Why Compare Dosing Strategies on the Same Population?
A fixed dose gives each patient the same amount of drug, whereas a weight-based dose scales the amount with body weight. Neither strategy is inherently preferable. Their performance depends on how body weight relates to pharmacokinetics, on the target being evaluated, and on the distribution of body weights in the population expected to receive treatment.
A useful comparison applies both strategies to the same simulated patients. Each patient retains the same body weight and the same simulated pharmacokinetic characteristics under both strategies. This paired design makes the difference in outcomes attributable to the dosing rule rather than to differences between two independently generated cohorts. The comparison should report more than target attainment: a regimen can improve attainment while also increasing average or total dose, and an average exposure can conceal patients at both low and high extremes.
Set Up a Fair, Interpretable Simulation
First specify the target population, the candidate doses, the exposure measure, and the target interval. The simulated body weights should represent the population relevant to the dose decision. For each simulated patient, generate or specify body weight and the pharmacokinetic quantities needed to calculate exposure. Then calculate exposure twice for that patient: once using the fixed dose and once using the weight-based dose.
The assumptions must match the calculation. In the examples below, pharmacokinetics are linear over the doses considered, daily AUC is daily dose divided by clearance, and clearance is \(CL_i=0.02W_i\eta_i\) L/h. Here \(\eta_i\) is a dimensionless individual clearance factor; a value above 1 represents greater clearance than the weight-only prediction, and a value below 1 represents lower clearance. These examples treat the specified weights and \(\eta_i\) values as the simulated cohort, not as observations from a real study. They assume the same \(\eta_i\) applies under either dosing strategy and that the target is daily AUC from 400 to 600 mg·h/L.
In a full population simulation, the distribution of \(\eta_i\), other relevant covariates, and any parameter uncertainty should be chosen consistently with the intended prediction. The simulation estimates performance under those assumptions; it does not establish that the assumptions are true. The introductory material on probability of target attainment defines the broader target-attainment framework; here the emphasis is on comparing two dosing rules within a shared cohort.
Here \(s\) denotes a dosing strategy, \(N\) is the number of simulated patients, and \(I(\cdot)\) equals 1 when the condition is true and 0 otherwise. A paired estimate of the difference in attainment is \(\widehat{PTA}_{W}-\widehat{PTA}_{F}\), calculated from the two outcomes for the same \(N\) patients. Also summarize exposure extremes, such as the simulated minimum and maximum or prespecified percentiles, and dose burden, such as mean daily dose and total daily dose across the cohort.
How Weight Scaling Changes Exposure
Under the example clearance model, a fixed dose gives \(AUC_i^F=d_0/(0.02W_i\eta_i)\). Thus, with \(\eta_i\) held constant, predicted AUC decreases as weight increases. A weight-based dose gives \(AUC_i^W=kW_i/(0.02W_i\eta_i)=k/(0.02\eta_i)\), so weight cancels from this particular equation. That cancellation explains why weight-based dosing can reduce exposure variation when clearance scales proportionally with weight. It does not guarantee uniform exposure: variation in \(\eta_i\) remains, and the cancellation depends on the assumed relationship between clearance and weight.
Worked Example: Comparing Both Doses for One Patient
Consider a hypothetical patient weighing 70 kg with \(\eta_i=1.1\). Compare a fixed daily dose of 600 mg with a weight-based daily dose of 10 mg/kg. The target is 400–600 mg·h/L.
First calculate clearance: \(CL_i=0.02(70)(1.1)=1.54\) L/h. The fixed dose is 600 mg/day, so \(AUC_i^F=600/1.54=389.6\) mg·h/L. This is below the lower target boundary of 400 mg·h/L.
The weight-based dose is \(10(70)=700\) mg/day. Its exposure is \(AUC_i^W=700/1.54=454.5\) mg·h/L, which lies within the target interval. The same simulated patient therefore attains the target under the weight-based strategy but not under the fixed strategy. This single comparison illustrates a possible difference; it does not estimate population target attainment.
Worked Cohort Comparison
The following hypothetical cohort uses a fixed dose of 600 mg/day and a weight-based dose of 10 mg/kg/day. For every patient, clearance is calculated as \(0.02W_i\eta_i\) L/h. Both strategies use the same patient-specific weight and \(\eta_i\). Exposures are rounded to one decimal place for display; target classifications use the unrounded calculations.
Worked Example: Target Attainment, Exposure Extremes, and Dose Burden
| Patient | Weight (kg) | \(\eta_i\) | Fixed-dose AUC | Weight-based AUC |
|---|---|---|---|---|
| 1 | 40 | 0.8 | 937.5 | 625.0 |
| 2 | 50 | 0.9 | 666.7 | 555.6 |
| 3 | 60 | 1.0 | 500.0 | 500.0 |
| 4 | 70 | 1.1 | 389.6 | 454.5 |
| 5 | 80 | 1.2 | 312.5 | 416.7 |
| 6 | 90 | 0.8 | 416.7 | 625.0 |
| 7 | 100 | 0.9 | 333.3 | 555.6 |
| 8 | 110 | 1.0 | 272.7 | 500.0 |
| 9 | 120 | 1.1 | 227.3 | 454.5 |
| 10 | 130 | 1.2 | 192.3 | 416.7 |
For patient 8, for example, clearance is \(0.02(110)(1.0)=2.2\) L/h. The fixed-dose AUC is \(600/2.2=272.7\) mg·h/L, while the weight-based dose is \(10(110)=1100\) mg/day and its AUC is \(1100/2.2=500.0\) mg·h/L.
