The discrepancy you are seeing represents a classic pitfall in pharmacokinetic (PK) modeling known as parameter unidentifiability due to sparse sampling.
The short answer is that both models can perfectly fit your single trough data point (resulting in a low Sum of Squared Errors and a favorable Akaike Information Criterion), but they make vastly different mathematical assumptions about what happens during the unmeasured distribution phase right after the drug is infused.
Vancomycin naturally exhibits multicompartment kinetics with a distinct distribution (α) phase that lasts 1 to 2 hours post-infusion.
Because AUC is the total integrated area under the concentration-time curve (∫ C · dt), the large, unmeasured concentration spike projected by the 2-compartment model right after each maintenance dose heavily inflates its calculated AUC relative to the 1-compartment model.
When sparse data (a single trough) is fed into PK software, the software relies almost entirely on its Bayesian population priors to guess the rest of the curve.
| Feature | 1-Compartment Model (with Trough-Only) | 2-Compartment Model (with Trough-Only) |
|---|---|---|
| Early post-infusion curve | Smooth, lower exponential curve | Sharp, high concentration spike (α-phase) |
| Calculated AUC | Moderately conservative (may underestimate by ~10%) | Frequently and significantly overestimated |
| Parameter stability | High; easier to identify with sparse data | Low; mathematically unidentifiable |
Pharmacokinetic literature explicitly warns that two-compartment models built from trough-only data should be avoided.
While vancomycin is biologically a 2-compartment drug, a 1-compartment model is significantly more reliable and clinically acceptable when dealing with sparse, trough-only data. To utilize a 2-compartment model accurately and capture the true AUC, a second data point is required — specifically a peak sample drawn 1–2 hours post-infusion — to anchor and verify the distribution phase.
Links open in a separate, closable popup window. Ctrl/Cmd-click, middle-click, or right-click → "Open link in new tab" to open a reference in a new tab instead.