Case Study 2 — Why Drugs Reach a Plateau: Dose Accumulation as a Sequence

DataField.Dev

Case Study 2 — Why Drugs Reach a Plateau: Dose Accumulation as a Sequence

Field: Medicine and pharmacology Calculus used: Recursive sequences and fixed points (§20.7), the Monotone Convergence Theorem (§20.6), and a finite geometric-sum limit (preview of Chapter 21)

A patient is prescribed a pill "twice a day." Behind that ordinary instruction lies a question every pharmacologist must answer before the drug is ever approved: if the patient keeps taking doses, does the amount of drug in their bloodstream keep climbing forever — eventually to toxic levels — or does it level off at a safe, steady plateau? The answer is not obvious. Each new dose adds drug, and the body has not finished clearing the previous doses. Intuition might fear an ever-rising staircase. The reassuring truth, and the reason chronic dosing is safe, is that the blood concentration follows a recursive sequence that converges to a fixed point. This case study is that calculation — the same fixed-point and monotone-convergence machinery of §20.6 and §20.7, now written in milligrams.

The Body as an Eliminator

Pharmacokinetics rests on one empirical law: the body eliminates most drugs by first-order kinetics, meaning a fixed fraction of the drug present is cleared per unit time, not a fixed amount. This is exactly exponential decay. If a dose leaves concentration $C$ in the blood, then after one dosing interval a fraction $f$ remains, where $0 < f < 1$. The fraction $f$ is set by the drug's half-life relative to the dosing interval.

To keep the arithmetic clean, take the common and clinically realistic case where the dosing interval equals the half-life. Then exactly half the drug is gone when the next dose arrives, so

$$f = \tfrac12.$$

(Many real drugs — certain anticonvulsants, beta-blockers, and antibiotics — are dosed near their half-life for precisely this reason: it keeps the swing between peak and trough modest.)

Building the Recursion

Let $C_n$ denote the peak concentration just after the patient takes the $n$-th dose, and let each dose raise the concentration by a fixed amount $D$. The first dose gives

$$C_1 = D.$$

Between doses, the body clears the drug, so by the time the second dose is taken, only $f C_1 = \tfrac12 C_1$ remains — and then the new dose adds $D$ on top:

$$C_2 = D + \tfrac12 C_1.$$

The pattern repeats forever. Just after the $n$-th dose,

$$\boxed{\,C_n = D + \tfrac12\, C_{n-1}\,}, \qquad C_1 = D.$$

This is a recursive sequence of exactly the form $C_n = g(C_{n-1})$ from §20.7, with $g(C) = D + \tfrac12 C$. Let us track it with a concrete number — say each dose contributes $D = 4$ (mg/L of peak concentration):

# Peak concentration after each dose: C_n = D + 0.5 * C_{n-1}, with C_1 = D
D = 4.0
C = D
for n in range(1, 8):
    print(f"After dose {n}: C = {C:.4f} mg/L")
    C = D + 0.5 * C

# After dose 1: C = 4.0000 mg/L
# After dose 2: C = 6.0000 mg/L
# After dose 3: C = 7.0000 mg/L
# After dose 4: C = 7.5000 mg/L
# After dose 5: C = 7.7500 mg/L
# After dose 6: C = 7.8750 mg/L
# After dose 7: C = 7.9375 mg/L

The peaks climb — $4, 6, 7, 7.5, 7.75, \ldots$ — but the increments are halving each time: $+2$, then $+1$, then $+0.5$, then $+0.25$. The staircase is flattening. The sequence is doing precisely what the compound-interest sequence of Case Study 1 did: rising, but throttled, toward a ceiling.

Finding the Plateau: The Fixed Point

Because $g(C) = D + \tfrac12 C$ is continuous, §20.7 tells us that if the sequence converges, its limit $C^*$ must be a fixed point of $g$ — a concentration the dosing cycle leaves unchanged. Set $C^* = g(C^*)$:

$$C^* = D + \tfrac12 C^* \;\;\Longrightarrow\;\; \tfrac12 C^* = D \;\;\Longrightarrow\;\; C^* = 2D.$$

With $D = 4$, the plateau is $C^* = 8$ mg/L — exactly the value the numbers above are creeping toward. This is the famous clinical fact that steady-state concentration is roughly twice the increment of a single dose when you dose at the half-life. The plateau is real, finite, and predictable from a one-line fixed-point equation.

Proving It Actually Converges

A fixed point is only a candidate limit — §20.7's Common Pitfall warns that solving $C^* = g(C^*)$ does not by itself prove the sequence gets there. We owe a convergence argument, and we have two clean ways to give one.

Route 1 — the stability criterion. Here $g'(C) = \tfrac12$, so $|g'(C^*)| = \tfrac12 < 1$. By the stability criterion of §20.7, the fixed point $C^* = 2D$ is attracting: the error $e_n = C_n - C^*$ shrinks by a factor of $\tfrac12$ each dose, $e_{n+1} = \tfrac12 e_n$, so $e_n = e_1 (\tfrac12)^{n-1} \to 0$. Convergence is guaranteed, and it is geometric.

