# Lesson 5: Opioids and the Toxic Unregulated Drug Supply

*Companion-podcast transcript • Sarah & Kiffer*  
*Office Hours episode to listen to after working through the lesson*

---

**Sarah:** Welcome back to Office Hours. I'm Sarah.

**Kiffer:** And I'm Kiffer. Between January 2016 and March 2026, the Public Health Agency of Canada counted fifty-seven thousand five hundred nine apparent opioid toxicity deaths, and each of them was a person with a family and a community. The support lines listed in the lesson, including nine eight eight and the National Overdose Response Service, are there for anyone who wants them. Lesson five is about opioids and the toxic unregulated drug supply, and some of what we know about that supply comes from the agency's percentages.

**Sarah:** In 2021, fentanyl was involved in eighty-six percent of the apparent opioid toxicity deaths the agency counted. In 2025, it was fifty-seven percent.

**Kiffer:** So is fentanyl leaving the supply? Tell me how you'd read that drop if those two figures were all you had.

**Sarah:** When I see a percentage in a surveillance report, I tend to read it as a fact about the drug. Fifty-seven percent sounds as though fentanyl has become less of a problem.

**Kiffer:** That's the natural reading, so it helps to ask three things of any percentage. Percent of what? Can the categories overlap? And what happened to the count underneath it? A share of deaths can fall because fewer deaths involved the drug, or because more deaths happened without it, and the share alone blends those together.

**Sarah:** Then let's get the counts. Here's one for everyone listening. In the Public Health Agency of Canada's data published in September 2026, there were seven thousand six hundred seventeen apparent opioid toxicity deaths in Canada in 2021, and eighteen percent involved a fentanyl analogue. In 2025, there were five thousand seven hundred four, and sixty percent involved one. Did the number of deaths involving an analogue go up or down, and by roughly how much? Pause the audio if you'd like more time. Take a few seconds.

*(Pause)*

**Kiffer:** Eighteen percent of seven thousand six hundred seventeen is about one thousand three hundred seventy. Sixty percent of five thousand seven hundred four is about three thousand four hundred twenty. So deaths involving an analogue rose about two and a half times, while total deaths fell by about a quarter.

**Sarah:** Let me do fentanyl itself. Eighty-six percent of the 2021 total is about six thousand five hundred fifty, and fifty-seven percent of the 2025 total is about three thousand two hundred fifty. So deaths involving fentanyl roughly halved.

**Kiffer:** Right, and part of that drop reflects fewer deaths overall, while part reflects analogues taking fentanyl's place. The fifty-seven percent on its own can't tell you how much of each. There's also a wrinkle in the count underneath. From 2025, the national total includes all of British Columbia's unregulated drug deaths and uses a broader definition for Quebec, so 2021 and 2025 aren't counted in quite the same way. That's one more reason to keep these estimates rough.

**Sarah:** The second check was overlap. For 2025, the BC Coroners Service's expedited toxicology found fentanyl in sixty-nine percent of the people tested, fluorofentanyl in fifty-four percent, cocaine in fifty-three and methamphetamine in fifty-two. Those add up to two hundred twenty-eight percent.

**Kiffer:** And that's fine, because one person can have several substances detected. In fact, a total of two hundred twenty-eight percent means that, on average, each person tested had 2.28 of those four substances detected. The sum can't tell you which substances turned up together, though.

**Sarah:** For that I'd look for a figure built for the question, like the Public Health Agency's national one: in 2025, seventy-one percent of apparent opioid toxicity deaths also involved a stimulant. And then there's the remaining check, percent of what. I'd be tempted to take sixty-nine percent of the one thousand eight hundred twenty-six deaths the Coroners Service reported for 2025 and say that about one thousand two hundred sixty people died with fentanyl in their system.

**Kiffer:** That's a fair estimate, as long as you say what it assumes. The sixty-nine percent describes the people who underwent expedited testing, and the one thousand eight hundred twenty-six was a preliminary count when it was released in February 2026. Your estimate assumes the tested group looks like everyone who died, and it will move as the count is revised.

**Sarah:** The same question matters for drug checking. When Toronto's Drug Checking Service found medetomidine in eighty-six percent of expected fentanyl samples in February 2026, that was eighty-two of ninety-five samples.

