The soaring death count of the opioid epidemic has spurred a search for novel solutions. One of the most controversial is to establish sites where people can use illegal drugs such as heroin under the supervision of trained staff who can offer clean injection equipment and intervene in the event of an overdose. Prominent Massachusetts physician Henry Dorkin spoke for many proponents when he argued supervised drug consumption facilities (SDCF) yield “substantial and evidence-backed reductions of death, disease, and expenditures,” whereas U.S. Deputy Attorney General Rod J. Rosenstein spoke for many critics when he asserted that SDCFs “are very dangerous and would only make the opioid crisis worse.” A just-released, comprehensive, review of the evidence by RAND Corporation scientists will disappoint both extremes in this debate, but is highly informative for any city contemplating whether to establish an SDCF.

The RAND team, led by Beau Kilmer, concluded that an important aspect of the experience of SDCF staff was echoed in scientific findings: a medically supervised episode of drug use is almost certainly less likely to lead to death or infectious disease transmission than is an episode elsewhere, such as alone in an alley. With provision of the rescue drug naloxone and other medical measures, for example, almost any opioid overdose can be reversed before it proves fatal.

However, existing research does not establish that drug users who access SDCFs are less likely to die of an overdose over time, or that opening an SDCF lowers a community’s rate of drug overdose fatalities. These two conclusions at first may seem like a contradiction of the findings on the reduced riskiness of supervised drug use sessions, but they are not.

The report found that many people use SDCFs intermittently, but do not adopt the safer use practices from SDCFs when using outside of it. An individual who injects heroin in the SDCF one day may thus avoid a fatal overdose that particular day, but have one the next day outside the SDCF. More importantly, no one knows whether becoming an SDCF user leads to longer drug use careers than do other interventions (e.g., methadone maintenance). If by making injection drug use safer and more positive (e.g., being surrounded by supportive people), SDCFs even modestly reduce the likelihood of an individual stopping injection use in the next week, or month, or year, the benefit of lower risk SDCF drug use now can be canceled out by an increased number of drug use episodes later.

Why don’t individuals’ SDCF experiences of lower-risk drug use clearly translate into lower community rates of overdose death? The RAND researchers highlight challenges of scalability: Even the busiest SDCF can oversee at most 150,000 injections a year, whereas American drug users inject heroin 1.1 billion times a year (and that astonishing number doesn’t even count injections of other drugs). It would require opening about 7,000 SDCFs to allow supervision of so many injections. If you want to appreciate how hard that would be, consider that three decades after the first SDCF opened, only about 100 exist in the entire world.

Perhaps the RAND team’s most important conclusion was that most SDCF evaluation studies are of poor methodological quality. When the scientists imposed even a minimal methodological bar for studies to be included in the review, only a handful made the cut. This should make anyone hesitant to make strong claims regarding what the evidence on SDCFs “clearly proves.” But with drug policy debate being what it is, there is no guarantee that many people will take that admonition to heart.