Daniel wants to know how a discipline that feels as permanent as gravity got built in living memory. Pharmacokinetics — what the body does to a drug — underpins every dosing schedule we've ever been handed, but it had to be invented, largely in the twentieth century, by people who decided you could describe the fate of a drug inside a body with equations. And then they had to convince actual clinicians that this wasn't a waste of time. Who were they, and how did they pull it off?
The prescription bottle says "take every eight hours" and it feels like Hippocrates carved that into stone. He didn't. The first pharmacokinetic equation was published in 1937, and the field wasn't even called pharmacokinetics until 1953. Before that, nobody had any real idea what happened to a drug after you swallowed it. Dosing was pure guesswork — trial and error, accumulated clinical folklore.
So we're talking about a discipline younger than the interstate highway system, and we've already built an entire regulatory state on top of it.
Right. And to understand why that's remarkable, you have to understand what pharmacokinetics actually is in one sentence: it's what the body does to a drug — absorption, distribution, metabolism, excretion. That's the ADME framework. The flip side is pharmacodynamics, what the drug does to the body. One is the journey, the other is the destination.
And the journey turns out to be mathematically describable. That's the leap.
That's the leap. And it was not obvious. The core tension that drove the whole field was this: clinicians had empirical dosing rules that worked okay — they knew digitalis accumulated, they knew alcohol got metabolized — but they had no way to predict what would happen if you changed the dose, or the interval, or the patient. If a seventy-year-old with failing kidneys needed the same drug as a twenty-year-old, you just... guessed. Pharmacokinetics promised prediction, and that was threatening to a medical establishment that prided itself on clinical intuition.
Nothing physicians love more than being told their intuition can be replaced by a differential equation.
The first person to try was a Swedish physiologist named Torsten Teorell. Nineteen thirty-seven. He published a paper called "Kinetics of Distribution of Substances Administered to the Body" in a Scandinavian journal, and it was almost completely ignored for a decade. Teorell proposed something radical: a two-compartment model. Imagine the body as two connected chambers. The central compartment is the blood and well-perfused organs — liver, kidneys, heart. The peripheral compartment is everything else — fat tissue, muscle, bone. You inject a drug into the central compartment, and it starts leaking into the peripheral one. Then it leaks back. Teorell wrote differential equations describing the rates of those leaks.
So he turned a human being into two buckets with a hole between them.
And that was the genius of it. The equations predicted something specific: if you measure blood concentration over time, you should see a biphasic curve. A rapid initial drop as the drug distributes into tissues, then a slower, steady decline as the body eliminates it. That shape — steep down, then gradual down — is now so familiar it's in every pharmacology textbook. Teorell predicted it from math before anyone had the tools to measure it properly.
Why was he ignored?
Several reasons. First, he was a physiologist, not a clinician. He published in a Scandinavian journal, not in English, and the war disrupted everything. But the deeper reason is that clinicians didn't think in equations. They thought in patients. Teorell's work looked like abstract mathematical play. And the tools to test his models — reliable blood assays, computing power — didn't exist yet. You couldn't measure drug concentrations at multiple time points with any precision in 1937. So the paper sat there, correct and useless, for a decade.
It's almost a perfect case study in how a scientific idea can be right but premature. I'm trying to think of an analogy — it's like someone in 1910 designing a working jet engine but having no aluminum alloys to build it with. The blueprint is sound, but the material reality isn't there yet.
Teorell had the conceptual framework, but he needed spectrophotometers that could detect drugs at microgram concentrations, and those didn't become routine until the 1950s. He needed computers to fit curves to data points, and those were still human beings with slide rules. The lag between theory and application was about fifteen years, which in the middle of the twentieth century might as well have been a century.
Until someone who actually treated patients picked it up.
Friedrich Hartmut Dost. German pediatrician. In the early 1950s, Dost independently rediscovered Teorell's work and immediately saw the clinical need. He was treating children, and children are not small adults. You can't just scale down the dose by weight and hope for the best — their kidneys and livers are developing, their body composition is different, their metabolism is different. Dost needed math to adjust doses properly, and he found Teorell's framework waiting for him. In 1953, he coined the term "Pharmakokinetik" and published a book that essentially founded the field.
He also gave us the concept of half-life as a clinically usable number.
Half-life wasn't new as a mathematical concept — it comes from physics, radioactive decay — but Dost turned it into something a doctor could use at the bedside. If a drug has a half-life of four hours, after four hours half is gone, after eight hours three-quarters is gone, after twelve hours seven-eighths. That simple rule replaced guesswork. He also introduced the area under the curve — AUC — as a measure of total drug exposure. Instead of just asking "how high does the concentration get," you ask "how much drug does the body see over time." That turned out to be the better predictor of both efficacy and toxicity.
