# Lesson 7: Cannabis, Psychedelics and New Psychoactive Substances

*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. Cannabis potency is the first thing I'd like to look at for Lesson seven, on cannabis, psychedelics and new psychoactive substances. On a page last updated in 2023, Health Canada says the average THC content of dried cannabis rose from about three percent in the 1980s to around fifteen percent today. That's five times as much THC in each gram.

**Sarah:** Question one. If cannabis products have become so much stronger, are people actually taking in more THC?

**Kiffer:** Researchers have looked at that in people's own joints. In a study published in 2014, ninety-eight people in the Netherlands who used cannabis heavily brought their own cannabis and smoked it as they normally would, while researchers measured what was in it and how much smoke they inhaled. People whose joints held more THC inhaled less smoke. That adjustment is called titration, matching intake to the strength of the product.

**Sarah:** So they compensated for the stronger product.

**Kiffer:** Partly. The adjustment didn't fully make up for the extra THC, and people with stronger cannabis also tended to put more of it in each joint, so the authors concluded that people using stronger cannabis were generally exposed to more THC. It's one small study of heavy use, so I wouldn't stretch it far, but it's the kind of evidence that tells you whether a rise in concentration reaches people.

**Sarah:** And after the amount comes the route. The standard THC unit of five milligrams counts the THC in a product, and Lesson three showed that the share reaching the bloodstream depends on route and bioavailability. So a unit that's eaten and a unit that's inhaled don't produce the same exposure.

**Kiffer:** Yes. Eating cannabis also delays the onset and lengthens the effects, which the 2022 Lower-Risk Cannabis Use Guidelines note can lead people to consume more than they intended. Then tolerance, and the shape of the dose-response curve from Lesson three, sit between the level in the blood and the effect.

**Sarah:** Let's put numbers on the amount side. Under the regulations in force since October 2019, a package of edible cannabis can hold at most ten milligrams of THC, and a package of cannabis extract at most one thousand milligrams.

**Kiffer:** Here's a question for everyone, then. How many edible packages at their limit would hold the same amount of THC as one extract package at its limit? Pause the audio if you'd like more time. Take a few seconds.

*(Pause)*

**Sarah:** One thousand milligrams divided by ten milligrams is one hundred, so one extract package at the limit holds the THC of a hundred edible packages at theirs. In standard units, that's one thousand divided by five, or two hundred units, compared with two units in an edible package. And the same regulations cap a single unit of extract, such as a capsule, at ten milligrams, which matches the limit for an edible package.

**Kiffer:** That's right. Now, someone could say the extract package is a hundred times as strong. It holds a hundred times as much THC, which is a statement about the amount in the package, and it says nothing yet about effect, because route, tolerance and the dose-response curve all come between the amount and what anyone feels. Calling it a hundred times as strong turns an amount into an effect, and that's the step I'd always stop and question.

**Sarah:** Question two. The EU-GEI study, published in 2019, reports its results as odds ratios, and I want to be sure we're reading them correctly. What exactly are odds?

**Kiffer:** Odds are the chance that something happens divided by the chance that it doesn't. A one in five chance is odds of one to four, or 0.25. An odds ratio divides the odds in people with an exposure by the odds in people without it. When an outcome is rare, the odds and the chance are almost the same number. A one in a thousand chance is odds of one to nine hundred ninety-nine.

**Sarah:** And a first episode of psychosis is rare. In Amsterdam, the study's adjusted incidence was about thirty-eight new cases a year for every hundred thousand people.

**Kiffer:** For an outcome that rare, the odds ratio of 3.2 for daily use is close to saying that daily use went with about three times the chance of developing a psychotic disorder, compared with never using. It's still an association, and whether cannabis causes it is a separate question.

**Sarah:** Then let me try an attributable fraction myself, for daily use in Amsterdam. The odds ratio was 7.1. Daily use was reported by 13.1 percent of the controls and fifty-one percent of the cases. The formula is the share exposed, times the odds ratio minus one, divided by the odds ratio. 7.1 minus one is 6.1, and 6.1 divided by 7.1 is about 0.859. Times 0.131 gives about 0.113. So about eleven percent of cases.

