# Lesson 1: Drugs and Society: Concepts, Patterns of Use and Language

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

---

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

**Kiffer:** And I'm Kiffer. Part of Lesson one is about what it means to call one drug more harmful than another, so here's a puzzle with made-up numbers. One substance leads to twice as many hospital admissions a year as another. The quick answer is that the first one is more dangerous, because its number is bigger.

**Sarah:** Question one. If a substance is legal, does that tell us it's less harmful? Before we get to the real studies, let's finish the puzzle, because the arithmetic matters.

**Kiffer:** Here are the full numbers, and they're hypothetical. The first substance is used by twenty million people and leads to two thousand hospital admissions a year. The second is used by fifty thousand people and leads to one thousand admissions a year. For one person who uses each substance, what's the chance of being admitted in a year, and how many times riskier is the second one? Take a few seconds. Pause the audio if you'd like more time.

*(Pause)*

**Sarah:** For the first, two thousand divided by twenty million is one in ten thousand. For the second, one thousand divided by fifty thousand is one in fifty. And ten thousand divided by fifty is two hundred, so the second substance is two hundred times riskier for the person using it.

**Kiffer:** Right. The first causes twice as many admissions in total, and the second is two hundred times riskier per person. Both statements are true at the same time. Now connect that to the harm ranking from the lesson. Nutt and colleagues scored alcohol highest overall, at seventy-two out of one hundred, ahead of heroin at fifty-five and crack cocaine at fifty-four.

**Sarah:** And when they split the scores, alcohol came out as most harmful to other people, while heroin, crack cocaine and methamphetamine came out as most harmful to the person using them.

**Kiffer:** That split fits the puzzle. Part of the reason alcohol causes so much harm in total is how many people drink. In Health Canada's 2025 survey, seventy-seven percent of people aged fifteen and older had drunk alcohol in the past year. So the study that put alcohol at the top of the overall ranking doesn't claim that a drink carries more risk for the drinker than a dose of heroin carries for the person taking it.

**Sarah:** Then there's the point Caulkins and colleagues made in their critique, that harm depends partly on the law itself. Alcohol is legal and sold almost everywhere, which raises the number of people who drink, and so its total harm.

**Kiffer:** And heroin is prohibited, so it's sold with no label and no quality control, and some of the risk to the person comes from not knowing its strength or what else is in it. A ranking measured under today's laws can't tell you what the harms would look like under different laws.

**Sarah:** They also pointed out that harm rankings leave out benefits. A prescribed opioid after surgery, or the caffeine in a morning coffee, would score some harm, and the ranking has nowhere to record what either one does for people.

**Kiffer:** Health Canada's spectrum does have a place for that, in its stage called beneficial use. A harm score describes a substance, and the spectrum describes how a person uses it.

**Sarah:** When students hear that drug categories are socially constructed, some of them hear that the categories are arbitrary. Is that fair?

**Kiffer:** Partly. The categories are made by people, and they're made for reasons. Some of those reasons are evidence about harm. Others come from history, culture and beliefs about who uses a substance, and Lesson two follows how anti-Asian racism shaped Canada's Opium Act of 1908. Calling a category constructed is an invitation to ask which reasons were at work.

**Sarah:** So if someone says a drug should be legal because it's less harmful than alcohol, I think the puzzle gives us three questions to ask. Harm to whom, the person using it or other people? Harm counted per person or in total? And harm measured under which laws?

**Kiffer:** Those are the three I'd ask too. A claim that can't answer them has squeezed a comparison with several dimensions into one number, and the same goes for the claim that a drug is illegal because it's the most dangerous.

**Sarah:** Question two. Where does lower-risk use end? I want to try this one myself, because I think it's where everyday intuition goes wrong.

**Kiffer:** Go ahead. Here's a hypothetical pattern. Someone drinks two tall cans of craft beer on Friday night and two more on Saturday night. Each can holds four hundred seventy-three millilitres, and the beer is seven percent alcohol. Where does that pattern sit under Canada's 2023 Guidance on Alcohol and Health, by the week and by the occasion? Take a few seconds.

*(Pause)*

**Sarah:** Okay. Two cans on each of two nights, so the weekly total is two times two, which is four drinks. Four falls in the moderate band, three to six. And each night is two drinks, which is right at the level the Guidance warns about going above. So moderate risk, and just inside the line on each occasion.

