Formulation and Administration Effects in Pediatric PK

Why Administration Conditions Belong in a Pediatric PK Model

A pediatric oral dose is not defined by the prescribed amount alone. A liquid may differ from a tablet in concentration, excipients, or absorption rate; food may alter the rate or extent of absorption; and a dose delivered through an enteral tube may be partly retained in the syringe or tubing. A child may also spit out part of a dose or receive it at a different time than recorded. If such conditions vary systematically between children or visits, the resulting concentration differences can be mistakenly attributed to age-related clearance.

The modeling task is to represent the processes separately. Formulation and administration can affect how much drug reaches systemic circulation, how quickly it enters, or when input begins. Clearance governs drug elimination after it reaches the systemic circulation. These processes may both change concentrations, but they do not generally leave the same signature in the concentration-time data.

Definition: Bioavailability, denoted \(F\), is the fraction of an administered dose that reaches systemic circulation, under a specified route and set of conditions. A formulation or administration effect is a systematic change in one or more absorption-related model quantities, such as \(F\), the absorption rate constant \(k_a\), or the absorption lag time \(t_{\mathrm{lag}}\). Clearance \(CL\) is the volume of plasma from which drug is eliminated per unit time; it is an elimination parameter, not an absorption parameter.

A Model That Separates Absorption From Elimination

A useful starting point for a single oral dose is a one-compartment model with first-order absorption and elimination. It assumes linear PK over the dose range being modeled, a single well-mixed distribution compartment with volume \(V\), first-order entry from the absorption site with rate constant \(k_a\), and first-order elimination with rate constant \(k=CL/V\). It also assumes the dose is represented correctly and that the formulation and administration conditions are constant over the dose interval. A lag time \(t_{\mathrm{lag}}\) allows input to begin after the recorded administration time.

C(t)= \begin{cases} \displaystyle \frac{F D k_a}{V(k_a-k)} \left[ e^{-k(t-t_{\mathrm{lag}})}-e^{-k_a(t-t_{\mathrm{lag}})} \right], & k_a\ne k,\\[6pt] \displaystyle \frac{F D k}{V}(t-t_{\mathrm{lag}})e^{-k(t-t_{\mathrm{lag}})}, & k_a=k, \end{cases} \qquad t>t_{\mathrm{lag}}, \quad k=\frac{CL}{V}.

Here \(D\) is the administered dose and \(C(t)\) is concentration at time \(t\). The formula applies after the lag; before the lag, the model predicts no drug input from that dose. For repeated doses, the contributions from previous doses are added, provided the model remains linear and the relevant dosing times and amounts are known.

The parameters have different roles. A lower \(F\) reduces the concentration profile proportionally, including AUC, if other parameters are held fixed. A lower \(k_a\) or a longer lag often delays and broadens the profile, changing peak concentration and peak time. A change in \(CL\) affects the elimination rate and, for a fixed systemic input, AUC. These patterns are useful clues, not proof of mechanism: sparse oral samples may not distinguish them.

Key result: Under linear PK with complete AUC measurement after a single oral dose, the dose-normalized exposure is \(AUC/D=F/CL\). Thus an oral AUC difference alone identifies a difference in the ratio \(F/CL\), not necessarily a change in \(F\) or \(CL\) separately.

Why Oral AUC Confounds Bioavailability and Clearance

The AUC result follows from mass balance. The amount entering systemic circulation is \(FD\). If elimination is linear, the elimination rate is \(CL\cdot C(t)\). Once the complete concentration-time curve has been observed, the total amount eliminated is \(CL\int_0^\infty C(t)\,dt\). Equating the amount eliminated to the amount that reached the circulation gives:

FD=CL\cdot AUC \qquad\Longrightarrow\qquad AUC=\frac{FD}{CL}.

Consequently, a comparison of oral AUCs can estimate relative bioavailability when clearance is effectively the same between the compared conditions—for example, in a suitable within-child crossover with stable health and dosing. If children of different ages are compared, clearance may differ with size and maturation, so attributing an AUC difference to formulation without accounting for clearance is not justified. An independent clearance estimate, an appropriate design, or strong external information may be needed to separate the effects.

