No federal standard requires a Canadian province to keep a supervised consumption site open — Ontario ordered 10 of 17 closed, and its auditor general reported in 2024 that the decision came without impact analysis

What is happening?

A supervised consumption site operates in Canada because the federal Minister of Health has granted it an exemption under section 56.1 of the Controlled Drugs and Substances Act. That section is written as a discretionary power: the Minister may exempt a site if, in the opinion of the Minister, the exemption is necessary for a medical purpose. The section consulted on 2026-08-18 carries no obligation to keep any site open, no minimum number of sites, and no service standard a province has to meet.

The harm the sites were built for is counted federally and is large. 56,631 opioid- and stimulant-related deaths were recorded across Canada from January 2016 through December 2025, a total that includes provisional figures and was last updated 2026-06-15. On the federal data page for the sites themselves, updated 2026-06-26, the network recorded 1,658,116 visits and 25,676 non-fatal overdose responses on site since March 2020.

Two provinces then moved in the opposite direction, and nothing federal stood in the way.

Ontario. The Safer Streets, Stronger Communities Act, 2024, introduced as Bill 223 and enacted as Statutes of Ontario 2024, chapter 27, barred a supervised consumption site from operating within 200 metres of a school or a child care centre. Ten of the seventeen sites then operating in Ontario fell under that provision. The Office of the Auditor General of Ontario reported in its 2024 annual report, released in December 2024, that the closure decision had been made without proper planning, impact analysis or public consultation. The same report recorded that those sites had responded to more than 1,500 overdoses on site in a single year across 2022 and 2023 with no deaths on site, in a year when Ontario recorded more than 2,500 opioid-related deaths overall. On 2025-04-01, nine of the ten reopened as provincially funded abstinence- and recovery-oriented HART hubs. One continued to operate as a supervised consumption site under an interim court order; that proceeding is unresolved and its merits are not described here. The replacement funding announced alongside the change grew from 378 million Canadian dollars for 19 hubs to 550 million Canadian dollars for 28 hubs, reported as roughly four times the budget of the sites being replaced.

British Columbia. A three-year pilot decriminalizing possession of small amounts, granted as a federal exemption, began 2023-01-31. The federal government amended the exemption on 2024-05-07 to exclude public places. The exemption expired 2026-01-31, and on 2026-01-14 the provincial government announced it would not seek a renewal, stating that the pilot had not delivered the results it had hoped for. British Columbia is the jurisdiction that declared this a public health emergency on its own in 2016; its statement on the tenth anniversary, 2026-04-13, recorded more than 18,000 deaths since that declaration and 1,826 in 2025 alone.

Whose problem is this?

The federal power that opens a site and the provincial power that can close it sit in different governments, and only one of the two is written down as a permission that can be withdrawn.

Where does this problem end?

The boundary here is not whether these services work but whether any level of government is obliged to keep them running once they exist.

What is the state now, and what should it be?

Now

Needs a new measurementthe target state: no source opened here names a target. Not the exemption section, which sets a permission and not a floor. Not the two federal data pages, which count harms and visits and set no goal against which those counts could fall short. Not the provincial statements, which announce decisions without a number to hit. The federal drugs and substances strategy page, the one document most likely to carry a coverage target, returned HTTP 403 to every fetch attempt, so whether it names one is unresolved rather than answered. Nothing is filled in here from advocacy or professional bodies, because a target chosen by anyone other than the government that would have to meet it is not a target.

How big is it?

Between about 178,000 and about 263,000 distinct people, over the period from March 2020 to the June 2026 federal data update. The full chain and its assumptions are in the population record; the short version is that the federal page counts visits and not people, so a second term is needed to turn 1,658,116 visits into a count of persons, and the only ratio available is a secondary figure for the Ontario network, where about 1.12 million visits corresponded to about 178,000 distinct clients.

The upper end is conservative in a specific and checkable way. The Ontario ratio of roughly 6.3 visits per client was measured over about five years, while the national visit total covers about six years and three months. A longer window lets one person accumulate more visits, so the true national ratio is higher than 6.3, which makes the true number of distinct people lower than 263,000, not higher.