**Kiffer:** Before you settle on that, look again at what counts as one standard drink, and compare it with the can.

**Sarah:** Oh. I counted cans as drinks. A Canadian standard drink is 17.05 millilitres of pure alcohol, which is what's in a three hundred forty-one millilitre bottle at five percent. This can is bigger and stronger. Four hundred seventy-three times 0.07 is about 33.1 millilitres of alcohol. Divided by 17.05, that's about 1.94, so each can is almost two standard drinks.

**Kiffer:** Good catch. So now redo the week and the single nights with that number.

**Sarah:** Each night is two cans times 1.94, which is about 3.9 standard drinks. The week is twice that, about 7.8. So the pattern sits in the increasingly high risk band of seven or more, and each night is well above the two-drink level for a single occasion. My first answer was off by almost a factor of two.

**Kiffer:** It's an easy mistake, because people count containers, and containers vary a lot in size and strength. For comparison, under the 2011 guidelines, 7.8 drinks a week sat inside both weekly limits, ten for women and fifteen for men. On a single night, 3.9 drinks was above the 2011 limit of three for women and just under the limit of four for men.

**Sarah:** There's another pair of numbers people mix up. The Guidance links more than two drinks on one occasion to more risk, and binge drinking is usually defined as five or more drinks in one setting for men, or four or more for women.

**Kiffer:** Those are two different lines drawn for two different purposes. The binge definition is mostly used to count heavy drinking occasions in surveys and research. The Guidance says the risk of harm on any occasion starts to rise with any drinking, and that more than two drinks brings a significantly higher risk of harm to yourself and others, including injuries and violence.

**Sarah:** So someone with 3.9 drinks a night is above the Guidance's level and below the binge definition, and both statements are accurate. The part of this I find most interesting, though, is that the boundary moved in 2023 partly because someone chose a level of risk.

**Kiffer:** Right. The final report behind the Guidance explains that advice about voluntary activities usually treats a lifetime risk of premature death of about one in one thousand as low, while alcohol guidelines had often used about one in one hundred. To see what that means, picture a hypothetical group of one hundred thousand people who drink at the same level for their whole adult lives. A risk of one in one thousand means about one hundred premature deaths caused by alcohol in that group. One in one hundred means about one thousand.

**Sarah:** So the evidence draws the curve that links drinking to deaths, and people decide where on that curve to put the word low.

**Kiffer:** That's true of every boundary on the spectrum. The line between lower-risk and higher-risk use combines evidence with a judgement about acceptable risk. You can accept the evidence and still argue about the judgement, and that's a legitimate argument to have.

**Sarah:** Question three is about the far end of the spectrum. If someone is physically dependent on a medication, are they addicted?

**Kiffer:** Let's work a hypothetical case. A person has taken a prescribed opioid for chronic pain for a year, exactly as prescribed. They now need a higher dose to get the same relief, and they feel unwell when a dose is late. How many of the eleven DSM-5 criteria for substance use disorder does this person meet? Take a few seconds.

*(Pause)*

**Sarah:** My first instinct is two, tolerance and withdrawal. And two is the threshold for a mild disorder.

**Kiffer:** That's the tempting answer, and it's a trap. The DSM-5 criteria for opioid use disorder include a note that tolerance and withdrawal are not counted for people taking opioids solely under appropriate medical supervision. So this person meets none of the criteria. What they have is physical dependence, an expected effect of taking the medication over time. Lesson eight explains the difference in detail.

**Sarah:** And on Health Canada's spectrum, taking a medication as prescribed is the first example of beneficial use. So the spectrum and the diagnosis agree here.

**Kiffer:** They do. Now change the case. Suppose the same person starts taking extra doses and runs out early, has tried several times to cut back and couldn't, and has strong cravings between doses. What changes?

**Sarah:** That's taking more than intended, unsuccessful efforts to cut down, and craving. Three criteria, so at least a mild disorder. And since they're no longer taking it only as prescribed, a clinician would have to judge whether tolerance and withdrawal now count as well, which could make it moderate.