Can we pause on the half-life thing for a second? Because I think most people, if they've heard the term at all, think of it as a property of the drug — like the drug has a half-life, the way an element has an atomic weight. But it's not. It's an interaction between the drug and the body.
Right. Half-life isn't stamped on the molecule. It emerges from two other numbers: volume of distribution and clearance. Volume of distribution is how widely the drug spreads through the body — if it mostly stays in the blood, the volume is small; if it sequesters in fat tissue, the volume is huge. Clearance is how fast the body eliminates it — through the liver, the kidneys, whatever. Half-life is proportional to volume divided by clearance. A drug with a large volume of distribution and slow clearance will have a long half-life. Change the patient's kidney function, and you change the half-life. The drug itself hasn't changed at all.
So when Dost gave clinicians half-life, he was giving them a number that looked fixed but was actually a moving target depending on who was taking the drug.
And that's the tension that runs through the whole field. The math looks clean, but the biology underneath is messy. Dost knew that — he was a pediatrician, he saw messy biology every day. But he also knew that a clean approximation was better than no approximation at all.
So Teorell built the mathematical skeleton, and Dost put clinical flesh on it. But there's still a gap between a German pediatrician publishing a book in 1953 and the FDA requiring pharmacokinetic data for every new drug approval.
That gap was filled by the Americans. In the 1960s, two figures turned pharmacokinetics from an academic curiosity into an engineering discipline. Sidney Riegelman at UCSF and John G. Wagner at Upjohn. Riegelman introduced the concept of biopharmaceutics — the idea that the formulation of a drug changes its pharmacokinetic profile. A tablet is not the same as a capsule is not the same as an injection, even if the active ingredient is identical. The excipients, the coating, the particle size — all of it affects how fast the drug dissolves and gets absorbed.
Which seems obvious now, but at the time it was a revelation that the pill itself mattered, not just the molecule inside it.
Wagner developed what's called the Wagner-Nelson method — a mathematical technique to calculate absorption rates from blood-level data alone. Before this, you had no way to know how much of an oral dose actually made it into the bloodstream. You could measure what went in and what came out, but the middle was a black box. Wagner-Nelson opened that box. It let you take blood concentration measurements at different time points and back-calculate the absorption rate at each point. Suddenly you could compare formulations — this tablet releases the drug twice as fast as that capsule.
And this is where the computational revolution comes in. Because these equations are not simple.
They're differential equations. Early pharmacokineticists solved them by hand — a single concentration-time curve could take days to compute. The arrival of mainframe computers in the 1960s and desktop computers in the 1980s made nonlinear regression fitting possible. You could take a set of blood concentration measurements, fit them to a compartment model, and estimate the rate constants — absorption rate, distribution rate, elimination rate — in minutes instead of days. That was the moment pharmacokinetics stopped being a niche academic pursuit and became something you could actually use in drug development.
I want to understand what "solving by hand" actually looked like. Are we talking about someone with a pencil and paper doing calculus for three days?
Yes. Literally. You'd have concentration measurements at maybe six or eight time points. You'd plot them on semi-log graph paper — concentration on the log scale, time on the linear scale. If the drug followed first-order elimination, the terminal portion of the curve would look like a straight line. You'd draw that line by hand, calculate its slope, and that gave you the elimination rate constant. Then you'd do something called "feathering" or "curve stripping" — you'd subtract that straight line from the original curve to reveal the distribution phase, which would also be a straight line on semi-log paper. Two straight lines, two rate constants. A skilled person could do it in a few hours. A computer does it in milliseconds.
So the entire field was bottlenecked by how fast a human being could draw a straight line through some dots.
And how consistently they could do it. Different people drawing lines through the same data would get slightly different slopes, slightly different half-lives. There's a whole literature from the 1960s and 1970s arguing about the "right" way to feather a curve. The computer didn't just make it faster — it made it reproducible.
And then the regulators noticed.
By the 1970s, the FDA was requiring pharmacokinetic data for new drug approvals. But the real institutional lock-in came with the 1984 Hatch-Waxman Act, which created the modern generic drug system. Hatch-Waxman said you don't need to repeat expensive clinical trials to prove a generic works — you just need to prove bioequivalence. The generic has to produce the same blood concentration curve as the brand-name drug, within a statistical boundary: the ninety percent confidence interval for the ratio of the means has to fall within eighty to one hundred twenty-five percent.
