I came across most of my sources while studying for proposals to open another one, and don't have them convenient anymore.
However, looking through NIH reports --
The site in Vancouver appears to lower the fatalities marginally (1-10 less fatalities per year). Several other studies indicate that fatalities related to opioid overdose, ambulance calls, etc decrease in the immediate area of various sites.
I can't find anything on the Vancouver site in the last decade, it's all circa 2006 or 2007 -- and seem to recall in my earlier research that the initial results looked good, but later studies basically reported that we saw a decrease in fatalities that was offset by an increase in usage, and the whole thing came out as sort of a wash.
RAND seems to suggest that supervised consumption is not settled in terms of scientific study, even now. [0]
> We conducted our own assessment of the individual studies and found that the evidence base concerning the overall effects of SCSs is limited in quality and location. Although we identified 65 outcome-related articles (as opposed to commentaries or studies that gauge opinions), none involved an RCT, and just nine employed a quasi-experimental design with control groups. These nine studies were based on just four SCSs in three cities: Sydney (five studies), Vancouver (two studies), and Barcelona (two studies). There was considerable overlap in the design,
methods, authorship, and data employed, so that four of the nine are arguably superseded by later studies using better methods or longer time series, in some sense reducing the effective number of quasi-experimental studies to five.
Worth noting, however, that many of these seem to indicate reduced petty crime in the area immediately around the site. Indicative I might be wrong, but not conclusive because of problems in the methods used, samples chosen, etc.
Looking at a compilation by Brookings, broadly on the topic [1] --
> New work by Packham and Wells (2018) suggests that syringe exchange programs—a staple of harm-reduction efforts—reduce HIV rates as intended but unintentionally increase opioid-related mortality by making it easier, cheaper, and safer to use heroin.
> Another popular harm-reduction effort is distributing naloxone, a drug that can save someone’s life if administered during an opioid overdose. Recent work by Doleac and Mukherjee (2018) finds that broadening naloxone access had mixed results. While some states saw beneficial effects of these policy changes, broad naloxone access increased opioid-related mortality by 14 percent in the Midwest, the region hardest-hit by the opioid epidemic.
So reports that I do find seem to suggest that this is a concern that is valid, at least in some contexts, and that the studies purporting to show outcomes are a) very narrowly done, b) often don't measure what I'd consider to be the complete set of externalities, and c) aren't particularly scientific.
It certainly seems reasonable that narrow focus on improving the outcomes of the addicts, particularly a narrowly defined set of metrics, actually lowers the group outcome.
Sorry I can't provide better sources to support that, though.
[0] -- https://www.rand.org/pubs/research_reports/RR2693.html
[1] -- https://www.brookings.edu/blog/up-front/2018/12/07/research-...