**Kiffer:** The published figure for daily use in Amsterdam is 43.8 percent. Before you redo it, what does a gap that size tell you?

**Sarah:** That I picked the wrong share. I used the controls, 13.1 percent, and the formula needs the share of cases who were exposed. So it's 0.51 times 0.859, which is about 0.438, or 43.8 percent. That matches.

**Kiffer:** And the reason it's the cases is worth saying aloud. The result is a fraction of cases. The 0.859 is the share of exposed cases that the exposure would account for if it were a cause, so you multiply it by the share of cases who were exposed. The controls tell you how common daily use is in the population, and that number belongs in a different formula. Comparing an answer with a published figure, whenever one exists, is a habit I'd recommend to anyone.

**Sarah:** And Amsterdam has a second published fraction, the 50.3 percent for high-potency use that we met in the lesson.

**Kiffer:** Then here's a harder one for everyone. Suppose a headline adds 43.8 percent for daily use and 50.3 percent for high-potency use, and says that about ninety-four percent of first-episode psychosis in Amsterdam came from cannabis. What's wrong with adding them? Take a few seconds.

*(Pause)*

**Sarah:** The two exposures overlap. In Amsterdam, 69.6 percent of cases used high-potency cannabis and fifty-one percent used cannabis daily. Together that's 120.6 percent, more than everyone, so at least 20.6 percent of the cases, about a fifth, must have been in both groups. Adding the two fractions counts those people twice.

**Kiffer:** Right, and each fraction answers its own what-if question. Take away high-potency cannabis, and you also take it away from many of the people who used daily. There's a general rule behind it, too. Attributable fractions for different factors can add up to more than one hundred percent, because a single case can have several contributing causes, and removing any one of them might have prevented it. To estimate the effect of removing both exposures at once, you'd need a combined analysis, and the two published numbers can't give you that.

**Sarah:** And both numbers still assume that cannabis is a cause.

**Kiffer:** They do, and the fraction is sensitive to that assumption. Suppose, hypothetically, that confounding by something like genetic risk had inflated the association, so that the true causal odds ratio for high-potency use was two. Then the formula gives 0.696 times one, divided by two, which is about thirty-five percent. Self-reported use, and potency inferred from the type of cannabis, add more uncertainty. HSCI 214 takes up the wider epidemiology of psychosis.

**Sarah:** Question three. Synthetic cannabinoids act on the same CB1 receptor as THC. So why are they so much more dangerous?

**Kiffer:** Because two drugs can act on the same receptor in quite different ways, which Lesson three described with two terms, affinity and efficacy.

**Sarah:** Affinity being how tightly a drug binds, and efficacy how fully it switches the receptor on once it's bound.

**Kiffer:** Exactly. One early synthetic cannabinoid, JWH-018, binds CB1 with four times the affinity of THC. Let's use the simplified receptor-share model from Lesson five. The share of receptors a drug occupies is its binding units divided by its binding units plus one, and one binding unit is the concentration at which that drug alone would occupy half the receptors.

**Sarah:** So at the same concentration, a drug that binds four times as tightly has four times as many binding units.

**Kiffer:** That's how the model works. So here's a question for everyone. Take a concentration that equals one binding unit for THC. What share of CB1 receptors would JWH-018 occupy at that same concentration? Take a few seconds.

*(Pause)*

**Sarah:** For THC it's one binding unit, so one divided by two, which is fifty percent. JWH-018 has four binding units, so four divided by five, which is eighty percent. So is it 1.6 times as strong?

**Kiffer:** That number describes binding. It takes about a quarter as much JWH-018 to occupy the same share of receptors as THC, which is a statement about amount, and the share occupied doesn't turn one for one into an effect, as Lesson five warned for opioids. The bigger difference is efficacy. THC is a low-efficacy agonist, so even the receptors it occupies are only partly switched on, and at the receptor its signal tops out below what a full agonist can produce. Many synthetic cannabinoids are full agonists.