**Kiffer:** That's a careful answer. Tolerance and withdrawal describe how the body adapts to a drug, while the disorder is defined mostly by loss of control and continued use despite harm. When those get mixed up, a patient who takes a medication exactly as prescribed can end up described as addicted, which is the kind of label the last part of the lesson is about. Even the word addiction is used in two ways. Health Canada's spectrum treats addiction and substance use disorder as one stage, while the Canadian Centre on Substance Use and Addiction describes addiction as characteristic of severe disorder, so it's clearer to name the disorder and its severity.

**Sarah:** One more thing about this end of the spectrum confused me. In the 2025 survey, about four percent of people who'd used a substance in the past three months screened at high risk of dependence, which was nine hundred seven thousand people. And four percent of people reported needing professional help, which was 1.4 million. Same percentage, different numbers of people.

**Kiffer:** Because the denominators are different. Try working backward from each number to the group it's a percentage of.

**Sarah:** Nine hundred seven thousand divided by 0.04 is about 22.7 million, so that must be roughly the number of people who'd used any substance in the past three months. One point four million divided by 0.04 is thirty-five million, which is close to the whole survey population of 34.9 million.

**Kiffer:** Right, with one caution. That's our own back-calculation, and it's rough, because four percent is a rounded figure. Anything from about three and a half to four and a half percent rounds to four, so the first group could be anywhere from about twenty to twenty-six million people.

**Sarah:** And a screening result is a different thing from a diagnosis.

**Kiffer:** Yes. Health Canada notes that a screening test can't diagnose a disorder, and that the true number at high risk is probably higher, because people whose use comes and goes may not have used in the previous three months. So before you compare two percentages, ask what each one is a percentage of.

**Sarah:** Question four, with a short note first. This part talks about deaths from the toxic drug supply. Support lines, including the nine-eight-eight Suicide Crisis Helpline, the National Overdose Response Service and the Hope for Wellness Help Line, are listed in the content notes on the lesson page. Here's the question. The 2025 Canadian Substance Use Survey put past-year heroin use at about 0.1 percent of people aged fifteen and older. Yet the Public Health Agency of Canada counted fifty-seven thousand five hundred nine apparent opioid toxicity deaths from January 2016 to March 2026. How can both numbers be right?

**Kiffer:** Let's build a hypothetical region to see how. It has one million adults. Ten thousand of them, which is one percent, use opioids from the unregulated supply. Six thousand of those ten thousand have no working phone or stable address, or are in jail or prison, so a telephone survey can't reach them. Everyone else can be reached. The survey randomly samples twenty thousand of the people it can reach, and everyone answers honestly. What prevalence will the survey report, and will its confidence interval include the true one percent? Take a few seconds.

*(Pause)*

**Sarah:** The survey can reach one million minus six thousand, which is nine hundred ninety-four thousand people. Among them, four thousand use opioids from the unregulated supply. Four thousand divided by nine hundred ninety-four thousand is about 0.4 percent. So in a sample of twenty thousand, I'd expect about eighty people to report that use, and an estimate of about 0.4 percent.

**Kiffer:** Right. And with eighty people out of twenty thousand, the usual ninety-five percent confidence interval runs from roughly 0.3 to 0.5 percent. It looks precise, and it misses the true value of one percent by a wide margin.

**Sarah:** Because the interval only describes the uncertainty from sampling some of the people the survey could reach. The six thousand outside the frame were never in the draw, so no amount of sampling would find them.

**Kiffer:** Now add the step from the lesson where you turn a percentage into a number of people. A planner multiplies 0.4 percent by one million and gets four thousand.

**Sarah:** When the real number is ten thousand. And the six thousand who are missing are in the groups where harms tend to be highest, such as people without stable housing and people in custody, who face a high risk of drug-related death after release.

**Kiffer:** So the trap has two parts. The frame removes people before any sampling happens, and the narrow interval makes the result feel more certain than it is. Real surveys add another problem. Rare behaviours produce very few respondents. In 2025, past-year heroin use came out at about 0.1 percent, which in a sample of about thirty-six thousand means only a few dozen respondents, so even inside the frame the estimate is imprecise.

**Sarah:** So what do public health agencies use alongside surveys?

**Kiffer:** Several kinds of data, each with its own blind spot. Coroners count deaths. The BC Coroners Service reported one thousand eight hundred twenty-six unregulated drug toxicity deaths in British Columbia in 2025, a preliminary figure about twenty-one percent lower than the two thousand three hundred fifteen deaths of 2024. Drug checking shows what's actually in the supply. And wastewater testing measures what a whole city consumes.

