Why Divide Concentration by Dose?
Concentration-time profiles measured after different doses can be difficult to compare on the same scale: a larger dose may produce larger concentrations even when the underlying concentration response is otherwise similar. Dividing each concentration by its associated dose provides a simple exploratory view of concentration per unit dose. The resulting plot can make some dose-related patterns easier to see, but it changes the scale and can conceal differences that matter in the original concentration data.
Dose normalization is a plotting transformation, not a pharmacokinetic model and not a test of dose proportionality. Its meaning depends on which dose is used, whether profiles are comparable in time and regimen, and whether concentration is expected to scale proportionally with dose. The earlier material on dose-stratified concentration-time profiles provides a complementary view: it preserves concentrations in their original units while organizing them by dose.
The denominator must correspond to the concentration being interpreted. For a single-dose profile, this is usually the administered dose that preceded the sample. For repeated dosing, concentration can reflect several preceding doses. Dividing by only the most recent dose is not automatically meaningful; dosing schedule, interval, and accumulation need to be comparable before a simple dose-normalized display can be interpreted. This tutorial uses examples with a clearly specified dose and comparable sampling times.
What the Normalized Profile Can Show
If concentration is proportional to dose at a given time, dividing by dose removes the dose multiplier. A useful idealized relationship is
where \(g(t)\) is the concentration per unit dose at time \(t\), assumed to be the same across the profiles being compared. Under this relationship, \(C_i(t;D_i)/D_i=g(t)\), so normalized profiles should coincide at matching times, apart from sampling and measurement variability. This is a conditional expectation, not a rule that normalization will make real profiles overlap.
The shared function \(g(t)\) requires strong comparability. Among other things, the profiles should have compatible routes, formulations, sampling-time definitions, and dosing conditions. The concentration response should scale proportionally with dose over the range considered; dose-dependent bioavailability, absorption, clearance, or other relevant processes can break that relationship. Individual differences in pharmacokinetics can also make \(g(t)\) differ between subjects even when dose proportionality holds within each subject.
Specify the dose linked to each concentration and its units. Check whether the intended denominator is the administered dose, a weight-normalized dose, or another prespecified quantity.
For each record, divide the observed concentration by its specified positive dose. Keep the concentration and dose units explicit.
Use elapsed time defined consistently across profiles. Label the vertical axis with the ratio units and distinguish dose groups or subjects where useful.
Check whether the normalized display clarifies a pattern or hides important absolute concentrations. Treat visual similarity as exploratory evidence, not proof of a mechanism.
Calculating and Reading the Ratios
Worked Example: Comparing Profiles at Matching Times
Suppose three hypothetical subjects receive single oral doses of 50 mg, 100 mg, and 150 mg. At 1 hour, their measured concentrations are 2.0 mg/L, 4.0 mg/L, and 6.0 mg/L, respectively. Assume the samples are all collected 1 hour after dosing and the profiles use the same formulation and route.
For each observation, calculate concentration divided by dose:
The normalized values are identical at this time point. The raw concentrations increase with dose, while concentration per milligram is the same in these hypothetical observations. This single-time comparison is consistent with proportional scaling at 1 hour, but it does not establish that the full profiles overlap or that the same relationship holds at other doses, times, or subjects.
A normalized plot should retain the information needed to interpret it. Label the vertical axis with the ratio units, identify the dose definition, and state whether points represent individual subjects or summaries. Use a time axis based on a common reference, such as time since the relevant administration. A plot without these details can make ratios look comparable when their denominators or time origins are not.
Why Proportionality Produces Overlapping Ratios
The normalization has a simple algebraic justification. Suppose that at a fixed time \(t\), each concentration follows \(C_i(t;D_i)=D_i g(t)\) with the same \(g(t)\). Dividing both sides by the positive dose gives
Thus, under the stated relationship, the ratio no longer depends on dose. The key assumption is not merely that each subject received a recorded dose. It is that the expected concentration response is proportional to that dose under otherwise comparable conditions. If instead \(C_i(t;D_i)=D_i g_i(t)\), then the ratio is \(g_i(t)\); subject-specific differences remain. If the response is nonlinear in dose, division by dose generally does not remove dose dependence.
Worked Example: Normalization Hides an Absolute Difference
Consider hypothetical concentrations at two matching times for two dose groups. One group received 50 mg and the other 100 mg. At 1 hour, concentrations are 4.0 mg/L and 8.0 mg/L; at 4 hours, they are 1.0 mg/L and 2.0 mg/L. The ratios are:
The dose-normalized profiles overlap at these times. However, the raw concentrations are twice as high in the 100 mg group at both times. If, purely for illustration, 3 mg/L were a relevant concentration threshold, the 8.0 mg/L observation is above it while the 4.0 mg/L observation is also above it, but the later concentrations of 2.0 mg/L and 1.0 mg/L are both below it. The ratio plot alone does not show these absolute values or their relationship to a threshold. Therefore, a normalized plot should not replace the concentration-scale display when absolute concentration is the question.
