Worked Problem: Standardizing Clearance to a Reference Weight

Why the Reference Weight Matters

A clearance estimate reported at a reference weight is not, by itself, a prediction for every person. The reference weight specifies where the estimate is anchored in a body-size model. To use that estimate at another weight, apply the model’s weight relationship; do not simply carry the reference value over unchanged.

This worked problem uses a power model to convert a clearance estimate from one reference weight to another and then calculate predictions for three hypothetical individuals. The focus is on keeping the reference weight, exponent, units, and arithmetic consistent. The predictions are conditional on the specified model; they are not new measurements of those individuals.

Definition: A clearance power model anchored at reference weight $W_{\mathrm{ref}}$ is $$ CL(W)=CL_{\mathrm{ref}}\left(\frac{W}{W_{\mathrm{ref}}}\right)^b, $$ where $W$ is body weight, $CL(W)$ is clearance predicted at weight $W$, $CL_{\mathrm{ref}}$ is clearance at $W_{\mathrm{ref}}$, and $b$ is the weight-scaling exponent. The weights in the ratio must use the same units, so the ratio is dimensionless. If clearance is expressed in L/h, then both $CL(W)$ and $CL_{\mathrm{ref}}$ are in L/h.

The reference weight is an anchor for expressing the same relationship, not an additional biological effect. If the weight relationship and exponent stay fixed, the model can be rewritten at a different reference weight without changing any predicted clearance. This tutorial holds other covariates, such as maturation, fixed; the conversion does not estimate or alter their effects.

Deriving the Conversion

Suppose the model is initially written at reference weight $W_1$, with clearance $CL_{\mathrm{ref},1}$. At the new reference weight $W_2$, the clearance must equal the prediction from the original model evaluated at $W_2$. Therefore,

CL_{\mathrm{ref},2} =CL_{\mathrm{ref},1}\left(\frac{W_2}{W_1}\right)^b.

This is the conversion formula. It follows directly by setting $W=W_2$ in the model anchored at $W_1$. The resulting model is

CL(W)=CL_{\mathrm{ref},2}\left(\frac{W}{W_2}\right)^b.

Substituting the conversion formula into this expression verifies that it reproduces the original prediction:

CL_{\mathrm{ref},1} \left(\frac{W_2}{W_1}\right)^b \left(\frac{W}{W_2}\right)^b = CL_{\mathrm{ref},1} \left(\frac{W}{W_1}\right)^b.

The equality uses the rule $a^b c^b=(ac)^b$ for positive ratios. The two models therefore make the same prediction at every positive weight, provided the exponent is unchanged. A conversion changes the numerical value reported at the anchor, but not the fitted weight-clearance relationship.

Assumptions for this conversion: The original and converted values refer to the same clearance parameter and model; the same exponent $b$ applies throughout; both weights are expressed in the same units; and any other covariates or model components are held fixed. If the exponent or the definition of the modeled clearance changes, this conversion alone does not make the models equivalent.

Worked Example: Converting a 70 kg Estimate to 60 kg

Suppose a hypothetical population model reports $CL_{\mathrm{ref},1}=5.60$ L/h at $W_1=70$ kg, with fixed exponent $b=0.75$. Convert the clearance estimate to a 60 kg reference weight.

The weights are both in kilograms, the clearance parameter is unchanged, and the stated exponent is held fixed, so the conversion formula applies:

\begin{aligned} CL_{\mathrm{ref},2} &=5.60\ \mathrm{L/h} \left(\frac{60\ \mathrm{kg}}{70\ \mathrm{kg}}\right)^{0.75}\\ &=5.60\ \mathrm{L/h}\,(0.89082)\\ &=4.9886\ \mathrm{L/h}\\ &\approx 4.989\ \mathrm{L/h}. \end{aligned}

Here, $(60/70)^{0.75}\approx0.89082$. Thus the clearance at the 60 kg anchor is about 4.989 L/h. It is lower than the 70 kg reference value because the exponent is positive and the new reference weight is smaller. The units remain L/h because the weight ratio has no units.

Predicting Clearance at Individual Weights

Once the clearance is expressed at 60 kg, use the same model to predict clearance at each individual’s weight. The following calculations use the unrounded converted value, $4.9886$ L/h, before rounding the final predictions. Keeping unrounded values until the final reporting step limits avoidable discrepancies.

Worked Example: Predictions for Three Individuals

Using the hypothetical 60 kg reference clearance from the previous calculation and $b=0.75$, calculate predictions for individuals weighing 20 kg, 40 kg, and 80 kg. Assume that all other model components are held fixed.

For a 20 kg individual, the weight ratio is $20/60$, and $(20/60)^{0.75}\approx0.438691$:

CL(20) =4.9886\ \mathrm{L/h} \left(\frac{20\ \mathrm{kg}}{60\ \mathrm{kg}}\right)^{0.75} \approx4.9886(0.438691)\ \mathrm{L/h} \approx2.188\ \mathrm{L/h}.

For a 40 kg individual, $(40/60)^{0.75}\approx0.737788$:

CL(40) =4.9886\ \mathrm{L/h} \left(\frac{40\ \mathrm{kg}}{60\ \mathrm{kg}}\right)^{0.75} \approx4.9886(0.737788)\ \mathrm{L/h} \approx3.681\ \mathrm{L/h}.

For an 80 kg individual, $(80/60)^{0.75}\approx1.24081$:

CL(80) =4.9886\ \mathrm{L/h} \left(\frac{80\ \mathrm{kg}}{60\ \mathrm{kg}}\right)^{0.75} \approx4.9886(1.24081)\ \mathrm{L/h} \approx6.190\ \mathrm{L/h}.