**Sarah:** That connects to the finding from the lesson that a CB1 blocker only modestly reduced the high from THC. Lu and Mackie read that as a sign that THC partly blocks the body's own stronger signal at CB1 as well as switching the receptor on weakly, which a low-efficacy agonist can do. And the gap in efficacy is why synthetic cannabinoids can have effects that differ in kind as well as in size, with agitation, seizures, high blood pressure and, rarely, heart attack and kidney failure.

**Kiffer:** Yes. And that ceiling for THC is at the receptor. Higher doses of THC still raise the risk of harm, which is what the evidence on potency and psychosis was about.

**Sarah:** The law has trouble with these substances too. Canada's drug law controls whole groups of substances defined by chemical structure, which lets it cover a substance before anyone finds it here. Health Canada's report on 2024 lists seven of the eleven new substances its Drug Analysis Service first identified that year as controlled. And yet a synthetic cannabinoid that the service first identified in 2023 wasn't controlled when Health Canada reported it.

**Kiffer:** That example shows what control status can and can't tell you. A substance outside the listing is no safer for it, since control describes chemistry and law. For people at risk, I'd say what matters first is how quickly a finding reaches clinicians, harm reduction services and people who use drugs, through detection, notification and assessment. In the European system, control is the last stage, and the slowest.

**Sarah:** Question four. Classic psychedelics have low physiological toxicity and aren't considered drugs of dependence. Does that make them low-risk?

**Kiffer:** It makes their risks different, and more dependent on context. Johnson and colleagues' 2008 safety review concluded that the main risks are psychological: distress during the drug's action, and rarely a prolonged psychosis, mostly in people with an existing vulnerability.

**Sarah:** And in research, prolonged psychosis was very rare. In a survey of researchers published in 1960, one psychotic reaction lasting more than two days occurred among twelve hundred research volunteers given LSD or mescaline.

**Kiffer:** Among research volunteers, it was, and Johnson and colleagues note that the one person affected would have been excluded under the screening they recommend. Their safeguards are screening out people with a personal or family history of psychotic disorders, preparing each volunteer, providing a safe setting and having at least two trained monitors present. Conditions like those are part of what a low rate in research describes.

**Sarah:** So the rate can't be carried straight over to someone using alone, with drugs from an unregulated supply whose contents they can't be sure of.

**Kiffer:** Right. The same molecule in a different setting carries different risks, and ketamine shows the same pattern with the bladder.

**Sarah:** And the estimates there range widely. A meta-analysis of recreational use reported lower urinary tract symptoms in forty-four to seventy-seven percent. Newer studies that found cases by different methods report two to twenty-seven percent, and a 2025 review of ketamine in psychiatric treatment found symptoms in zero to about twenty-five percent of patients, mostly mild or moderate. Why would the way a study finds cases matter that much?

**Kiffer:** Let's make up an example. Imagine, hypothetically, a thousand people who use ketamine regularly, and a hundred of them have urinary symptoms. That's ten percent. Now a study recruits through clinics and services, where people with symptoms are more likely to turn up. Suppose it reaches eighty of the hundred people with symptoms and a hundred twenty of the nine hundred without. What does the study report?

**Sarah:** The sample is eighty plus a hundred twenty, which is two hundred people, and eighty of them have symptoms. Eighty divided by two hundred is forty percent. That's four times the true ten percent, and nobody's bladder is any different.

**Kiffer:** That's one way the method of finding cases can push a figure up. A 2025 review judged the high recreational figures potentially misleading, and for people who need long-term ketamine treatment, it named higher doses, more frequent dosing, longer treatment and oral use as potential risk factors. Dose, frequency and context shape the risk in both settings. HSCI 485 takes up ketamine and the psychedelics as treatments.

**Sarah:** Now the part where we disagree. As of 2026, MDMA, known as ecstasy, sits in Schedule I of the Controlled Drugs and Substances Act with heroin and cocaine, where possession prosecuted by indictment carries a maximum of seven years. In Nutt and colleagues' 2007 expert ratings, ecstasy was in the bottom five of twenty substances for all three categories of harm, and alcohol and tobacco, both legal, scored higher on dependence than LSD or ecstasy. I think that mismatch is a real finding. A structured comparison of harms like this one should be one input whenever drug schedules are reviewed.