Worked Example: Estimating Relative Bioavailability From AUC

In a hypothetical within-child comparison, the same child receives a 10 mg dose of a reference formulation and, after an adequate washout, a 10 mg dose of a liquid formulation. The complete AUCs are 4.00 mg·h/L for the reference and 3.40 mg·h/L for the liquid. Assume linear PK, the same clearance in both periods, comparable adherence, and no period-specific illness or interacting medication.

With equal doses and clearance, the ratio of AUCs estimates the ratio of bioavailabilities:

\frac{F_{\mathrm{liquid}}}{F_{\mathrm{reference}}} = \frac{AUC_{\mathrm{liquid}}/D}{AUC_{\mathrm{reference}}/D} = \frac{3.40\ \mathrm{mg\cdot h/L}}{4.00\ \mathrm{mg\cdot h/L}} =0.850.

Under the stated assumptions, the liquid has an estimated relative bioavailability of 0.850, or 85.0% of the reference formulation. This does not mean its absolute \(F\) is 0.850: the reference formulation may itself have incomplete bioavailability. Nor does the AUC ratio show whether the difference arises from absorption extent, dose loss, or another difference in systemic input. If clearance had changed between periods, the ratio would also reflect that change.

Representing Formulation and Administration Effects

A population model can include recorded formulation and administration conditions as covariates on absorption parameters. For example, let \(F_i\) be the bioavailability for observation \(i\), with \(z_{\mathrm{form},i}\), \(z_{\mathrm{feed},i}\), and \(z_{\mathrm{tube},i}\) indicators for specified formulation, feeding, and tube conditions. A bounded model for absolute bioavailability is:

\operatorname{logit}(F_i)= \beta_{F,0} +\beta_{F,\mathrm{form}}z_{\mathrm{form},i} +\beta_{F,\mathrm{feed}}z_{\mathrm{feed},i} +\beta_{F,\mathrm{tube}}z_{\mathrm{tube},i} +\beta_{F,\mathrm{form,feed}}z_{\mathrm{form},i}z_{\mathrm{feed},i} +\eta_{F,i}.

The logit link keeps \(F_i\) between 0 and 1. The interaction term allows the feeding effect to differ by formulation; it should be included when scientifically plausible and supported by data, rather than automatically. The random effect \(\eta_{F,i}\) represents between-individual variation under the chosen model. Analogous covariate models can be used for \(k_a\) and \(t_{\mathrm{lag}}\), often on a log scale for positive parameters. Coefficients then describe model-based changes on the selected parameter scale, not direct causal effects unless the study design and assumptions support a causal interpretation.

For an enteral tube, prescribed dose and dose delivered are not always the same. A useful extension is to represent the amount reaching the absorption site as \(D_{\mathrm{site}}=qD\), where \(q\) is the delivered fraction, and then represent absorption from that amount with \(F_{\mathrm{abs}}\). The net systemic amount is \(F_{\mathrm{abs}}qD\). If only plasma concentrations are available, \(q\) and \(F_{\mathrm{abs}}\) may be difficult or impossible to estimate separately; dose-preparation records, recovery measurements, or other information can help distinguish them.

Worked Example: Calculating a Peak Concentration Under First-Order Absorption

Consider a hypothetical child receiving an oral dose \(D=20\) mg. Suppose \(F=0.80\), \(V=10\) L, \(CL=2.0\) L/h, \(k_a=1.2\) h\(^{-1}\), and \(t_{\mathrm{lag}}=0.25\) h. Assume the one-compartment, linear, first-order model is appropriate and these parameters do not change over the dose interval. Then \(k=CL/V=2.0/10=0.20\) h\(^{-1}\).

For \(t>t_{\mathrm{lag}}\), the maximum occurs when the derivative of the concentration with respect to time since absorption began is zero. Writing \(\tau=t-t_{\mathrm{lag}}\), differentiation gives a peak at:

t_{\max}=t_{\mathrm{lag}}+ \frac{\ln(k_a/k)}{k_a-k} = 0.25+\frac{\ln(1.2/0.20)}{1.2-0.20} = 0.25+\ln(6) \approx 2.042\ \mathrm{h}.