The predicted clearances increase with weight because $b=0.75$ is positive. They do not increase in direct proportion to weight: the model uses a power of 0.75, not 1. The calculations are model-based predictions, and the displayed values are rounded to four significant figures.

Checking the Conversion by Changing the Anchor

A useful arithmetic check is to express the same model at a second reference weight and verify that the predictions agree. Starting from the 60 kg reference clearance, convert to 40 kg:

Worked Example: Verifying Predictions With a 40 kg Anchor

Use $CL_{\mathrm{ref},60}=4.9886$ L/h and $b=0.75$ to calculate the equivalent 40 kg reference clearance. Then check the predictions at 20 kg and 80 kg.

The converted reference clearance is

\begin{aligned} CL_{\mathrm{ref},40} &=4.9886\ \mathrm{L/h} \left(\frac{40\ \mathrm{kg}}{60\ \mathrm{kg}}\right)^{0.75}\\ &\approx4.9886(0.737788)\ \mathrm{L/h}\\ &\approx3.681\ \mathrm{L/h}. \end{aligned}

Now predict at 20 kg using the 40 kg anchor. The weight ratio is $20/40=0.5$, and $0.5^{0.75}\approx0.594604$:

CL(20) \approx3.6805\ \mathrm{L/h}(0.594604) \approx2.188\ \mathrm{L/h}.

At 80 kg, the ratio is $80/40=2$, and $2^{0.75}\approx1.68179$:

CL(80) \approx3.6805\ \mathrm{L/h}(1.68179) \approx6.190\ \mathrm{L/h}.

These agree, to the reported precision, with the 60 kg-anchor predictions. The 40 kg clearance used in these checks is carried as an unrounded value of approximately 3.6805 L/h. Small differences in the last displayed digit can occur if intermediate values are rounded earlier.

Diagnostics, Alternatives, and Limitations

Several checks help detect a conversion or prediction that has gone wrong:

Check the direction. With a positive exponent, moving the reference weight downward should lower the anchored clearance, and moving it upward should raise it. This is a directional check, not a substitute for calculation.
Check the ratio. The conversion ratio is new reference weight divided by old reference weight. Reversing the ratio reverses the conversion.
Check units. Both weights must use the same unit before forming the ratio. A ratio such as $60\ \mathrm{kg}/70\ \mathrm{kg}$ is dimensionless; mixing kilograms and pounds without converting is not.
Check the exponent. Use the exponent specified by the model. Changing it while changing the reference weight changes the model’s predictions, not merely its anchor.
Check predictions under both anchors. A large disagreement beyond rounding indicates a likely arithmetic, ratio-direction, or parameter-definition error.

Re-anchoring is an algebraic way to report the same power model. An alternative is to retain the original reference weight and calculate each prediction directly from that anchor. Both approaches are equivalent when they use the same exponent and model. Re-anchoring may be convenient when results need to be reported at a conventional weight, while direct prediction avoids an extra conversion step.

A different alternative is to use a per-kilogram normalization, which effectively imposes a linear weight relationship when clearance is calculated as a constant times weight. That is not generally the same model as a power relationship with exponent 0.75. The choice of relationship must come from the model being used; changing the normalization convention does not justify silently changing its exponent.

The conversion also does not establish that the power model is biologically correct, that the exponent is known without uncertainty, or that a prediction is accurate for a particular individual. It only translates the same specified model to a new anchor. In an applied analysis, the model’s fit and parameter uncertainty still matter; the earlier material on propagating uncertainty in allometric scaling addresses uncertainty in predictions.

Common Pitfall

Common mistake: “The estimate is 5.60 L/h at 70 kg, so the 60 kg reference estimate is $5.60(70/60)^{0.75}$, because 70 kg is the original reference.” The ratio is backward. To find the clearance at 60 kg from the 70 kg anchor, evaluate the original model at the new weight: $5.60(60/70)^{0.75}\approx4.989$ L/h. A quick check is that a positive exponent should give a lower clearance at 60 kg than at 70 kg.

A related reporting error is rounding the converted reference clearance and then mixing that rounded value with predictions calculated from an unrounded one. For example, $4.9886$ L/h may be displayed as $4.989$ L/h, but calculations should retain $4.9886$ (or greater precision) until the final results are rounded. This keeps comparisons between reference weights internally consistent.

Check Your Understanding

Try each problem, then compare your reasoning with the solutions below.

A model has $CL_{\mathrm{ref}}=8.00$ L/h at 80 kg and $b=0.75$. What is the equivalent clearance at 60 kg?
For the model in Question 1, what clearance does the 60 kg-anchored model predict at 40 kg?
Why must the exponent remain unchanged when converting a reference clearance if the goal is to preserve the model’s predictions?
A student converts from a 50 kg anchor to a 70 kg anchor using $(50/70)^b$. Identify the error and state the correct ratio.

Solutions. 1. $CL_{\mathrm{ref},60}=8.00(60/80)^{0.75}=8.00(0.805927)\approx6.447$ L/h. 2. First use the unrounded converted value: $CL(40)=6.4474(40/60)^{0.75}\approx6.4474(0.737788)\approx4.757$ L/h. 3. The conversion is derived by evaluating one fixed power model at the new reference weight; changing $b$ changes the weight relationship and generally changes predictions. 4. The ratio is reversed. For conversion from 50 kg to 70 kg, use $70/50$, so $CL_{\mathrm{ref},70}=CL_{\mathrm{ref},50}(70/50)^b$.