# Lesson 3: How Drugs Work: Routes, Pharmacokinetics and Pharmacodynamics

*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. We'll be discussing overdose deaths, and support lines are listed on the lesson page. Lesson three keeps coming back to whether one number can tell you how dangerous a drug is. In 2004, the researcher Robert Gable divided the usual lethal dose of each of twenty substances by the dose people most commonly take outside medicine, and called the result a safety ratio.

**Sarah:** Alcohol came out at ten, MDMA at sixteen, and smoked cannabis at more than a thousand, a figure he marked as uncertain. So which of those three is the most dangerous drug?

**Kiffer:** The ratio can't answer that question. It describes one substance, taken alone, at a dose the person knows. In British Columbia's unregulated drug supply, a person can't know the dose, and in 2025 many of the deaths there involved more than one drug. That brings us to question one, about the number people quote most often. Fentanyl is fifty to one hundred times more potent than morphine. What does that actually tell you?

**Sarah:** I think most people hear it as fentanyl being fifty to one hundred times stronger, as if the effect were that much bigger, or the drug that many times more dangerous. I'll admit that's roughly how I heard it before I studied any pharmacology.

**Kiffer:** That's the common reading, and it mixes up two properties of the dose-response curve. Potency is how much drug you need for a given effect. Efficacy is the largest effect the drug can produce. Fentanyl and morphine are both full agonists at the mu-opioid receptor, so the potency figure tells you that fentanyl's curve sits far to the left of morphine's. A much smaller amount produces the same effect.

**Sarah:** And the drug with a lower ceiling is buprenorphine, the partial agonist. Its effect on breathing levels off at higher doses, and that's a difference in efficacy.

**Kiffer:** Right. So here's a question for everyone listening. If potency only tells you how much drug is needed for an effect, why should a highly potent drug in the unregulated supply raise the risk of overdose at all? Take a few seconds. Pause the audio if you'd like more time.

*(Pause)*

**Sarah:** My answer is that the person never measures the fentanyl itself. They measure a powder, and the fentanyl is a small share of it.

**Kiffer:** That's the heart of it, and Canadian drug checking data show how small and how variable that share is. Larnder and colleagues analysed four hundred fifty-four samples that people brought to a drug checking service in Victoria between November 2020 and July 2021, all described as opioids. Fentanyl was in ninety-six percent of them. Its concentration had a median of nine percent, and a standard deviation, which measures how spread out the values are, of seven percentage points.

**Sarah:** Let me work an example with those numbers. Suppose, hypothetically, that two portions of powder weigh exactly the same. One is four percent fentanyl and the other is twelve percent. Both sit within one standard deviation of the median. The second portion holds twelve divided by four, which is three times as much fentanyl.

**Kiffer:** So a person who repeats exactly what they did last time could take three times the drug with no visible difference. A threefold change would matter with morphine too. What potency adds is a question of scale. When the active amount is tiny, it's a small share of anything a person can see or weigh, so the amount of powder says very little about the amount of drug.

**Sarah:** And then there's the second drug. More than half of the Victoria samples contained another active drug the person didn't expect, most often etizolam, a sedative that acts like a benzodiazepine.

**Kiffer:** Which is why Larnder and colleagues concluded that the overdose crisis was better explained by the general variability of the market than by occasional bad batches. The fifty to one hundred figure is accurate, and it answers a narrow question about the amount needed for the same effect. For understanding risk in the unregulated supply, I'd say the numbers that do more work are how much the concentration varies and how often an unexpected second drug turns up. HSCI 402 develops the drug checking services that produce data like these.

**Sarah:** Question two. Half-life. I think the most common mistake is to treat it as the time a drug lasts. If injected naloxone has a half-life of thirty to eighty minutes, people assume it works for thirty to eighty minutes.

**Kiffer:** The half-life tells you how fast the level falls. How long the effect lasts also depends on how far above the effective level the peak starts. Here's a hypothetical drug to show it. Its half-life is two hours, it's absorbed instantly, and one dose gives a peak level four times the lowest level that produces any effect. How long does the effect last? Take a few seconds.

*(Pause)*

**Sarah:** Okay. The level halves every two hours. It starts at four times the threshold, so after one half-life it's at two times, and after two half-lives it's down to the threshold. So the effect lasts about four hours.

**Kiffer:** Good. And what happens when the dose is doubled?

**Sarah:** Now the peak is eight times the threshold. Eight falls to four, four to two, two to one. That's three half-lives, so six hours. Doubling the dose added two hours, which is a single half-life.

**Kiffer:** That's the pattern. Each doubling of the dose buys about one more half-life above the threshold. The same logic explains why an antagonist with a short half-life, such as naloxone, can wear off while a longer-acting opioid is still active, which is why people who respond to naloxone are monitored for hours afterward. Lesson five develops that.

**Sarah:** Steady state is where I want to try something harder, because I suspect I've had it slightly wrong. Take a hypothetical drug with a half-life of thirty-six hours, taken once a day. How high does the level climb compared with the first dose?