At the peak, \(\tau\approx1.792\) h. Substituting into the concentration equation, the prefactor is:

\frac{F D k_a}{V(k_a-k)} = \frac{(0.80)(20\ \mathrm{mg})(1.2\ \mathrm{h}^{-1})} {(10\ \mathrm{L})(1.0\ \mathrm{h}^{-1})} =1.92\ \mathrm{mg/L}.

The exponential terms are \(e^{-0.20\ln(6)}\approx0.6988\) and \(e^{-1.2\ln(6)}\approx0.1165\). Their difference is approximately \(0.5824\), so:

C_{\max}\approx (1.92\ \mathrm{mg/L})(0.698827-0.116471) =(1.92)(0.582356)\ \mathrm{mg/L} \approx1.118\ \mathrm{mg/L} \approx1.12\ \mathrm{mg/L}.

The units are concentration, as expected. The calculation illustrates how a lag shifts the peak time, while \(F\), \(k_a\), and \(CL/V\) jointly determine the peak height. It does not establish that these parameter values describe any actual pediatric population.

Worked Example: Distinguishing an AUC Difference Due to Clearance

Suppose two hypothetical age groups receive the same 10 mg dose of the same formulation under comparable administration conditions. Their estimated clearances from suitable independent information are 2.0 L/h and 2.5 L/h. The observed oral AUCs are 4.00 and 3.20 mg·h/L, respectively. Assume linear PK, complete AUC measurement, and that the clearance estimates apply to the periods of concentration measurement.

For the first group, the implied bioavailability is:

F_1=\frac{AUC_1 CL_1}{D} =\frac{(4.00\ \mathrm{mg\cdot h/L})(2.0\ \mathrm{L/h})}{10\ \mathrm{mg}} =0.800.

For the second group:

F_2=\frac{AUC_2 CL_2}{D} =\frac{(3.20\ \mathrm{mg\cdot h/L})(2.5\ \mathrm{L/h})}{10\ \mathrm{mg}} =0.800.

The AUC ratio is \(3.20/4.00=0.800\), matching the inverse clearance ratio \(2.0/2.5=0.800\). Given the stated clearance information and assumptions, the AUC difference is consistent with equal bioavailability and higher clearance in the second group—not with an absorption difference. If clearance were not independently known or estimable, the oral AUCs alone would not support this separation.

Practical Design, Diagnostics, and Alternatives

The model is only as informative as the administration data. Record the actual formulation, concentration, dose preparation, route, tube type and flushing procedure where relevant, feeding status and timing, dose time, and whether the full dose was taken. Where possible, distinguish a prescribed dose from a dose believed to have reached the child. A covariate such as “fed” is ambiguous if the meal timing, content, or definition of feeding differs across observations.

Useful diagnostics include comparing observed and model-predicted profiles by formulation, feeding condition, and route; checking residuals against time after dose and against those covariates; and examining whether one condition systematically produces delayed peaks or biased predictions. Look for concentration records inconsistent with dose times, unexpectedly low exposure after tube administration, implausible estimated \(F\) or \(k_a\), and correlations between formulation and age that make their effects hard to separate. Sparse samples near the peak can obscure absorption-rate differences, while samples only in the terminal phase may provide little information about \(k_a\).

The main alternatives answer different questions. A simple categorical covariate model is often appropriate when the goal is to estimate average differences between observed conditions. A model with separate effects on \(F\), \(k_a\), and \(t_{\mathrm{lag}}\) can distinguish exposure extent from input timing when sampling and design support that distinction. A direct delivered-dose model is more appropriate when dose loss can be measured or credibly characterized. A noncompartmental comparison of dose-normalized AUC can summarize relative exposure under suitable conditions, but it does not by itself estimate absorption kinetics or resolve \(F\)-versus-\(CL\) confounding. These approaches are not interchangeable: choose based on the question, design, and information in the data.