When a Normalized Pattern Can Mislead
Worked Example: A Ratio Difference Does Not Establish a Dose Effect
Suppose two different hypothetical subjects are sampled 2 hours after dosing. Subject A received 50 mg and has a concentration of 1.0 mg/L; subject B received 100 mg and has a concentration of 4.0 mg/L. Their normalized concentrations are:
The ratio for the higher-dose subject is twice the ratio for the lower-dose subject. It would be incorrect to conclude from these two observations that dose caused a more-than-proportional concentration increase. The subjects may differ in bioavailability, clearance, absorption, or other characteristics; the observations may also differ in conditions not captured by the dose label. The calculation is correct, but the causal interpretation is unsupported. A normalized plot describes these records and can flag a pattern for investigation; it does not isolate a dose effect.
Other failure modes are more mechanical. If dose units are inconsistent, ratios are not comparable. If one profile is plotted against time since dose and another against time since study start, matching horizontal positions do not represent matching pharmacokinetic times. With repeated dosing, different intervals or prior-dose histories can produce different concentrations even at the same time after the latest dose. A sensible plot therefore begins with a check that both the numerator and denominator are defined consistently.
Censoring also matters. A concentration reported below the quantification limit is not an observed zero. Dividing a reported limit by dose gives a bound on the normalized concentration, not an exact ratio. Substituting zero or the limit as though it were the measured concentration can create artificial patterns. In addition, if measurement error is additive on the concentration scale, dividing by different doses changes the scale of that error: for an error term \(\epsilon\), the ratio error is \(\epsilon/D_i\). Consequently, low-dose ratios may show more spread from the same absolute concentration error.
Choosing the Display for the Question
Dose-normalized plots are most useful when the question is whether concentration profiles have a similar shape and magnitude per unit dose under comparable conditions. They can help reveal departures from proportional scaling that are difficult to see when high-dose concentrations dominate the vertical range. They are less suitable when the question concerns absolute concentrations, clinical thresholds, or differences between dose groups that should remain visible.
| Display | What it preserves | Useful question | Main caution |
|---|---|---|---|
| Raw concentration-time profiles by dose | Absolute concentration and dose-group differences | How do observed concentrations compare across dose groups? | Large concentrations can make smaller profiles harder to inspect on a shared scale. |
| Dose-normalized concentration-time profiles | Concentration per specified dose unit | Do profiles look similar after removing the dose multiplier? | Can hide absolute differences; overlap does not prove proportionality. |
| Subject-specific profiles with dose labels | Observed within-subject time pattern and dose context | Are patterns consistent within and between subjects? | Unequal sampling and subject differences still affect visual comparison. |
A practical exploratory report often presents the raw and normalized views together. The raw display protects against losing scale information; the normalized display can make per-dose patterns easier to compare. Neither view alone resolves confounding between dose and subject characteristics, and neither substitutes for an analysis designed to estimate dose-related effects.
Common Mistakes and Limitations
- Assuming the ratio is automatically comparable. A correct division does not make route, formulation, timing, regimen, or subject characteristics equivalent.
- Using an ambiguous denominator. “Dose” could mean dose per administration, daily dose, or dose per kilogram. State which quantity is used and keep its units consistent.
- Calling overlapping curves proof of dose proportionality. Overlap is a visual pattern in the displayed data. Sparse sampling, variability, and the selected subjects can all affect it.
- Interpreting a ratio difference as a causal dose effect. Differences between subjects or dosing conditions can produce different ratios. The plot alone cannot separate these explanations.
- Forgetting what normalization removes. Dividing by dose can conceal absolute concentration differences that may matter for exposure or decision-making.
Dose-normalized plots are therefore descriptive tools with a narrow but useful role. Their strongest interpretation is conditional: given a consistent dose definition and sufficiently comparable profiles, the ratios show how observed concentration varies per unit dose. The plot can suggest questions about proportionality, but its visual pattern is not itself a model-based estimate or proof.
Check Your Understanding
Answer each question using the definitions and cautions above.
- A hypothetical sample has concentration 3.0 mg/L after a 75 mg dose. Calculate its dose-normalized concentration and give the units.
- Two profiles have identical concentration-per-dose ratios at every displayed time. What does that show, and what does it not establish?
- Why can dividing by the most recent dose be misleading for a concentration measured during repeated dosing?
- Two subjects have different normalized concentrations at the same nominal time, but received different doses. Give one reason this does not by itself establish a dose-dependent pharmacokinetic process.
- What should be plotted or reported alongside a dose-normalized display when absolute concentration differences matter?
Solutions. (1) \(3.0/75=0.0400\ \mathrm{L}^{-1}\). (2) It shows that the displayed ratios coincide for those observations; it does not prove dose proportionality generally, establish a mechanism, or rule out differences outside the observed subjects, times, or conditions. (3) A repeated-dose concentration can reflect prior doses and accumulation, not only the latest administration; the regimen and dosing history need to be considered. (4) The subjects may differ in clearance, bioavailability, absorption, or another characteristic, so the dose groups are not necessarily comparable. (5) Show the raw concentration-time profiles with their concentration units and dose context, so absolute values and differences remain visible.