**Kiffer:** Go ahead and work it through out loud.

**Sarah:** Steady state takes four to five half-lives. So the doses keep stacking up for four or five half-lives, which means the level ends up about four to five times the level after the first dose.

**Kiffer:** Check that with the simplest case in the lesson, where a dose is taken once every half-life. What did the peak level off at there?

**Sarah:** Two. It went one, 1.5, 1.75, and then crept toward two. And that also took four to five half-lives. So I've mixed up how long it takes with how high it goes. Four to five half-lives is a time. It says nothing about the height.

**Kiffer:** Right. So what does set how high the level climbs?

**Sarah:** How much is left when the next dose arrives. If a dose comes every half-life, half is left, and the level settles where each dose just replaces what was lost, at twice a single dose. Here a dose comes every twenty-four hours, which is two thirds of a half-life. One half to the power of two thirds is about 0.63. So sixty-three percent is still there when the next dose comes, and thirty-seven percent is lost each day.

**Kiffer:** And at steady state, what's lost each day has to equal one dose.

**Sarah:** So the peak times 0.37 equals one dose, and the peak is one divided by 0.37, about 2.7 times the first day's peak.

**Kiffer:** That's right. And how long does the level take to get there?

**Sarah:** Four to five half-lives of thirty-six hours, so one hundred forty-four to one hundred eighty hours, which is six to seven and a half days. By the third dose the peak is about double the first one, and by the fifth dose it's within ten percent of where it settles.

**Kiffer:** So here's the trap the example hides. Someone judging a dose on the first day sees only about thirty-seven percent of where the level will end up. The same arithmetic applies after any change in dose, because the new level takes another four to five half-lives to settle. It's the same pharmacology behind clinical warnings that methadone can build up to harmful levels early in treatment, since its half-life runs from about eight to sixty hours. The thirty-six-hour drug in our example is hypothetical, and decisions about methadone treatment belong to the person receiving it and their prescriber.

**Sarah:** Question three is about enzymes, and it has a Canadian origin story I really like. In 1989, David Bailey and colleagues at a hospital in London, Ontario, were studying whether alcohol changed the effects of felodipine, a blood pressure drug. They gave the drinks in grapefruit juice, which hid the taste of the alcohol.

**Kiffer:** And the felodipine levels came out far higher than expected, and the authors suspected the juice. In a small follow-up published in The Lancet in 1991, six men with borderline high blood pressure took felodipine with water, grapefruit juice or orange juice. With grapefruit juice, the amount of felodipine reaching the circulation averaged two hundred eighty-four percent of the amount with water, nearly three times as much. Orange juice had no such effect.

**Sarah:** The same paper tested nifedipine, a related blood pressure drug. With grapefruit juice, its bioavailability was one hundred thirty-four percent of the amount with water, about a third higher.

**Kiffer:** So why would two related drugs respond so differently? Here's a worked example with two hypothetical drugs. Drug A has an oral bioavailability of ten percent, because the gut wall and liver break down ninety percent of each swallowed dose. Drug B has a bioavailability of ninety percent. Suppose something blocks enough of that breakdown to let an extra ten percentage points of each dose through. Which drug's level changes more? Take a few seconds.

*(Pause)*

**Sarah:** Drug A goes from ten percent to twenty percent. With a hypothetical dose of fifty units, that's five units reaching the circulation, rising to ten, so the amount doubles. Drug B goes from ninety percent to one hundred percent, forty-five units to fifty. That's a rise of about eleven percent.

**Kiffer:** The same ten points produce very different changes. A drug that loses most of a swallowed dose on the first pass has the most room to rise. Felodipine's prescribing information puts its bioavailability at about twenty percent, and nifedipine's is roughly forty-five to sixty percent, which is consistent with what the 1991 study found. In 2013, Bailey's group reviewed these interactions in the Canadian Medical Association Journal. The drugs that interact with grapefruit share three features: they're swallowed, they have very low to intermediate bioavailability, and they're broken down by the enzyme CYP3A4.

**Sarah:** So an enzyme inhibitor matters most for a drug the body already breaks down heavily. Does it work the other way for a prodrug like codeine?

**Kiffer:** The mechanism predicts that it does. Codeine needs CYP2D6 to turn part of each dose into morphine, so anything that slows that enzyme makes the person behave a little more like a poor metabolizer, and less morphine forms. For an active drug, inhibiting the enzyme that clears it raises the level. For a prodrug, inhibiting the enzyme that activates it lowers the effect. That's why it helps to ask which molecule is doing the work before predicting what an interaction will do.

**Sarah:** Question four. Expectation can change what a drug does, and students sometimes take that to mean the effects are imaginary, all in someone's head.

**Kiffer:** The study by Levine, Gordon and Fields in 1978 is a good answer to that. Naloxone increased pain after dental surgery only in the people whose pain a placebo had relieved, which points to the body's own opioids carrying the placebo relief. In that study, the expectation appears to have worked through opioid receptors. So an expectancy effect can be a physical effect that begins with what the person believes.