The weaker end is the floor. If the 178,000 figure is the sum of unique clients counted site by site rather than distinct people across the whole network, it double counts anyone who used more than one site, and then it is not a strict floor for Canada at all.

This count measures people who used a service that a province can withdraw, not people harmed in any single year. The annual harm is counted separately and is far larger in scope: 56,631 opioid- and stimulant-related deaths from January 2016 through December 2025, a figure that includes people who never went near a site. The two numbers answer different questions, and neither of them counts the people who would use a site if one existed near them.

Under what conditions does it arise?

1. The permission is discretionary in both directions. Section 56.1 lets a federal Minister grant an exemption where it is judged necessary for a medical purpose. A power written that way can be granted, narrowed or allowed to lapse, and the British Columbia exemption went through all three: granted 2023-01-31, narrowed 2024-05-07, expired 2026-01-31 with no renewal requested.

2. A province does not need the federal power in order to defeat it. Ontario did not ask for the exemptions to be revoked. It legislated where a site may stand, and a distance rule of 200 metres from a school or child care centre reached ten of seventeen sites without touching the federal instrument at all. A permission that is silent about location can be emptied by a rule about location.

3. The decision needs no impact assessment before it takes effect. The Office of the Auditor General of Ontario reported in December 2024 that the closures had been ordered without proper planning, impact analysis or public consultation, and the sites closed on schedule the following spring regardless. An audit finding published after a decision does not reverse it.

4. Replacement is measured in money, not in the function that was removed. The provincial funding grew from 378 million Canadian dollars for 19 hubs to 550 million Canadian dollars for 28 hubs, an amount reported as several times the budget of the sites replaced. Whether the replacement absorbs the same on-site overdose response that produced more than 1,500 interventions in a year is a different question from whether it costs more.

A permission that can be granted at discretion can also lapse at discretion, and nothing in the statute fixes a floor beneath it.

What has been tried?

Two of the attempts here run against each other inside the same federal instrument. The exemption that opened the sites and the exemption that opened the pilot were granted by the same power, and both were reversed without that power being used.

What was found?

Why is it still unsolved?

Institutional gap — the federal government can permit a service, and only a province can decide whether it survives.

The instrument in use is a permission. Section 56.1 answers the question of whether a site may lawfully operate, and it answers it one site at a time, at ministerial discretion, on a medical-purpose judgment. It does not answer the question this problem is about, which is whether anyone must keep the service running once it exists. Those are different questions, and the statute only contains an answer to the first.

That means a province needs no federal cooperation to reverse the outcome. Ontario legislated about distance from schools and child care and reached ten of seventeen sites without disturbing a single exemption. British Columbia let an exemption run out. In both cases the federal instrument stayed exactly as it was and the service on the ground did not. An instrument that can be defeated without being touched has no floor under it.

The third part is that the reversal carries no procedural cost. The Office of the Auditor General of Ontario reported in December 2024 that the closures had been ordered without proper planning, impact analysis or public consultation, and the closures proceeded anyway in the spring of 2025. Where an audit arrives after the decision and no rule requires the assessment beforehand, the finding becomes a description of what happened rather than a condition on what may happen. A discretionary permission is not a standard, and a service that exists only while a government wants it to exist is not a floor.

What observation would mean it is solved?

Candidates — (a) a federal instrument states a minimum that a province cannot unilaterally remove, whether as a service standard attached to the exemption power or as a condition on health transfers (b) a decision to close or convert an approved site requires a published impact assessment before it takes effect (c) the count of people with access to a supervised consumption service stops moving with provincial policy cycles.

(a) alone can be written thin. A minimum stated at a level below what already exists changes nothing, and a standard whose only sanction is a report keeps the same structure this document describes. A federal floor also has to survive the division-of-powers question that no source opened here addresses.

(b) alone measures paperwork. An impact assessment that is produced, published and then disregarded satisfies the observation while leaving the outcome identical. The Ontario audit shows what a finding is worth when it arrives after the fact; requiring it beforehand only helps if the finding can stop the decision.

(c) alone is the hardest to read and the most important. Access is not the count of open sites, because sites differ in hours, capacity and reach, and the national count of operating sites could not even be established here. It is also confounded in both directions: a fall in overdose deaths could follow better access, a changed supply, or a change in how deaths are attributed, and the federal count moves for all three reasons. The three have to be read together, and (c) has to be read against the year it came from.