For the fixed strategy, patients 3 and 6 attain the 400–600 mg·h/L target. Thus \(\widehat{PTA}_F=2/10=0.20\), or 20%. For the weight-based strategy, patients 2, 3, 4, 5, 7, 8, 9, and 10 attain the target. Patients 1 and 6 have AUC of 625.0 mg·h/L, above the upper boundary. Thus \(\widehat{PTA}_W=8/10=0.80\), or 80%. The paired estimated difference is \(0.80-0.20=0.60\), a 60-percentage-point higher attainment proportion for the weight-based strategy in this simulated cohort.
The fixed-strategy exposure range is 192.3–937.5 mg·h/L; the weight-based range is 416.7–625.0 mg·h/L. The weight-based strategy narrows the extremes in this example, but it still produces two values above target. Dose burden also differs: the fixed strategy gives 600 mg/day to each patient, for a cohort total of 6000 mg/day and a mean of 600 mg/day. The weight-based doses sum to \(10(40+50+\cdots+130)=8500\) mg/day, for a mean of 850 mg/day. The attainment improvement therefore comes with a higher administered amount in this comparison.
The estimated difference is a description of this simulated cohort under its specified model. In Monte Carlo work, a finite simulated cohort also introduces simulation error: repeating the simulation with another random sample may yield a somewhat different estimated proportion. More simulations can reduce this numerical error, but cannot correct a misspecified weight-clearance relationship or an unrepresentative weight distribution.
Dose Caps, Diagnostics, and Alternatives
Worked Example: Evaluating a Maximum Dose
Suppose the weight-based strategy above is capped at 900 mg/day, so the administered dose is \(D_i=\min(10W_i,900)\). For a 120 kg patient with \(\eta_i=1.1\), clearance is \(0.02(120)(1.1)=2.64\) L/h. Without the cap, the dose is 1200 mg/day and AUC is \(1200/2.64=454.5\) mg·h/L. With the cap, the dose is 900 mg/day and AUC is \(900/2.64=340.9\) mg·h/L, below the target.
Applying the cap to the full cohort reduces the doses for patients weighing 100, 110, 120, and 130 kg to 900 mg/day. Their capped AUC values are 500.0, 409.1, 340.9, and 288.5 mg·h/L, respectively. Patients 1, 6, 9, and 10 are then outside the target interval, leaving six of ten patients attaining it: \(6/10=60\%\). The capped total daily dose is \(7500\) mg and the mean is \(750\) mg/day. In this illustration, the cap reduces dose burden relative to uncapped weight-based dosing, while reducing target attainment from 80% to 60%.
A practical comparison should check whether the conclusions are being driven by modeling or implementation details. Inspect simulated weights and covariates against the intended population; confirm units, dose calculations, and exposure calculations; and examine exposure distributions by weight strata as well as overall. Very low or high exposures concentrated in a subgroup can be concealed by an acceptable overall average. Also verify that a weight-based dose is not being treated as continuously deliverable if actual formulations require rounding, and represent caps or other dose rules explicitly when they are part of the candidate strategy.
A common mistake is to conclude that weight-based dosing is superior because it removes weight from the exposure equation. That reasoning is only valid under a particular clearance scaling relationship and does not account for individual clearance variability, target boundaries, dose caps, or dose burden. The correction is to simulate both strategies under the same patient-level pharmacokinetic model and compare the outcomes that matter, including exposure extremes and administered dose.
The main alternative is not limited to choosing between one fixed dose and one weight-based formula. One can compare several fixed doses, different weight-based coefficients, capped weight-based dosing, or a small set of weight bands. The alternatives should be evaluated on the same simulated population and against the same target and burden summaries. A fair decision also requires clarity about the comparison: two candidate strategies may be chosen to have the same mean dose, the same total drug budget, or different doses reflecting their intended use. These are different comparisons and can produce different conclusions.
This approach has important limits. The result is conditional on the simulated population, the pharmacokinetic model, and the target definition. If clearance does not scale proportionally with weight, the cancellation in the weight-based exposure equation will not hold. If the model omits clinically relevant sources of variability, the simulated extremes may be misleading. Finally, a higher simulated target-attainment proportion is not by itself a complete clinical decision: it must be interpreted alongside dose burden and the relevance of the selected exposure target.
Specify the target population, exposure target, candidate fixed and weight-based dose rules, and any caps or rounding rules.
Use the same body weight and patient-level pharmacokinetic characteristics for both strategies.
Compare target attainment, exposure extremes or distributions, and mean and total dose burden.
Inspect subgroup patterns and confirm that conclusions are not artifacts of a dose cap, a narrow simulated weight range, or an assumed clearance-weight relationship.
Check Your Understanding
Answer each question, then compare your reasoning with the solutions.
- In a paired simulation of 200 patients, 130 attain the target under a fixed dose and 150 under a weight-based dose. What are the two estimated attainment proportions and their difference?
- Under \(CL_i=cW_i\eta_i\), explain why a dose of \(kW_i\) removes the direct weight term from AUC but does not remove all exposure variability.
- Why is it potentially misleading to compare a fixed-dose strategy and a weight-based strategy using separate simulated cohorts?
- A weight-based strategy has higher target attainment but a higher mean daily dose. What should be reported before describing it as preferable?
Solutions. (1) The fixed-dose estimate is \(130/200=0.65\), or 65%; the weight-based estimate is \(150/200=0.75\), or 75%. The estimated difference is \(0.75-0.65=0.10\), or 10 percentage points. (2) AUC is \(kW_i/(cW_i\eta_i)=k/(c\eta_i)\); the weight term cancels, but variation in \(\eta_i\) remains. (3) Differences between independent cohorts can affect the comparison, so it may not isolate the dosing-rule difference. A shared cohort pairs each patient's outcomes under both rules. (4) Report the magnitude of the attainment improvement, the dose burden, and exposure extremes or distributions, and check that the simulation assumptions and target are appropriate. Higher attainment alone does not establish that a strategy is preferable.