**Sarah:** I didn't realize wastewater testing was done for drugs in Canada.

**Kiffer:** It is. In September 2024, Statistics Canada reported that levels of a fentanyl breakdown product in Metro Vancouver's wastewater were four to five times higher than in the other participating cities through 2022 and 2023.

**Sarah:** And wastewater can't tell you whether that's a few people using a lot or many people using a little.

**Kiffer:** Right, and deaths data can't tell you how many people use without dying, and drug checking only sees the samples people bring in. Each source answers a different question. Read together, they describe a pattern that no household survey could show on its own.

**Sarah:** Okay. Now the part where we disagree, and it's about the words for these deaths.

**Kiffer:** Overdose, drug toxicity or poisoning. You lean toward poisoning, so make the case.

**Sarah:** Overdose suggests the person took too much of something they understood, as if they misjudged a known dose. In the unregulated supply, people often can't know the strength of what they're taking or what else is in it. In May 2022, British Columbia's chief coroner at the time, Lisa Lapointe, told CBC News that the word overdose suggests that if people had just been more careful, they wouldn't have died.

**Kiffer:** That argument has real force, especially coming from a chief coroner. What else would you point to?

**Sarah:** The toxicology. Statistics Canada, citing the Public Health Agency of Canada, noted that fentanyl was implicated in more than eighty percent of apparent accidental opioid toxicity deaths in 2023. Poisoning puts the cause where the toxicology points, in the supply.

**Kiffer:** I agree about the cause, and I'm less sure about the word. The International Overdose Awareness Day statement from 2024 noted that in the United States, poisoning has been used to support expanding criminal penalties, including for people who supplied drugs socially or were simply present when someone overdosed. A word chosen to lift blame from the person who died can be used to place it on the people around them.

**Sarah:** That's a real risk, but it's a risk about how other people use the word, and it doesn't change what happened to the person who died.

**Kiffer:** There's also the question of fit. Think of a hypothetical person who knowingly bought fentanyl and died because one batch was much stronger than the last. That's harm from an unpredictable supply, and whether poisoning is the right word for it is itself a judgement. The toxicity terms used by the BC Coroners Service and the Public Health Agency of Canada place the cause in the drug without implying anyone's intent.

**Sarah:** But toxicity is a word most people never use. Overdose Awareness Day kept overdose in 2024 partly because it's so widely recognized. If the goal is to change how the public thinks about these deaths, a technical term won't reach many people, and poisoning is plain language.

**Kiffer:** That's a fair point about plain language, and the evidence from Kelly and Westerhoff gives your side some weight. If one label can shift trained clinicians toward blame and punishment, the word for a death might shift how the public assigns blame too. Where I'd hold back is that the experiments in the lesson tested labels for people. They didn't test these three words for deaths, so we're both reasoning from related evidence.

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

**Kiffer:** When I'm reporting numbers, I'd use the term of the agency I'm citing and say what it means, and I'd keep overdose for the physical event in the body, which Lesson five explains. When I'm writing for the public, I can see the case for poisoning, provided it's tied clearly to the supply.

**Sarah:** I'd reach for poisoning more often than Kiffer would, but we agree on the practical rule. Whatever term you choose, give your source's own term when you cite a number, say what that term means, and ask what cause and what response the word points toward.

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

**Sarah:** First, when anyone ranks drugs by harm, ask harm to whom, counted per person or in total, and measured under which laws.

**Kiffer:** Second, placing a pattern on the spectrum starts with counting the right thing. Count standard drinks by the alcohol they contain, count DSM-5 criteria with the note about medical supervision in mind, and check what each percentage is a percentage of.

**Sarah:** And third, a precise survey estimate can still be wrong about the people outside its frame, so read surveys together with deaths data, drug checking and wastewater testing.

**Kiffer:** If you'd like more practice, rework the hypothetical region from question four with a different number of people outside the survey's reach, and watch how far the estimate moves while the confidence interval stays narrow.

**Sarah:** Next time, it's Lesson two, Drugs in History: How Substances Became Licit or Illicit, where the Opium Act of 1908 and the Le Dain Commission come in.

**Kiffer:** Take care, everyone.

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