**Kiffer:** Those were ninety-five samples brought in by people who chose to use the service. That makes it a timely early signal, tied to what people expected to buy, and it's a long way from a random sample of a city's drug market. I'd put more trust in the trend, from twelve percent in August 2024 to eighty-six percent in February 2026, than in any single month's level.

**Sarah:** One more on numbers. In the same national data, deaths fell by twenty-nine percent between 2023 and 2025, and people will want to credit whichever policy they favour.

**Kiffer:** The counts can't settle that, because several things changed at once, including the supply itself. Working out why deaths fell takes the policy evaluation methods that HSCI 402 develops. What the counts do show clearly is that the supply in 2025 looked very different from the supply in 2021.

**Sarah:** Question two. What does a potency figure like ten thousand times actually tell you?

**Kiffer:** It tells you how much mass two drugs need to produce the same effect, and each figure carries its conditions with it: the route, the effect measured and the people studied. The heroin figure is a measured comparison. In a 1981 trial among one hundred sixty-six patients with cancer who had pain after surgery, intramuscular heroin was about twice as potent as morphine, with confidence limits from 1.6 to 2.6.

**Sarah:** And the ten thousand for carfentanil?

**Kiffer:** That figure comes from a public warning issued by the United States Drug Enforcement Administration in September 2016. It's a different kind of number from the heroin figure, which came from a trial with confidence limits. I'd treat it as an order of magnitude, a way of saying that carfentanil is roughly a hundred times as potent as fentanyl.

**Sarah:** The nitazenes seem even harder to pin down.

**Kiffer:** They are, and they show why the method matters. In laboratory assays on cells, several nitazenes were more potent than fentanyl at the mu-opioid receptor. A review of fatal cases found blood concentrations consistent with a potency similar to fentanyl's for the analogues it examined. One study measured the receptor and the other measured people who died, so the two findings can differ and both be correct.

**Sarah:** Here's a problem that uses the rough figures anyway. Picture the opioid in a hypothetical batch as a pile of one hundred grains of fentanyl, the teaching device from the lesson. Now suppose one of those grains is carfentanil.

**Kiffer:** So here's the question for everyone. Counted in morphine grains, does the amount of opioid in that pile go up by about one percent, about double, or about a hundredfold? Take a few seconds.

*(Pause)*

**Sarah:** My gut says about one percent, since only one grain in a hundred changed.

**Kiffer:** That's the intuitive answer, so let's check it. Put everything in morphine grains, using the figures from the lesson. At the upper figure, fentanyl is a hundred times as potent as morphine, so the original pile is worth one hundred times one hundred, or ten thousand morphine grains.

**Sarah:** And after the swap, ninety-nine grains of fentanyl are worth nine thousand nine hundred, and the one grain of carfentanil is worth ten thousand on its own. That's nineteen thousand nine hundred, so the amount of opioid in the pile roughly doubles. At the lower figure, fifty times morphine, the original pile is worth five thousand, and the new one is worth four thousand nine hundred fifty plus ten thousand, which is fourteen thousand nine hundred fifty. That's about three times as much.

**Kiffer:** So one percent of the mass changed, and the amount of opioid, counted in morphine grains, doubled or tripled, with nothing about the pile looking any different. That's the point Ciccarone made, that rapid changes in potency, purity and mixture may be even more dangerous than fentanyl's potency itself. It connects to question one, too. As analogues move through the supply, products that look the same can carry very different amounts of opioid. And since the inputs are rough, the honest answer here is roughly double to triple, which is as precise as those figures allow.

**Sarah:** So the pile has two to three times the effect?

**Kiffer:** That's the step to be careful with. It has two to three times the opioid, and effects don't rise in step with dose. The dose-response curves in Lesson three showed that the same added amount can change the effect a lot or a little, depending on where on the curve a person starts and on their tolerance. What we can say is that the pile carries two to three times the opioid that anyone expected, with no way to see it.

**Sarah:** Question three. If naloxone works, why would anyone need a second dose, and why might a person keep breathing and still not wake up?

**Kiffer:** Let's use the receptor-share model from the lesson. The opioid's share of receptors is its units divided by the sum of its units, the naloxone units and one. The numbers here are hypothetical, and the model is a simplified teaching tool. It assumes the drugs have settled into a steady balance at the receptor, and the effect on breathing doesn't rise in exact proportion to the share.