That's it. That's the legal definition of equivalence. Two curves that overlap within that window.
And that standard is entirely pharmacokinetic. It assumes that if the blood levels match, the clinical effects will match. That assumption holds up remarkably well for most drugs, but it's still an assumption. And it's why pharmacokinetics now feels like a permanent fixture of medicine — it's baked into the regulatory framework of every major drug market on earth.
When does it break down?
It breaks down when the relationship between blood concentration and effect isn't straightforward. Some drugs have active metabolites — the thing you measure in the blood isn't the thing that actually does the work. Some drugs bind irreversibly to their target, so the effect persists long after the drug is cleared from the blood. Proton pump inhibitors work that way — they shut down stomach acid pumps permanently, and the body has to make new ones. The blood level of the drug doesn't predict the duration of effect at all. For those drugs, bioequivalence based on blood levels is a useful fiction, but it's still a fiction.
So Hatch-Waxman essentially took a useful fiction and made it the law of the land.
It worked. Generic drug adoption in the US went from something like nineteen percent of prescriptions before Hatch-Waxman to over ninety percent today. That's hundreds of billions of dollars in savings. All built on Teorell's two buckets and Dost's half-life.
Let's talk about the people themselves. Who were they?
Outsiders. Almost to a person. Teorell was a physiologist, not a clinician. Dost was a pediatrician who taught himself mathematics. Wagner had a PhD in pharmaceutical chemistry and worked in industry, not academia. Riegelman was a pharmacist by training. They were often self-taught in math, and they faced real resistance from medical faculty who called their work "curve-fitting" or — I've seen this phrase in the historical record — "mathematical masturbation."
Charming.
The criticism wasn't entirely unreasonable. Early pharmacokinetics did involve a lot of curve-fitting — you'd take data, pick a model, and tweak parameters until the curve matched. Critics said this was just describing data, not explaining it. The models were oversimplified — two compartments can't capture everything that happens in a human body. And the early practitioners sometimes overclaimed what they could predict.
But they won anyway.
They won because they made predictions that held up. Teorell predicted the biphasic curve before anyone could measure it. When better assays arrived in the 1950s and 1960s, the curves looked exactly like his equations said they would. Dost showed you could optimize penicillin dosing using half-life calculations — give it every four hours instead of every six, and you maintained effective concentrations without wasting drug. That saved money and improved outcomes. Wagner's methods let generic manufacturers prove their products were equivalent without repeating full clinical trials, which made cheap drugs possible. The field won not by arguing theory, but by being useful.
By being right in ways that mattered to patients. Too little drug meant treatment failure. Too much meant toxicity. Empirical dosing was killing people at both ends.
There's a story about Wagner at Upjohn. He was asked to figure out why a particular drug formulation wasn't working — patients weren't responding the way the dose predicted. He ran the pharmacokinetics and found that the tablet was barely dissolving in the stomach. The active ingredient was fine, but the formulation was broken. They changed the excipients, fixed the dissolution, and suddenly the same dose worked. That's the kind of practical problem pharmacokinetics solved.
It's almost like a detective story. The crime is "the drug isn't working," and the detective is a guy with a differential equation and some blood samples.
The culprit wasn't the molecule — it was the pill. That distinction between the active ingredient and the delivery system is one of those things that's so obvious in retrospect that we forget it had to be discovered. Before biopharmaceutics, people thought a milligram of drug was a milligram of drug, regardless of how you packaged it.
By the 1980s, we've got a mature discipline. Teorell's two compartments, Dost's half-life and AUC, Wagner's absorption calculations, Riegelman's biopharmaceutics, all baked into FDA regulations through Hatch-Waxman. Case closed.
Except the models are still simplifications. Two compartments are better than one, but they're still a cartoon of human physiology. Real bodies have fat that changes with age, liver enzymes that vary genetically, kidney function that declines, drug interactions that compete for the same metabolic pathways. The next frontier is something called physiologically-based pharmacokinetics — PBPK modeling — where instead of abstract compartments, you model actual organs with actual blood flows and actual enzyme concentrations.
Which requires orders of magnitude more computation and data.
It's still early. But the promise is individualized dosing — not "take every eight hours" for everyone, but "take this amount at this interval based on your specific genetics, organ function, and concurrent medications." Teorell's 1937 vision of a fully predictive model might finally become real, just with a lot more variables than he ever imagined.
All of that math and regulation eventually trickles down to someone who has to actually draw the blood.
