecting agency still has to say what happens when people don’t respect others.
This is not the same as ignoring consequences. A successful harm reduction strategy potentially narrows the case for intervening in a person’s own choices. It does not narrow the case for intervening on behalf of at-risk third parties. Not all harms, after all, are limited to the person making the choice. Youth initiation, secondhand exposure, impaired driving, and neighborhood effects all complicate a purely individual risk-reward calculation. Respecting an individual’s agency to make their own choices does not eliminate legitimate concern about the costs those choices impose on others.
What a Good Harm Reduction Strategy Looks Like
Harm reduction is evaluated based on real-world evidence and results, rather than on assumptions or beliefs made in advance. We ask questions such as: Would this policy substantially reduce an important risk? Will people actually use the resource or strategy that the policy recommends, and reduce their risk, or will they increase their overall risk now that the behavior is “safer”? Does the harm reduction program create significant new harms for nonusers? Does the harm reduction policy complement or displace cessation, treatment, and prevention?
My own research with Bhai and McMichael (2025) published in Medical Care Research and Review analyzed fentanyl test strips. Fentanyl test strips (FTS) are a particularly clear example of harm reduction because they address a severe information problem in an illicit market. The problem is that illicitly manufactured fentanyl can appear in opioids, counterfeit pills, stimulants, and other drugs. Users cannot reliably detect fentanyl by sight, taste, or smell. Drug prohibition leaves consumers without ordinary product labeling, quality control, or legal remedies.
But test strips can reduce harm. They provide inexpensive, rapid information about whether fentanyl is present. A positive result can prompt someone to discard the substance, use less, avoid using alone, carry naloxone, or alert another person. They do not require a person to abstain. Instead, they provide potentially lifesaving information.
Currently most states explicitly treat FTS as a legally permitted harm reduction strategy. In the past, however, FTS were treated as drug paraphernalia, so they made possessors liable to drug charges. Various states legalized FTS at different times, creating a natural experiment that compares changes in overdose mortality in states that changed their laws with changes in states that have not done so. States that treated FTS as harm reduction still had overdoses, but the policy led to a 7% reduction in overdose mortality.
Some lawmakers worried that if people believed they could use drugs more safely, they would feel less urgency to seek treatment. Another worry was that removing the fear of a fentanyl overdose might push people toward methamphetamines. These are exactly the kind of behavioral offsets that could, in principle, erase the benefit of better information. But the reduction in overdose mortality shows that this offset did not overwhelm the direct benefit of the policy. We cannot rule out that some individuals may have responded exactly as these lawmakers feared.
FTS have limits, of course. A strip only indicates presence, not potency or distribution throughout a sample. Users could get a false negative. Moreover, some fentanyl-like substances, or other dangerous adulterants, like xylazine, may not be detected. If fentanyl is pervasive, users may have few viable alternatives. Information only reduces harm if users possess the means and willingness to respond.
Test strips reduce uncertainty about a dangerous product. Vaping presents a slightly different model. Vaping replaces a highly dangerous nicotine delivery system with a less dangerous one. Cigarettes are especially harmful because burnt tobacco creates toxic smoke and carcinogenic chemicals. Nicotine creates dependence, while combustion produces much of smoking’s risk. Noncombustible nicotine products therefore occupy a different position on the risk continuum.
The potential benefit is that an adult who would otherwise continue smoking may reduce exposure by switching completely to vaping. Current randomized evidence indicates that nicotine e-cigarettes can help some adults stop smoking and may outperform traditional nicotine-replacement therapy for cessation (Hartmann–Boyce et al., 2020). Harm reduction asks whether vaping is safer than continued smoking, not whether vaping is perfectly safe.
Concerns do exist about users’ full transition to vaped nicotine. Dual use, in which users in a harm reduction scenario for a time consume both cigarettes and vapes, may preserve much of smoking’s risk, especially its cardiovascular risk. Cutting cigarette consumption without eliminating combustion should not automatically be treated as proportional risk reduction. Vaping may not be harm-reducing if people simply use it as a way to get their nicotine fix until they can smoke.
There is a second concern about vaping. Vaping could initiate nicotine dependence among adolescents and nonsmokers. Products and marketing designed to help adult smokers may attract new users. While we do have more than twenty years of vaping data, the long-term risks are less certain than those of cigarettes. There is also the potential that dangerous chemicals can be inhaled through vapes. To be effective, vaping regulations need to provide attractive lower-risk alternatives for adult smokers while limiting youth access and deceptive health claims.