**Sarah:** I'd like to try one, because I suspect my intuition about this is off. Give me a problem.

**Kiffer:** Suppose an opioid is present at forty-nine units. Fifty units of naloxone bring its share down to forty-nine divided by one hundred, or forty-nine percent. How much naloxone, in total, would bring the share down to twenty-five percent? Everyone listening can try this one too. Take a few seconds.

*(Pause)*

**Sarah:** Okay. Fifty units took it to forty-nine percent, and I want about half of that. So I'd double the naloxone to one hundred units.

**Kiffer:** Check that by putting your answer back into the equation.

**Sarah:** Forty-nine divided by forty-nine plus one hundred plus one, so forty-nine divided by one hundred fifty. That's about thirty-three percent. Doubling the naloxone didn't halve the share.

**Kiffer:** Good catch. So what went wrong in the reasoning?

**Sarah:** I assumed the share would fall in proportion to the naloxone. The naloxone is only one part of the bottom of the fraction. So I'll solve it properly. If forty-nine divided by the total is one quarter, the total has to be four times forty-nine, which is one hundred ninety-six. Take away the forty-nine opioid units and the one, and the naloxone has to be one hundred forty-six units.

**Kiffer:** And forty-nine divided by one hundred ninety-six is exactly a quarter, so that checks out. It takes nearly three times the original naloxone. Look at what each block of fifty units bought. The first fifty took the share from ninety-eight percent to forty-nine. The next fifty took it only to about thirty-three. Each extra unit lowers the share less than the one before.

**Sarah:** Which is one reason a larger or more potent exposure, like a nitazene or carfentanil, can need more naloxone and repeated doses. The other reason is timing. Health Canada says naloxone is active for about thirty to one hundred twenty minutes, and fentanyl's half-life is three to seven hours.

**Kiffer:** So when the naloxone fades faster than the opioid, the share climbs back up, because the opioid was there the whole time. A review of modelling studies by Algera and colleagues also found that how quickly drugs reach and leave the receptor limits reversal, and that in some conditions reversal is less effective or even impossible. The simple model leaves all of that out.

**Sarah:** And the second half of the question, why someone can keep breathing and still not wake up.

**Kiffer:** Because the model has only one receptor in it. Medetomidine and xylazine act on alpha-two adrenergic receptors, and bromazolam acts on the GABA-A receptor. However far naloxone pushes down the opioid's share, those receptors are untouched. That's why Canadian guidance judges naloxone by breathing, and warns that giving dose after dose to wake someone can trigger precipitated withdrawal in a person who is dependent on opioids.

**Sarah:** The medetomidine report had a detail that surprised me. In British Columbia, nine one one calls for ingestion poisonings reached a one-day high of one thousand four hundred seventy-one on February twenty-second, 2026, while deaths in the province had stayed relatively steady since a decline in October 2024.

**Kiffer:** It's a good reminder that deaths are one measure of harm among several. Drugs that cause long sedation, a slow heart rate and severe withdrawal can fill emergency departments while the death count holds steady. In Alberta, reports described sedation that sometimes lasted up to a week. So before anyone says the supply has become safer, look at more than one measure.

**Sarah:** Question four. Why is the first week after release from prison so dangerous, and what does a number like one hundred twenty-nine times actually mean? Binswanger and colleagues followed thirty thousand two hundred thirty-seven people released from Washington State prisons between 1999 and 2003. In the first two weeks after release, their risk of death from a drug overdose was one hundred twenty-nine times that of other state residents. When I first read that, I pictured a huge share of people dying.

**Kiffer:** That's a common misreading. A relative risk tells you how many times higher one group's risk is than another's, so it multiplies a baseline, and the baseline risk of dying of an overdose in any given two weeks is small. Across the whole study, with an average follow-up of 1.9 years, four hundred forty-three people died from any cause, about one and a half percent of those released. What the one hundred twenty-nine shows, together with the study's finding that risk was highest in those first two weeks, is how concentrated the danger is in time, which tells you when help matters most.

**Sarah:** The Ontario study lets us look at the timing with Canadian data. Groot and colleagues reviewed coroners' files for six thousand nine hundred seventy-eight adults who died of drug toxicity in Ontario from 2006 to 2013. Seven hundred two of those deaths came within a year of release from provincial incarceration, and one hundred thirty-seven came in the first week.

**Kiffer:** So here's a question for everyone. If those seven hundred two deaths had been spread evenly over the fifty-two weeks of the year, how many would you expect in the first week, and how does one hundred thirty-seven compare? Take a few seconds.

