Distinguishing Maturation from Size Effects

Why Separate Size From Development?

In children, body weight and age often increase together, while pharmacokinetic processes may also change with development. A higher clearance in an older child could therefore reflect greater body size, developmental change, or both. A comparison that does not account for these linked changes cannot determine which explanation is more consistent with the observed data.

Two useful comparisons are children of similar age but different weight, and children of similar weight but different age. The first holds age approximately constant while examining a weight contrast. The second holds weight approximately constant while examining an age contrast. These comparisons help distinguish patterns, but they do not by themselves prove that weight or age caused a difference.

Use a Common Weight Scale

Clearance is often expected to increase with body size. A power-function relationship represents that pattern: clearance is proportional to weight raised to an exponent. For the examples below, use the conventional allometric exponent \(0.75\) for clearance. The exponent is a modeling choice, not a universal biological law; the allometric material earlier in the course explains its role and alternatives.

A useful conceptual decomposition is \(CL_i=CL_{\mathrm{ref}}(W_i/W_{\mathrm{ref}})^{0.75}M(A_i)\exp(\eta_i)\), where \(A_i\) denotes age, \(M(A_i)\) represents an age-related factor without specifying its functional form, and \(\eta_i\) represents individual deviation. This expression is a way to reason about contrasts, not a complete model for every drug or population.

If age is held constant, the age-related factor cancels in a comparison, so the remaining systematic contrast is consistent with a size effect under the assumed model. If weight is held constant, the size factor cancels, so a remaining systematic contrast is consistent with an age-associated effect beyond size. In real data, individual variation and other covariates remain, so one pair of children is not enough to establish a general pattern.

Worked Comparisons

Assumptions and a Short Justification

The cancellation underlying these comparisons is simple but depends on the setup. Suppose the conceptual decomposition above applies, the allometric exponent is common to the children being compared, and other relevant covariates are sufficiently similar or accounted for. At the same age, the ratio is

because the shared \(CL_{\mathrm{ref}}\) and age factor cancel. At the same weight, the ratio instead becomes

because the reference clearance and weight factor cancel. These equations justify the comparison logic: matching on one variable removes its shared factor from the ratio. They also show why differences need not be entirely attributable to the other variable: individual deviations and unaccounted covariates can remain.

The logic is most useful when there is genuine overlap in the data: children with different weights at similar ages, and children with different ages at similar weights. “Similar” should be judged relative to the age and weight ranges and the expected rate of change, not treated as an arbitrary universal cutoff. If children are not comparable on important factors such as disease status, dose history, or concomitant treatment, the contrast can reflect those differences as well.

Diagnostics, Alternatives, and Limitations

A practical analysis should inspect the data before interpreting a contrast. Plot weight against age and mark the observations contributing to each comparison. Then inspect clearance against weight within age bands, and inspect standardized clearance against age among children with overlapping weights. A remaining trend after standardization may indicate an age-associated pattern, an unsuitable size relationship, or omitted covariates; the plot alone cannot decide among these explanations.

• Check overlap. If all older children are heavier than all younger children, a same-weight age comparison is unsupported by the observed data. Any such comparison would rely on extrapolation.
• Check the size adjustment. A residual relationship between standardized clearance and weight suggests the chosen exponent or size measure may not adequately describe the size pattern.
• Check influential observations and uncertainty. A few children may determine a narrow matched comparison. Report how many observations support it and avoid treating point estimates as exact.
• Check comparability. Similar age or weight does not guarantee similarity in illness, treatment, or other determinants of clearance.

Matched comparisons are transparent and help explain what a covariate adjustment is intended to do. A joint regression or population PK model can use more observations and adjust for age and size simultaneously, but its conclusions depend on adequate data support and appropriate model assumptions. A crude comparison of all younger with all older children is simpler, yet it combines age and size changes and is usually less informative for separating them. Neither a matched comparison nor a fitted model can resolve a distinction that the data contain too little overlap to support.

A common mistake is to see that older children have higher raw clearance and conclude that clearance increased because of maturation. That reasoning overlooks the fact that older children may also be heavier. The correction is to compare children of similar weight or to standardize clearance for size, then describe any remaining age pattern as age-associated rather than automatically causal. The reverse mistake is also possible: treating a successful size adjustment as proof that development has no role. A close match in one comparison does not rule out developmental change elsewhere in the age or weight range.