**Sarah:** I want to try the balanced placebo design with numbers, because its logic is easy to get backwards. These numbers are made up. Four groups do a taste test, and we measure how many millilitres each person drinks. People told they had alcohol who really got alcohol drank three hundred millilitres on average. Told alcohol but given tonic, two hundred ninety. Told tonic but given alcohol, two hundred. Told tonic and given tonic, one hundred ninety.

**Kiffer:** Here's the question for everyone. Which groups would you compare to measure the effect of expectancy alone, and how big is that effect? Take a few seconds.

*(Pause)*

**Sarah:** My first instinct is to compare the first group with the last, three hundred compared with one hundred ninety, a difference of one hundred ten. But those two groups differ in two ways at once, the belief and the drink, so that difference mixes both effects. To isolate expectancy, I hold the drink fixed. Among people who got alcohol, being told it was alcohol raised drinking from two hundred to three hundred. Among people who got tonic, it raised drinking from one hundred ninety to two hundred ninety. Both differences are one hundred.

**Kiffer:** And how would you measure the effect of the alcohol itself?

**Sarah:** Hold the belief fixed. Among people told alcohol, it's three hundred compared with two hundred ninety, a difference of ten. Among people told tonic, two hundred compared with one hundred ninety, also ten. So in these made-up numbers, belief moves drinking ten times as much as the drink does.

**Kiffer:** In direction, that matches what Marlatt and colleagues reported in 1973: the amount people drank followed what they believed was in the glass. Two limits are worth keeping in mind. The finding concerns one outcome, the amount drunk in a short taste test, so each other effect of alcohol needs its own test. And researchers who later extended the design noted that the original version was limited to low doses of alcohol, which matters because the whole design depends on people believing what they're told about the drink.

**Sarah:** Okay. Back to where we started, with Gable's safety ratios, because this is where we disagree.

**Kiffer:** Go ahead and make the case for them, then.

**Sarah:** My case is that they correct a picture most people carry around, in which a drug's legal status tells you how dangerous it is. In Gable's table, alcohol's ratio of ten is narrower than intranasal cocaine at fifteen or MDMA at sixteen, and intravenous heroin was the narrowest at six. The table is built from about three thousand records, and Gable listed his own cautions right alongside the numbers. For a first-year student, it opens up the question of why the law treats these substances so differently, which Lesson two traced through history.

**Kiffer:** I agree the table is a good prompt for that question, and I'd keep it in the course. My worry is what happens when one of these numbers travels without its conditions. Each ratio assumes one substance, a known dose, a person without tolerance, and acute poisoning only. Gable himself found that most of the lethal cases he reviewed involved a second substance, most often alcohol. And the heroin ratio describes heroin. In the Victoria drug checking data, fentanyl turned up in ninety-six percent of the samples people described as opioids.

**Sarah:** But every summary number leaves things out. Life expectancy hides enormous variation, and we still teach it. When a number has conditions, the answer is to teach the conditions with it.

**Kiffer:** That's fair, and it's the strongest version of your argument. Where I'd push is on the large ratios. A ratio of more than a thousand for smoked cannabis tends to be heard as safe, when it says nothing about long-term harms. Gable made the same point about alcohol and tobacco: given how widely they're used, direct acute deaths are rare, yet both are among the leading long-term causes of preventable death. And the ratio assumes no tolerance, when tolerance itself shifts with the setting, as Siegel's work on conditioned tolerance suggests.

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

**Kiffer:** I'd use safety ratios to answer one question. For a single substance, taken alone at a known dose by someone without tolerance, how big is the gap between a usual dose and a lethal one? I'd avoid using them to rank drugs by overall harm, and I'd be cautious about applying them anywhere the dose is unknown, which is the situation in the unregulated supply.

**Sarah:** I'm more enthusiastic about them than Kiffer is, but we agree on the practical rule. When you quote a safety ratio or a therapeutic index, state its four conditions along with it: one substance, a known dose, no tolerance, and acute effects only. Then ask which of them fails in the case in front of you.

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

**Sarah:** First, potency is the amount of drug needed for an effect. In an unregulated supply, it matters because the active drug is a small and variable share of what a person can see or measure.

**Kiffer:** Second, the half-life sets how fast a level falls and how long repeated doses take to settle, while the dose and the interval between doses set how high the level goes. Four to five half-lives is a time, and the height depends on how much is left when the next dose arrives.

**Sarah:** And third, every summary number in this lesson comes with conditions. Bioavailability shifts with enzymes and interactions, an estimate of expectancy depends on which groups you compare, and a safety ratio holds only for one known dose of one substance in a person without tolerance.

**Kiffer:** If you'd like more practice, rework the thirty-six-hour half-life problem from this episode with a dose every twelve hours, and check your answer by asking how much is left each time a new dose arrives.

**Sarah:** Next time, it's Lesson four, Alcohol and Other Depressants, where these tools meet alcohol, benzodiazepines and the risks of combining depressants.

**Kiffer:** Take care, everyone.

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