What is it connected to?

Fills with researchfederal and provincial division of powers over health services, the funding conditions attached to federal health transfers, drug policy in the provinces not examined here, the criminal law treatment of possession after the end of the British Columbia pilot, and the same permit-without-floor structure in other services delivered under federal exemption. Relation type and evidence grade were not confirmed in this round.

What these sources do not say

• Whether any federal document sets a minimum. The negative in this document rests on one section of one statute. The federal drugs and substances strategy page, which is where a coverage target or a service standard would most plausibly appear, returned HTTP 403 to every fetch attempt, so no source opened here either states such a minimum or states that none exists.
• How many supervised consumption sites operate in Canada today. The federal data page publishes visits and overdose responses, not a current count of open sites. The professional-body statement that carries a figure returned HTTP 404, so the number is unconfirmed here.
• What happened after the Ontario closures. No government post-closure impact assessment was found. The only post-closure overdose figures found anywhere here were compiled by a local advocacy organization and reported in one outlet, and no opened source establishes a causal link between the closures and any later change.
• Why the province chose 200 metres. The statute states the distance. No source opened here carries a stated basis for it, an estimate of how many sites it would reach, or a record that either was prepared before the vote.
• Why the audit finding did not change the outcome. No source opened here carries an on-the-record response from the provincial government to the auditor general finding.
• What the replacement model delivers. The funding figures for the HART hubs are reported, but no opened source reports what those hubs recorded against the on-site overdose response the closed sites performed, which is the comparison that would say whether the function was replaced or removed.
• Beyond a single line, why British Columbia did not seek renewal. The provincial statement of 2026-01-14 says the pilot had not delivered the results hoped for. No evaluation, no data series and no criterion of success accompanies it in anything opened here.
• The federal position on calls for emergency powers. The page carrying federal comments on that subject returned HTTP 403, so no account of the federal position is established here.
• Provinces other than Ontario and British Columbia. Their site policies and any changes to them were not examined in this round.

See the evidence

No primary audit or rulemaking document was read in full. The two federal data pages and the statute section were opened directly and carry the national counts and the text of the exemption power. The two provincial statements were opened directly. The provincial statute page was opened and confirms the 200 metre provision, but not the count of sites it reached. Everything about the audit itself is one step removed: the annual report index page was opened, the audit PDF could not be extracted, and the finding and the on-site figures come from a secondary report of it. Where sources overlap they agree — the closure of ten sites and the conversion of nine appear in two opened sources, and the 2025-04-01 date appears in one with the statutory timeline supporting it. Where they do not overlap the gap is left visible rather than filled: three sources could not be opened at all and their rows carry no URL and no claim, the count of sites now operating nationally is unestablished, and the only post-closure overdose figures anywhere here are an advocacy compilation carried in one outlet and are graded low for that reason. This is a Path A output (research-based definition), so observation_refs is empty and provenance_mode: press-derived.

People affected

Estimated range 178,000–263,000 As of 2020-03 through the 2026-06 federal data update

Sensitivity The interval is 178,000 to about 263,000 and both ends rest on one secondary figure for one province. The upper end is conservative in a direction that can be checked: the Ontario ratio was measured over about five years while the national visit total covers about six years and three months, and a longer window lets one person accumulate more visits, so the true national ratio is above 6.29 and the true count of distinct people is below 263,000 rather than above it. The weaker end is the floor. If the 178,000 figure is the sum of unique clients counted site by site rather than distinct people across the whole network, it double counts anyone who used more than one site and is not a strict floor for Canada. Neither end counts people who would use a site if one existed near them, people who used a site before March 2020, or people in provinces that never opened one. Neither end is a count of people harmed: 56,631 opioid- and stimulant-related deaths were recorded from January 2016 through December 2025 on a separate federal page, and that count includes people who never went near a site.

Regional breakdown The federal data page publishes national visit and overdose-response totals and does not break clients out by province or territory, and the only provincial client figure found is a secondary count for Ontario. Splitting the national total by provincial population would be apportionment, which this layer does not accept, and it would be wrong here in a specific way: sites are not distributed in proportion to population, and several provinces have never had one.