FTS and vaping both teach us a few things. First, they both reject the false choice between perfect safety and doing nothing. Both give individuals tools to reduce risk without requiring immediate abstinence. Both depend on accurate information about relative risks.
There are important differences, of course. Test strips detect a hazard in an existing product, while vaping substitutes one product for another. Test strips address immediate overdose risk. Tobacco reduction concerns a disease accumulating over years. The principal beneficiary of test strips is usually the existing user. But vaping policy could lead to youth initiation. Some people may become addicted to nicotine because they pick up vaping. A positive FTS test result does not guarantee safer behavior, while access to vaping does not guarantee that anyone ceases smoking.
A Framework, Not a Guarantee
Harm reduction is a framework. It’s not proof that every intervention bearing the label produces a net benefit. It has real limits. Policies should communicate residual risk honestly. Overstatement can destroy trust; understatement can discourage beneficial switching. There is also a behavioral response to consider. People may ignore warnings, use two products, compensate for lower perceived risk, or take greater risks now that the risk feels smaller. When risk-takers become aware that their overall risks declined, they may respond by increasing their risk in other ways. The correct question is not whether this kind of risk compensation is possible, but how large it is relative to the direct benefit.
Economists call this behavioral offset the Peltzman effect. Peltzman (1975) showed that mandatory automobile safety equipment did not reduce driving deaths as much as engineering estimates predicted. Safer cars encouraged more aggressive driving. The change in safety moved risk from drivers to pedestrians and cyclists, who bore none of the benefit. People’s responses did not completely offset the gains in safety, but there was some offset. Drivers were safer, but pedestrians were less safe than expected. The lesson is not that policies mandating safety equipment failed, but that people adjusted their behavior in risk, which ate into the expected gain. That effect forces others to also adjust their behavior: to walk and bike less, to use other routes, or for cities to create protected intersections and better enforce traffic laws.
Individual and population effects can also diverge. An intervention that clearly benefits current high-risk users may simultaneously create new risk for people who would not otherwise have been exposed at all. Vaping makes this tension especially visible. Vapes help adult smokers quit, but the same product can introduce nicotine to teenagers who never smoked. Evaluating harm reduction from only one vantage point can miss half the picture.
Harm reduction also doesn’t solve the underlying conditions that drive risky behavior in the first place. Fentanyl test strips do not treat substance-use disorder, provide stable housing, or create a safe drug supply. Vaping does not eliminate nicotine dependence. These interventions reduce the harm of a behavior without addressing why the behavior persists. That is a real limitation. Harm reduction is meant to function alongside treatment, prevention, and recovery, not to substitute for them.
We must also be on guard for supposed harm reduction strategies that simply miss the mark. Firms have an incentive to market a profitable product as harm reduction, whether or not the evidence supports the claim. Advocates sometimes treat any criticism of a specific intervention as opposition to saving lives. But we need to test these policies empirically to make sure they work as hoped. Policymakers should demand evidence about actual behavior and net outcomes, not good intentions.
In an ideal world, we would eliminate all harm. In the world we actually live in, some people will keep using drugs, and some smokers will not quit. Fentanyl test strips supply information in an illicit market that otherwise offers none. Vaping offers a lower-risk alternative to combustion of tobacco. But vaping only delivers its full benefit when the switching is complete, and if it does not pull in new users. Neither intervention makes the underlying behavior safe. A test strip does not treat substance-use disorder, and a vape does not end nicotine dependence. Both work by moving people along a continuum of risk, instead of pretending the continuum doesn’t exist.
Harm reduction, properly understood, calls for supplying a non-judgmental alternative to people participating in risky behaviors to reduce the harm those choices carry. This is not the same as eliminating the harm itself. Risky behaviors have real, tragic, and sometimes permanent consequences. A serious harm-reduction framework says so plainly rather than looking away from it. What harm reduction refuses is the false choice between endorsing a behavior and simply abandoning the people who engage in it. Harm reduction is neither surrender nor a cure-all. It is the disciplined effort to make people safer while continuing to pursue prevention, treatment, and hopefully cessation.
References
Bhai, M., McMichael, B. J., & Mitchell, D. T. (2025). Impact of fentanyl test strips as harm reduction for drug-related mortality. Medical Care Research and Review, 82(3), 240-251.
Hartmann-Boyce, J., McRobbie, H., Lindson, N., Bullen, C., Begh, R., & Theodoulou, A. (2020). Can electronic cigarettes help people stop smoking, and do they have any unwanted effects when used for this purpose? Cochrane Database Syst Rev, 10(250), 10-1002.
Peltzman, S. (1975). The effects of automobile safety regulation. Journal of Political Economy, 83(4), 677-725.
