Maturation Functions for Pediatric Clearance

Why Represent Maturation as a Function?

A child’s clearance may change with development, so applying one clearance value across all ages can misrepresent drug elimination. A maturation function expresses this age-related change as a fraction of a reference clearance. It gives a smooth, interpretable relationship between age and the fraction of mature clearance, while making clear which assumptions are built into that relationship.

This tutorial isolates the age-related multiplier. The clearance value to which it is applied is held as the specified reference; body-size scaling and its combination with maturation are separate considerations. The chosen age variable must have a clearly stated origin and unit, and must be appropriate to the population and developmental process being represented.

The usual interpretation assumes \(M(a)\) is nonnegative, increases with age, and approaches 1 as the modeled maturation process reaches its asymptote. These assumptions make the reference clearance the limiting clearance for this multiplier. They do not establish that every child reaches that value at a particular age, or that the value is appropriate if other modeled conditions differ.

Sigmoid Emax or Hill Maturation Function

A widely used form is the sigmoid Emax, also called Hill, maturation function. For age \(a\geq 0\), half-maturation age \(T_{50}>0\), and Hill coefficient \(H>0\), it is

Here \(T_{50}\) is the age at which the maturation fraction is one-half, and \(H\) controls the steepness of the curve around that age. The age and \(T_{50}\) must be expressed in the same units. The ratio raised to \(H\) is dimensionless, so \(H\) itself is dimensionless. If the model includes a nonzero fraction at birth or a different upper asymptote, it requires a modified function; those features are not represented by this simple form.

What the Parameters Mean

Setting \(a=T_{50}\) gives \(M(T_{50})=1/2\), because the numerator and the second term in the denominator are then equal. Thus \(T_{50}\) is a half-maturation age, not an age of complete maturation. As age becomes much larger than \(T_{50}\), the fraction approaches 1. A larger \(H\) makes the transition around \(T_{50}\) steeper; it does not change the fraction at \(T_{50}\) or the limiting values.

The inverse relationship helps answer questions such as “At what age does the model reach 90% of its asymptote?” Starting with a target fraction \(q\), where \(0<q<1\), solve the defining equation:

This is an algebraic property of the function, not a claim that a particular age is biologically universal. For example, the answer depends on the fitted or specified \(T_{50}\) and \(H\).

The sigmoid curve is monotone increasing under the stated parameter conditions. One way to verify this is to differentiate with respect to age:

The positive derivative shows that the function increases for positive ages. Also, \(M(0)=0\), while dividing numerator and denominator by \(a^H\) shows that \(M(a)\) approaches 1 as \(a\) increases without bound. These properties justify interpreting the function as an increasing fraction with a mature asymptote.

Exponential Maturation Function

Another standard form is the exponential maturation function:

The rate constant \(k\) must have units reciprocal to the age unit. For age in years, \(k\) is in year\(^{-1}\). At age zero the fraction is zero, and as age increases the fraction approaches 1. Its half-maturation age follows from \(1-\exp(-ka)=1/2\), giving \(a=\ln(2)/k\). Unlike the sigmoid form, this model has no separate Hill coefficient to control the steepness around a half-maturation age.

Choosing and Checking a Function

Both forms impose a smooth, increasing trajectory from zero to a limiting fraction of 1. The sigmoid form allows the location and steepness of the transition to be represented separately by \(T_{50}\) and \(H\). The exponential form uses one rate parameter and has a more constrained shape. Neither function is automatically preferable: the choice should reflect the scientific model and whether the available age range can support the parameters.

An age-band or other piecewise specification can represent abrupt changes, but it does not impose the same smooth developmental trajectory. More flexible curves may describe a wider range of shapes, but can require more data and may be harder to interpret. The comparison should be based on the intended biological representation, parameter support, and predictive performance—not on which curve gives the most appealing fit to a small sample.

Useful checks include confirming the age origin and units, plotting the maturation fraction over the observed age range, and checking that predicted clearances remain plausible. Inspect whether the data cover ages near the transition: if all observations are far below or above \(T_{50}\), the data may provide little information about its location. For a sigmoid model, \(T_{50}\) and \(H\) can be difficult to distinguish when the transition is sparsely observed. For an exponential model, a poorly supported rate constant can likewise make the inferred curve uncertain. A smooth curve can still be wrong if the true process is not monotone, does not begin at zero, or does not share the assumed asymptote.

Common Pitfalls and Limitations

• Calling \(T_{50}\) the age of full maturation. It is the age at which the sigmoid function equals one-half of its asymptote. Full maturation is not reached at a finite age in this simple function.
• Ignoring units. The age and \(T_{50}\) must use the same units, while \(k\) must be in inverse age units. Substituting age in months into a function parameterized in years changes the calculation unless the parameter is converted too.
• Interpreting the fraction as a measured biological percentage. \(M(a)\) is a model-defined multiplier. Its meaning depends on the chosen function and reference clearance.
• Assuming a good-looking curve proves the model. Sparse or narrow age coverage can leave the parameters weakly determined, and extrapolation beyond observed ages depends strongly on the assumed form.
• Confusing age maturation with every cause of clearance variation. This multiplier describes only the specified age relationship. It does not automatically account for body size, disease, or other determinants.

The central limitation is structural: each function prescribes a particular starting value, monotonic direction, and limiting behavior. If those features are not scientifically defensible, parameter estimates cannot repair the mismatch. The function should be interpreted within its stated population, age definition, and reference conditions.