*(Pause)*

**Sarah:** Seven hundred two divided by fifty-two is about thirteen and a half deaths a week. One hundred thirty-seven divided by thirteen and a half is about ten. So the first week carried roughly ten times its even share of these deaths. But is that the same kind of number as Binswanger's one hundred twenty-nine? It sounds like a relative risk, too.

**Kiffer:** It's a different kind. Binswanger compared risks, which means deaths divided by the number of people released, with the same figure for other residents. Our ten times compares timing among people who died. On its own it can't give anyone's risk, because a risk needs the number of people released in its denominator. The two still point the same way, and they fit the same mechanism.

**Sarah:** And the mechanism the lesson gives for both is lost tolerance.

**Kiffer:** Right. Tolerance fades during a period without use. In the receptor-share picture, the same number of opioid units takes the same share of receptors, and with tolerance gone, that share has a much larger effect on breathing. The supply may also have changed in the meantime. In Toronto's drug checking results, medetomidine went from twelve percent of expected fentanyl samples in August 2024 to eighty-six percent in February 2026. A year and a half in custody could span that whole change.

**Sarah:** And Groot's team found that in fifty-nine percent of the opioid-related deaths after release, the circumstances suggested another person could have intervened.

**Kiffer:** Which is part of why the authors saw room for prevention. The same kind of risk has been linked to other transitions, such as discharge from hospital. HSCI 402 develops care after release and naloxone distribution.

**Sarah:** I want to push on something now, because I think we see it differently. I'd say the toxic supply explains most of this crisis.

**Kiffer:** Make the case, and then I'll tell you what I'd add to it.

**Sarah:** In British Columbia, fentanyl went from about five percent of illicit drug overdose deaths in 2012 to ninety-four percent of opioid deaths in 2018, by two different measures. I find it hard to believe that people's reasons for using drugs changed that much in six years. The RAND study led by Bryce Pardo concluded that supplier decisions drove the move to fentanyl, and it found that problems in parts of Canada were as severe as in the eastern United States, despite very different drug policies. And Canadian reports describe most exposure to xylazine and nonmedical benzodiazepines as unintended.

**Kiffer:** I agree that the supply explains the timing and the toxicity. What I'd add is who gets exposed, and under what conditions. Ciccarone himself argued that demand mattered, because the prescription pill era created a population dependent on opioids, which sustained demand for heroin and then for whatever was sold as heroin. And question four showed that lost tolerance, and whether anyone is there to respond, shape who dies from the same supply.

**Sarah:** My worry is that once we talk about tolerance and circumstances, the conversation slides back toward the individual. The research on language in Lesson one showed how quickly a framing can turn into blame.

**Kiffer:** That's a fair worry, and it's the real risk on my side of the argument. The conditions I have in mind are things like release from custody, a discharge from hospital, or a supply that changed while someone was away. Those are circumstances that health and justice systems shape, and they can be changed.

**Sarah:** So where do you land?

**Kiffer:** I'd call the supply the main reason deaths rose so quickly over the past decade, and I'd still want any explanation of who died to include tolerance, timing and whether someone could respond.

**Sarah:** I'd put more weight on the supply than Kiffer does, but we agree on the practical rule. When someone explains a change in deaths with a single cause, ask two questions. What changed in the supply, and what changed in who was exposed and how? A good explanation needs evidence on both.

**Kiffer:** Let's pull it together with three things to take away.

**Sarah:** First, before you read a percentage of deaths, ask percent of what, whether the categories overlap, and what happened to the count underneath it.

**Kiffer:** Second, potency figures are comparisons with conditions attached, and in an unmeasured supply a change in a tiny fraction of the mass can double or triple the amount of opioid, though effects don't rise in step with the dose.

**Sarah:** And third, naloxone acts on one receptor. It can wear off before the opioid does, it can't reach the sedatives now common in the supply, and in the receptor-share model each extra unit lowers the opioid's share less than the one before. Tolerance and timing decide much of the risk, especially in the first week after release.

**Kiffer:** If you'd like more practice, rework the naloxone problem from this episode with a different target share, say twenty percent, and then put your answer back into the equation to check it.

**Sarah:** Next time, it's Lesson six, Stimulants, Caffeine, Tobacco and Vaping, which takes up the stimulants now involved in most opioid deaths in Canada.

**Kiffer:** Take care, everyone.

**Sarah:** See you in Lesson six.
