Therapeutic Drug Monitoring
On this page
Direct answer
A witnessed seizure in a compliant epileptic on "correct" phenytoin doses is the canonical TDM problem: phenytoin saturates its metabolising enzymes (Michaelis-Menten kinetics), so a small dose increase near saturation can double levels — toxicity without prescription error. Therapeutic drug monitoring fits drugs combining a narrow therapeutic index with concentration-linked efficacy and toxicity, variable pharmacokinetics and no easily observed endpoint. Core targets: phenytoin 10-20, carbamazepine 4-12, valproate 50-100 micrograms/mL; lithium 0.6-1.2 mEq/L (12-hour trough); digoxin 0.5-0.9 ng/mL; theophylline 10-20; vancomycin trough 10-20; amikacin peak 20-30. Sampling discipline is half the science: troughs just before the next dose at steady state, peaks for aminoglycosides, free phenytoin when albumin is low.
What you must remember
- Indications: narrow therapeutic index, unpredictable PK (nonlinear metabolism, interactions, organ dysfunction), concentration-related toxicity, and target-organ damage that only shows late.
- Antiepileptic ranges: phenytoin 10-20, carbamazepine 4-12, valproate 50-100, phenobarbitone 15-40 micrograms/mL — phenytoin alone is nonlinear at therapeutic doses.
- Lithium discipline: 12-hour post-dose trough, 0.6-1.2 mEq/L (higher band for acute mania); excreted unchanged by the kidney, so NSAIDs, ACE inhibitors, thiazides and dehydration all raise it — the interaction list is the viva.
- Digoxin 0.5-0.9 ng/mL: sample at least 6-8 hours post-dose; levels interpret symptoms (nausea, xanthopsia, arrhythmia) with potassium and renal function, never alone.
- Free phenytoin logic: only unbound drug crosses to brain and receptor; in hypoalbuminaemia, pregnancy or uraemia, total levels understate toxicity — request free level (target 1-2 micrograms/mL) or correct with the Sheiner-Tozer equation.
- Aminoglycoside strategy: traditional peak (gentamicin 5-10, amikacin 20-30) and trough (below 2 and below 10 respectively); extended-interval dosing monitors a single trough that should be undetectable before re-dosing.
- Vancomycin evolution: trough 10-20 mg/L historically, with nephrotoxicity risk above 20; current guidance moves toward Bayesian AUC-guided dosing (AUC/MIC 400-600) — quote both to look current.
- Timing rules: trough = just before next dose; steady state = about 5 half-lives; post-distribution digoxin and lithium = 6-12 hours; never sample right after a dose.
Interpreting a level, stepwise
An elderly woman on phenytoin 300 mg daily for years is added cimetidine (or becomes hypoalbuminaemic after a pneumonia) and arrives with nystagmus and ataxia. Step one, suspect: the interaction or protein-binding change alters free drug even if total level reads "therapeutic". Step two, sample correctly: trough, at steady state, noting the dose and any missed doses. Step three, interpret against the clinical state — a level of 18 with ataxia in a uraemic, hypoalbuminaemic patient means free phenytoin near toxic. Step four, act pharmacokinetically: hold doses, recheck in days (phenytoin's half-life lengthens as levels rise, so recovery is slower than the toxic climb — the definition of nonlinear kinetics), and adjust maintenance using the Michaelis-Menten relationship rather than proportional arithmetic. The same four steps run for lithium tremor confusion after a thiazide, or a digoxin level of 2.4 with hypokalaemia: the number is a coordinate, not a verdict.
Where students slip
Treating the range as the diagnosis is the cardinal error: a digoxin level of 2 is therapeutic in atrial fibrillation control for one patient and toxic in another with hypothyroidism and hypokalaemia — symptoms, electrolytes and ECG interpret the number. The second slip is sampling at the wrong time — a "trough" drawn an hour after the morning dose is meaningless, a common ward reality in India where phlebotomy timing is the intern's responsibility. Third, forgetting that aminoglycoside toxicity (nephro- and ototoxicity) correlates with troughs, not peaks — the once-daily dosing logic is precisely trough minimisation.
Frequently asked questions
What makes phenytoin the classic drug for therapeutic drug monitoring?
It exhibits dose-dependent Michaelis-Menten kinetics — metabolism saturates within the therapeutic range — so level rises are disproportional to dose increases, and its toxic-therapeutic margin is narrow.
When and how is a trough sample drawn?
Just before the next scheduled dose, once steady state (about five half-lives) has been reached — the point of minimal concentration that best reflects accumulation.
Why monitor free rather than total phenytoin in hypoalbuminaemia?
Phenytoin is 90 percent albumin-bound; with less albumin (or uraemic displacement), total levels fall while free, pharmacologically active drug stays the same or rises — target free 1-2 micrograms/mL.
Which drugs and states raise lithium levels dangerously?
Renal impairment, dehydration, NSAIDs, ACE inhibitors/ARBs, and thiazides — all reduce lithium clearance, since lithium is excreted entirely unchanged by the kidney.
What is the current monitoring target for vancomycin?
Traditionally a trough of 10-20 mg/L, increasingly replaced by AUC-guided dosing targeting AUC/MIC of 400-600, which achieves efficacy with less nephrotoxicity than fixed high troughs.