Hilbert: Nineteen eighty-eight. I spent two years drawing blood at a contract research organization in New Jersey that ran bioequivalence studies for generic drug companies. My job was to hit the time points. Zero hours, half hour, one hour, two, four, eight, twelve, twenty-four. Label the tubes. Put them in the centrifuge. If I was five minutes late on the two-hour draw, the whole study could be invalidated. I had no idea who Teorell or Dost were. I just knew that if I missed the four-hour window, the statisticians would call my boss and I'd be in trouble. The math was invisible to me — I was just the guy with the needle. But I was part of the machine that made those curves real.
What drug were you testing?
Hilbert: A generic blood pressure medication. Half-life of exactly six hours, which meant the dosing schedule was every twelve hours. The volunteers had to come back for a second day of draws. One of them never showed up. Ruined the entire dataset. Twelve volunteers, two days of blood draws, hundreds of labeled tubes, and the whole thing was garbage because one person decided to sleep in.
That's the fragility of real-world data collection. The math is elegant, but it depends on someone being punctual with a needle.
Hilbert: The protocol was unforgiving. You needed complete concentration-time curves for every subject. One missing point and the statisticians couldn't calculate the AUC. No AUC, no bioequivalence. No bioequivalence, no generic drug approval. The whole edifice — Teorell's equations, Dost's half-life, Wagner's absorption rates, the FDA regulations — it all rested on me not missing the two-hour draw.
You never missed it.
Hilbert: I was the most punctual person in America for two years. I still set my watch five minutes fast.
It's strange to think about. We treat dosing schedules as if they're laws of nature, but they're contingent on a chain of events that includes a Swedish physiologist who was ignored for a decade, a German pediatrician who taught himself math, and a lab tech in New Jersey who never hit snooze.
Hilbert: The tubes are still in a box somewhere. The unused ones. I kept a sleeve of them when the study ended. Vacutainers with the lavender tops. EDTA anticoagulant. Never opened.
I have to ask — what was the actual physical experience of doing those timed draws? You're sitting there with a stopwatch and a room full of volunteers. Walk us through it.
Hilbert: The study would start at seven in the morning. Volunteers arrive fasting. You take the zero-hour draw — baseline blood, before they get the drug. Then they swallow the pill, and the clock starts. The half-hour draw is chaos because you've got twelve people and maybe three phlebotomists, and everyone's veins are different. Some people are easy — you could hit their vein with your eyes closed. Others are what we called "tough sticks" — rolling veins, deep veins, veins that collapse as soon as you apply vacuum. By the time you finish the twelfth volunteer for the half-hour draw, it's almost time to start the one-hour draw on the first volunteer. For the first two hours, you don't sit down.
It's not just punctuality — it's physical skill under time pressure.
Hilbert: You're doing it in a windowless room with fluorescent lights. By hour four, everyone's tired. By hour eight, the volunteers are cranky because they haven't eaten all day — we could only give them standardized meals after certain time points. By hour twelve, you're drawing blood under what felt like battlefield conditions. And then they go home, sleep, and come back at seven the next morning for the twenty-four hour draw. If someone doesn't show up, you've lost not just that time point but every subsequent one for that subject. The whole curve is gone.
Where does that leave us, a century after Teorell's first equations? We've solved the problem of "how much drug is in the blood," but we still struggle with "what is the body doing to the drug in this specific patient." PBPK modeling tries to answer that by modeling actual organs instead of abstract compartments, but we're not there yet. Most dosing is still population-based — one size fits most, with adjustments for weight and kidney function if you're lucky.
The next step is probably real-time monitoring. Continuous glucose monitors already exist. Wearable sensors that track drug concentrations hour by hour are in development. Imagine a dosing schedule that adapts in real time — your insulin pump already does this, but extend that to antibiotics, chemotherapy, immunosuppressants. Teorell's 1937 vision of a fully predictive model might finally become real, just not the way he imagined it.
If you take one thing from this, it's that the dosing schedule on your prescription bottle was invented, not handed down. It took outsiders, equations, and a lot of punctual lab techs to make it work. That number — every eight hours, every twelve hours — is a human achievement, not a fact of nature.
It's fragile. It depends on the whole chain holding — from the mathematician's equation to the tech's stopwatch. Next time you take a pill, remember that.
This has been My Weird Prompts. Thanks to our producer Hilbert Flumingtop for keeping us on time, as always. If you want to think differently about something you take for granted, email the show at show at my weird prompts dot com. We'll be back soon.