Common pitfall: “The younger children had lower AUC, so their clearance must be higher.” That conclusion does not follow from oral AUC alone, because \(AUC=FD/CL\). Lower bioavailability, incomplete administration, or higher clearance can all lower AUC. The correction is to model or otherwise account for dose delivery and absorption conditions, and to use appropriate information on clearance before assigning the difference to developmental elimination.

This distinction matters in dose development because an apparent age trend can be created or exaggerated by different formulations, feeding practices, or routes of administration across age groups. Conversely, a real administration effect can be hidden if clearance differences are not considered. The aim is not to add every possible covariate: it is to use well-defined, biologically interpretable effects that the design can distinguish, and to report where the data leave mechanisms unresolved.

Worked Example: Accounting for Incomplete Tube Delivery

In a hypothetical administration assessment, a 10 mg dose is prescribed through an enteral tube. Measurement indicates that 8.8 mg reaches the absorption site, so the delivered fraction is \(q=8.8/10=0.88\). Suppose absorption from the delivered amount has relative bioavailability \(F_{\mathrm{rel}}=0.95\) compared with the reference administration. Assume linear PK and the same clearance under both conditions.

The relative systemic input factor, compared with the reference administration, is the product of the delivered fraction and relative bioavailability:

qF_{\mathrm{rel}}=(0.88)(0.95)=0.836.

Thus the relative systemic input is \(10\ \mathrm{mg}\times0.836=8.36\) mg relative to a reference dose with complete delivery and unit relative bioavailability. The absolute systemic amount also depends on the reference's absolute bioavailability \(F_{\mathrm{reference}}\): it is \(10\ \mathrm{mg}\times0.836\times F_{\mathrm{reference}}\), and equals 8.36 mg only if \(F_{\mathrm{reference}}=1\). For the same clearance, the dose-normalized AUC relative to that reference is 0.836. This calculation separates measured dose loss from the relative absorption effect. If \(q\) had not been measured, plasma exposure alone would generally identify only their product, not the two components separately.

Limitations and Interpretation

A one-compartment, first-order model is a starting approximation, not a universal description of oral absorption. Dissolution, gastric emptying, saturable absorption, enterohepatic recycling, food-dependent changes in drug solubility, or multiple absorption sites may require a different structure. In addition, pediatric feeding and administration practices can vary within a child over time. Treating such time-varying conditions as fixed individual traits can misrepresent their effects. If formulation and age are strongly associated—for example, if only the youngest children receive a liquid—the data may not separately identify both effects without overlap or external information.

A parameter estimate should therefore be interpreted conditionally: it describes the modeled population, study conditions, and assumptions. Check plausibility, uncertainty, and predictive performance before using it to inform a dose or administration recommendation. An estimated formulation effect is not automatically evidence that the formulation caused the observed difference; study design, co-occurring changes, and measurement quality matter.

Check Your Understanding

Answer each question, then compare your reasoning with the solutions.

Under linear PK, what can an oral AUC comparison identify if dose is known but clearance may differ?
A formulation change lowers \(k_a\) while \(F\) and \(CL\) remain fixed. Which features of the profile may change, and should total AUC change?
A 12 mg tube dose has a delivered fraction of 0.90 and relative absorption fraction of 0.80. What fraction of the prescribed dose reaches systemic circulation, and what systemic amount does that represent?
Why is a comparison of oral AUC between age groups not, by itself, proof of an age-related clearance difference?

Solutions. 1. It identifies \(F/CL\) through \(AUC/D=F/CL\), not \(F\) and \(CL\) separately. 2. Peak timing and peak height can change because the input rate changes; under the assumed linear model with complete AUC, total AUC remains \(FD/CL\), so it does not change when \(F\), dose, and clearance are fixed. 3. The relative exposure factor is \((0.90)(0.80)=0.72\). The absolute systemic amount is \(12\ \mathrm{mg}\times0.72\times F_{\mathrm{reference}}\), where \(F_{\mathrm{reference}}\) is the reference's absolute bioavailability; it is 8.64 mg only if \(F_{\mathrm{reference}}=1\). 4. Oral AUC also depends on bioavailability and delivered dose. Without accounting for them or using suitable independent clearance information, a lower AUC cannot be uniquely assigned to higher clearance.