**Kiffer:** I'd use the ratings more cautiously, and the authors named several reasons themselves. These were judgements by experts in the United Kingdom in 2007, about drugs in the forms usually used there. Cannabis is often smoked with tobacco, which may have raised its scores, and the harms of ecstasy, ketamine and alcohol are tangled up with the use of other drugs. Legal status changes harm too, since easy availability shapes the scores for alcohol and tobacco. And all nine parameters were weighted equally and averaged.

**Sarah:** Those are limits of one study. A better-designed rating could fix most of them.

**Kiffer:** Some of them. I think the harder problem is the one the Toronto pharmacologist Harold Kalant raised in 2010: any ranking has to decide how much each kind of harm counts, and that's a value judgement. Here's a made-up example for everyone. Two hypothetical drugs are rated from zero to three on three parameters: acute harm, chronic harm and dependence. Drug A scores three for acute harm and zero on the other two. Drug B scores one on all three. If acute harm counts twice as much as each of the others, which drug ranks as more harmful? Take a few seconds.

*(Pause)*

**Sarah:** With equal weights, A is three divided by three, which is one, and B is three divided by three, also one, so they tie. If acute harm counts twice, the weights add up to four. A becomes six divided by four, which is 1.5, and B becomes two plus one plus one, four, divided by four, which is one. So A ranks as more harmful.

**Kiffer:** And if chronic harm counted twice, B would rank higher, at four divided by four, which is one, compared with three divided by four, or 0.75, for A. The arithmetic is simple, and the choice of weights decides the order. Averaging also blurs the real ratings. Alcohol's mean physical harm score was 1.4, lower than both its acute score of 1.9 and its chronic score of 2.4, because the mean also included harm from intravenous use, which doesn't apply to alcohol.

**Sarah:** But the current schedules contain value judgements too, and they're rarely set out as explicit weights. A published rating at least puts its weights where people can argue about them.

**Kiffer:** That's the strongest version of your argument, and I agree that stated weights are better than hidden ones. In 2011, three policy researchers added that a single score leaves out the benefits some drugs have, and treats harms as fixed when they change with the policy in force. So I'd use ratings like these to ask why a drug sits where it does. Before drawing a conclusion about any schedule, I'd want direct evidence on specific harms, like the cannabis review and the psychosis study, and some sense of what a change would do to supply and to people who use drugs. Lesson eleven looks at how the schedules were built.

**Sarah:** I'd still give the ratings more weight than Kiffer would, but we agree on the practical rule. When you meet a harm ranking, ask who did the rating, where and when, for which forms of use, and how the parameters were weighted. Then treat any mismatch with the law as a question that needs direct evidence before it settles anything.

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

**Sarah:** First, keep the amount of a drug separate from its effect. A percentage or a milligram count describes the product, and route, tolerance and how much a person uses all come between that number and what they feel.

**Kiffer:** Second, an attributable fraction is a fraction of cases. Build it from the share of cases who were exposed, remember that it assumes the association is causal, and keep fractions for overlapping exposures apart, because they can't be added.

**Sarah:** And third, ask what kind of strength or harm a number describes: how tightly a drug binds, how fully it switches a receptor on, a risk measured in one setting, or an expert's weighted judgement. Each one answers a narrower question than it first seems to.

**Kiffer:** If you'd like more practice, rework the attributable fraction problem from this episode with the figures for daily use across all eleven sites: 29.5 percent of cases, and an odds ratio of 3.2. Compare your answer with the published 20.4 percent, allowing for a small rounding difference, then repeat it with the 6.8 percent of controls to see how far that mistake moves the result.

**Sarah:** Next time, it's Lesson eight, The Neurobiology of Addiction: Reward, Learning and Neuroadaptation, where we look at what all these drugs share in the brain's reward system.

**Kiffer:** Take care, everyone.

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