A Hypothetical Renal-Clearance Model
A child’s renal clearance can differ from an adult reference for two distinct reasons: body size differs, and renal function matures with developmental age. A useful model represents these effects as multiplicative factors. This worked example uses hypothetical parameter estimates and patient characteristics; none of the values describe a real study or drug.
Suppose a nonlinear mixed-effects model fit to hypothetical data gives an adult-reference clearance of 6.00 L/h, a fixed allometric exponent of 0.75, a maturation half-age of 48 weeks postmenstrual age, and a Hill coefficient of 3.40. The exponent is treated as fixed here to keep the example focused on the joint size-and-maturation prediction. Postmenstrual age (PMA) is gestational age at birth plus chronological age after birth, expressed in weeks.
For the illustrative fit, \(CL_{70}=6.00\) L/h, \(PMAT_{50}=48\) weeks, and \(\gamma=3.40\). The maturation fraction is between 0 and 1 for positive PMA, increases with PMA, and equals 0.5 at 48 weeks. At ages far above 48 weeks it approaches, but does not exceed, 1. The reference clearance \(CL_{70}\) is therefore a mature-plateau reference, not the predicted clearance of every 70-kg subject at every age.
The formula gives a typical prediction, with the individual random effect set to zero. A corresponding mixed-effects specification could include \(\exp(\eta_{CL,i})\) as a multiplicative individual effect, where \(\eta_{CL,i}\) is a subject-level random effect. The calculations below omit that term; they are not individual predictions adjusted for a subject-specific random effect, nor do they include residual measurement error.
Why the Two Factors Multiply
The model uses a separable structure: the allometric factor adjusts the mature reference clearance for size, and the maturation fraction scales that size-adjusted value down when renal function is not yet mature. This structure assumes that size and maturation contribute through these specified factors rather than through an additional interaction or a separate age-dependent allometric exponent.
For two subjects with the same PMA, the maturation factor cancels in their typical-clearance ratio. For two subjects with the same weight, the allometric factor cancels. These cancellations make the model’s comparisons interpretable:
For example, doubling weight at the same PMA multiplies typical clearance by \(2^{0.75}\), approximately 1.682, not by 2. This is a direct consequence of the chosen exponent. At the same weight, a subject at the maturation half-age has half the typical clearance of a subject at the mature plateau, because \(F_{\mathrm{mat}}(PMAT_{50})=0.5\).
Worked Predictions Across Development
Worked Example: A Preterm Infant at 28 Weeks PMA
Consider a hypothetical preterm infant with a current PMA of 28 weeks and a body weight of 1.00 kg. The infant’s preterm birth history is relevant to the clinical context, but the model’s maturation input here is the current PMA. Calculate the typical clearance using the illustrative parameter values.
First calculate the allometric factor and the clearance at the mature-function plateau for this weight:
Next calculate the maturation fraction. The ratio \(28/48\) is dimensionless because both ages are in weeks:
Finally, multiply the weight-scaled plateau clearance by the maturation fraction:
The predicted typical clearance is approximately 0.0342 L/h. The small value reflects both the low body weight relative to 70 kg and a maturation fraction of about 0.138. It is not a clearance estimate for all preterm infants; it is the model prediction for these specified hypothetical inputs and parameters.
Worked Example: A Term Infant at 40 Weeks PMA
Now consider a hypothetical term infant at 40 weeks PMA with a body weight of 3.40 kg. This example changes both weight and PMA from the previous case, so the difference in predicted clearance cannot be attributed to either factor alone.
The weight-scaled plateau clearance is:
For maturation, first evaluate the power:
Combining these factors gives:
The typical prediction is approximately 0.2172 L/h. At 40 weeks PMA, the maturation fraction is below one-half because 40 weeks is below the model’s 48-week half-age. The result is lower than the subject’s weight-scaled plateau clearance of 0.6208 L/h.
Worked Example: An Older Child Near the Maturation Plateau
Suppose a hypothetical 8-year-old child weighs 20.0 kg. Using a gestational age of 40 weeks for this example, the child’s PMA is approximately 456 weeks: 40 weeks at birth plus 416 weeks of chronological age. The large PMA makes the maturation fraction close to its plateau.
The weight factor and corresponding plateau clearance are:
The maturation power is very large relative to 1:
Thus:
The prediction is approximately 2.344 L/h, nearly equal to the child’s weight-scaled plateau clearance. The maturation factor is almost one, but the clearance is still below the 6.00 L/h 70-kg reference because this child weighs 20.0 kg rather than 70 kg.
Calculation Checks and Interpretation
A useful way to check each prediction is to keep the two multipliers visible until the final step. The maturation fraction must lie between 0 and 1, so multiplying by it cannot increase the weight-scaled plateau clearance. In the three examples, the predicted values are each below their corresponding plateau clearances: 0.0342 is below 0.2479 L/h, 0.2172 is below 0.6208 L/h, and 2.344 is just below 2.345 L/h. A prediction above its plateau value would signal an arithmetic, unit, or implementation error under this model.