Route 2 — the Monotone Convergence Theorem. Independently, the sequence is increasing: if $C_{n-1} < 2D$, then $C_n = D + \tfrac12 C_{n-1} < D + \tfrac12(2D) = 2D$, and also $C_n - C_{n-1} = D - \tfrac12 C_{n-1} > 0$. So every term is below $2D$ (bounded above) and each exceeds the last (increasing). By the Monotone Convergence Theorem of §20.6, the sequence converges — and Route 1 already identified the limit as $2D$. Two independent paths, one plateau.

We can even write the limit in closed form, previewing the geometric series of Chapter 21. Unrolling the recursion gives

$$C_n = D\left(1 + \tfrac12 + \tfrac14 + \cdots + \tfrac{1}{2^{\,n-1}}\right) = D\cdot\frac{1 - (1/2)^n}{1 - 1/2} = 2D\left(1 - \tfrac{1}{2^n}\right),$$

which makes both facts transparent at a glance: the bracket rises toward $1$, so $C_n \nearrow 2D$, and the leftover gap is exactly $2D/2^n$, vanishing geometrically.

The Clinical Payoff: "Four to Five Half-Lives"

Doctors are taught a rule of thumb: a drug reaches steady state after about four to five half-lives. That rule is a theorem about this sequence. Ask when the concentration first reaches 95% of its plateau:

$$C_n \ge 0.95\,(2D) \;\;\Longleftrightarrow\;\; 2D\left(1 - \tfrac{1}{2^n}\right) \ge 0.95\,(2D) \;\;\Longleftrightarrow\;\; \tfrac{1}{2^n} \le 0.05.$$

Solving $2^n \ge 20$ gives $n \ge \log_2 20 \approx 4.32$, so $n = 5$ doses. Since each dose is one half-life apart, that is five half-lives — exactly the clinical rule, derived not from memorization but from the geometric decay of the error term $2D/2^n$. The same calculation explains the mirror-image fact that a drug is essentially gone (95% cleared) about five half-lives after the last dose, since the leftover then decays as $(1/2)^n$.

It also explains loading doses. If a critically ill patient cannot wait five half-lives for an antibiotic to build up, the clinician gives one large initial dose engineered to land near $2D$ at once, then switches to ordinary maintenance doses to hold the level. In the language of §20.7, a loading dose starts the iteration already sitting at its fixed point — so the sequence has nowhere to climb and the patient reaches therapeutic concentration immediately.

Why the Sequence View Matters

A purely intuitive clinician might fear that "drug plus leftover drug, forever" must accumulate without bound. The recursive-sequence model replaces that fear with a guarantee. Because the per-interval retained fraction $f = \tfrac12$ satisfies $|g'| = f < 1$, the iteration is a contraction: it converges, to the explicit value $D/(1 - f)$, geometrically fast. The boundary between a safe drug and a dangerous one that accumulates is, quite literally, whether $f < 1$ — whether the elimination process is a contraction in the sense of §20.7. The mathematics of this chapter is what lets a pharmacologist promise that "twice a day, indefinitely" will plateau rather than poison.

Discussion Questions

The plateau formula. For a general retained fraction $f$ (not just $\tfrac12$), the recursion is $C_n = D + f\,C_{n-1}$. Find the fixed point $C^* = D/(1-f)$ and verify it is attracting whenever $0 < f < 1$. What happens to $C^*$ as $f \to 1$ (a very slowly eliminated drug)?
Two routes, same answer. This study proved convergence both by the stability criterion ($|g'| < 1$) and by the Monotone Convergence Theorem. State the hypotheses each route needs, and explain why it would be a logical error to solve $C^* = g(C^*)$ first and call that a proof of convergence.
Choosing the interval. If a drug's half-life is 24 hours but it is dosed every 12 hours, then $f = (1/2)^{12/24} = 1/\sqrt2 \approx 0.707$. Recompute the plateau $C^* = D/(1-f)$ and the number of doses to reach 95% of it. Why does dosing more frequently than the half-life raise the plateau?
Loading dose. Using $C^* = 2D$ for the half-life-interval case, explain in sequence language what a loading dose of $2D$ followed by maintenance doses of $D$ does to the iteration $C_n = D + \tfrac12 C_{n-1}$.
Comparison to Case Study 1. Both the compound-interest sequence and the drug-accumulation sequence are increasing and bounded above, and both converge by the MCT. Identify one structural difference between them (for instance, the form of the recursion or the role of the index $n$).

A Short Annotated Reading

Rowland, M., and Tozer, T. N. (2011). Clinical Pharmacokinetics and Pharmacodynamics (4th ed.). Lippincott. The standard text; the chapters on multiple dosing and accumulation are this case study in clinical detail.
Strogatz, S. (2015). Nonlinear Dynamics and Chaos (2nd ed.). Westview. For the general theory of fixed points and stability that underlies the $|g'| < 1$ criterion used here.
OpenStax, Calculus, Volume 2 (Strang & Herman), §5.2 (Geometric Series). Connects the closed-form plateau $C^* = D/(1-f)$ to the geometric-series sum reached in Chapter 21.

A prescription label hides a recursive sequence. Because the body's elimination is a contraction — it retains a fraction less than one each interval — repeated dosing converges to a finite, computable plateau. The Monotone Convergence Theorem and the fixed-point criterion of this chapter are what make "twice a day, indefinitely" a safe instruction.