Vancomycin TDM · external comparison

DoseMeRx comparison sheet

For capturing paired outputs at the workstation, so divergence between the two tools can be traced to a layer rather than argued about. De-identified inputs only.

No PHI. Use synthetic or fully de-identified demographics. Nothing here should identify a patient — no MRN, name, date of birth, admission date, or real dose timestamps tied to a person. Shift the clock to an arbitrary day-1 and keep only the intervals, which is all the pharmacokinetics needs. Everything you type stays in this browser's local storage; nothing is transmitted.

Before you start: four things to pin down

A comparison that does not control these will show differences that are artifacts, and you will not be able to tell which layer disagrees.

Control	Why it matters	What to do
Model pairing	The two tools must be running the same published model, or you are comparing papers, not implementations.	DoseMeRx Standard Adult (1-comp) = Buelga 2005 → select Buelga 2005. DoseMeRx Complex / Critically Ill (2-comp) = Goti 2018 → select Goti 2018. DoseMeRx Enhanced Obese = Sabourenkov 2019, unpublished and proprietary. There is no equivalent here — Hughes 2024 is a different model. Expect divergence; it is not a defect.
Creatinine clearance	Cockcroft-Gault weight basis and any SCr floor may differ. If CrCl differs, everything downstream differs, and you learn nothing about the Bayesian fit.	Record DoseMeRx's displayed CrCl, then force the same number into this calculator with the manual CrCl override. That isolates the fitting layer from the covariate layer. Run it both ways if you want to measure the covariate difference too.
Horizon	DoseMeRx simulates a finite window — its alternative-regimen matrix is stated "over N days" and its own numbers confirm it (1500 mg q12h reads 859, not 2 × 456.69 = 913, because accumulation has not settled). This calculator reports steady state.	Record the Regimen days field. A 5–10% AUC gap on a short horizon is expected and is not a disagreement.
Target	Both default to AUC₂₄ 450, but a site protocol can change theirs.	Read the target off their screen and enter the same one here.

Also worth recording, because it is a claim we can test. DoseMeRx's own model guidance says the 2-compartment model "does not generally work as well with single levels or when levels are troughs", and recommends two levels at different points initially. If you can, capture at least one case with a single trough on the 2-comp model — that is where the two tools are most likely to part company, and where knowing who is right matters most.

Cases

Fill the left column from DoseMeRx and the right from this calculator. Differences compute as you type. A case is only interpretable if the controls above match.

Getting the data out of a locked-down workstation. Download may be blocked there. Copy all as JSON puts everything on the clipboard — paste it into an email to yourself and nothing is lost. If the clipboard is blocked too, the text appears in the box below ready to select. Your entries also survive a page reload in this browser, but not a different browser or a cleared profile, so copy before you walk away.

Reading the differences

Quantity	Agreement	What a gap points at
CrCl	should be exact once overridden	If it differs without an override, the covariate layer differs — weight basis, SCr floor, or rounding. Fix this before reading anything else.
Posterior CL	<5% close · 5–15% worth a look · >15% investigate	Same model + same CrCl + same data should give nearly the same CL. A gap means the objective, the prior variances, or the optimiser differ.
Volume	looser — V is weakly identified	With one level CL and V are not separately identifiable, so V can differ a lot while AUC agrees. Do not over-read it.
AUC₂₄	<10% good · 10–20% check horizon · >20% investigate	The number that drives dosing. Check the horizon control first — a steady-state vs 2-day gap lands right in the 5–10% range.
Trough	looser still	Very sensitive to volume and to 1-comp vs 2-comp. A trough gap with matching AUC is expected between compartment models, not a defect.
Recommended regimen	may legitimately differ	At steady state AUC₂₄ is daily dose ÷ clearance and carries no interval term, so several intervals reach the same exposure. Two tools can pick differently and both be right. Compare the daily dose first; only then the interval.

Advisory tooling — not a prescription. Kept out of version-controlled patient data by design; everything typed here lives only in this browser. Pair with docs/audit/probe-v3-case-34m.cjs to reproduce any case against the shipped engine.