The components also clarify comparisons. The older child’s prediction is much larger than the preterm infant’s, but the examples differ in both weight and PMA. It would be incorrect to interpret the overall ratio as a pure age effect. To isolate one modeled effect, hold the other input constant or use the factor ratios derived above.
In an analysis, the parameter estimates are usually obtained jointly from the data by fitting the NLME model; the arithmetic here starts from hypothetical fitted values. The formula alone does not show that the estimates are precise or that the relationship fits well. A fitted curve should be assessed against the observed data and the model assumptions, including whether a common maturation curve is plausible across the weights and developmental stages represented.
Diagnostics, Alternatives, and Limits
Several patterns can suggest that this relationship is not describing the data adequately. Residuals or prediction errors that vary systematically with PMA may indicate that the maturation curve is too steep, too shallow, or has an unsuitable half-age. Systematic trends with weight may indicate that the allometric exponent or reference scaling is inadequate. Persistent underprediction in one age or weight range and overprediction in another is more concerning than random scatter around predictions.
Also inspect the coverage of the data. If there are few observations near the maturation transition, estimates of \(PMAT_{50}\) and \(\gamma\) may be weakly informed and strongly correlated: different combinations can yield similar curves over the observed ages. Predictions far outside the observed age or weight range then depend heavily on the assumed functional form. Physiological plausibility and the observed data should both be considered rather than treating a converged fit as proof of adequacy.
The principal alternatives answer related but different modeling needs:
| Approach | What it represents | Main limitation relative to this model |
|---|---|---|
| Weight allometry only | Clearance changes with weight, without an explicit developmental maturation multiplier. | May not capture low clearance associated with immature renal function at a given weight. |
| Maturation function only | Clearance changes with developmental age, without explicit weight scaling. | Does not account for size differences among subjects at the same PMA. |
| Combined allometry and maturation | Applies a weight adjustment and a bounded developmental multiplier. | Assumes the chosen factors and their multiplicative combination are adequate. |
| A different maturation function | Can represent a different shape, such as a linear or exponential increase over a selected range. | May impose unsuitable behavior near the age boundaries or outside the fitted range. |
The combined model is useful when both size and renal development are expected to matter, but it is not a complete physiological account of renal drug elimination. It does not automatically represent changes in renal function due to disease, fluid status, concomitant treatment, or other unmodeled covariates. Nor does a single maturation fraction distinguish multiple biological pathways. Those extensions require additional data and modeling choices, not merely a different interpretation of this calculation.
A Common Pitfall
A frequent mistake is to apply the maturation fraction to the reference clearance and forget to scale for weight, or to apply the allometric factor twice. The correct sequence is explicit in the model: calculate \(CL_{70}(WT/70)^{0.75}\), then multiply once by \(F_{\mathrm{mat}}(PMA)\). For the term-infant example, using maturation alone would give \(6.00(0.3498)=2.099\) L/h, which is not the model prediction for a 3.40-kg infant. Applying the weight factor as well gives the correct \(0.2172\) L/h.
A related interpretation error is to call \(CL_{70}\) the predicted clearance at every age for a 70-kg subject. It is the mature-plateau reference: at a specified PMA, the model predicts \(CL_{70}F_{\mathrm{mat}}(PMA)\) for a 70-kg subject with \(\eta_{CL}=0\). Reference values and age-specific predictions should not be conflated.
Use kilograms for weight and the same units for PMA and \(PMAT_{50}\).
Evaluate \((WT/70)^{0.75}\) and multiply by \(CL_{70}\) to obtain the weight-scaled plateau clearance.
Evaluate the sigmoidal fraction at the subject’s PMA and check that it lies between 0 and 1.
Multiply the two factors, report clearance in L/h, and state that the result is a typical prediction under the assumed model.
Check Your Understanding
Use the hypothetical model and parameter values stated above. Solutions are included so you can check both the calculation and its interpretation.
- At 48 weeks PMA, what is the maturation fraction, and why?
- At the same PMA, how does the model’s typical clearance change if weight doubles?
- For a 70-kg subject at 48 weeks PMA with \(\eta_{CL}=0\), what is the typical clearance?
- Why can the difference between the preterm and term examples not be described as an age effect alone?
Solutions. (1) The fraction is \(0.5\), because \((48/48)^{3.40}=1\), so \(F_{\mathrm{mat}}=1/(1+1)=0.5\). (2) With PMA fixed, maturation cancels in the ratio; doubling weight multiplies clearance by \(2^{0.75}\approx1.682\). (3) The weight factor is \((70/70)^{0.75}=1\), so clearance is \(6.00(1)(0.5)=3.00\) L/h. (4) The examples differ in both weight and PMA, so both the allometric factor and maturation fraction change; the contrast combines size and maturation effects.