Therapeutic drug monitoring (TDM) is the clinical practice of measuring drug concentrations in biological fluids, typically plasma or serum, and interpreting the results to individualize drug therapy. TDM aims to maintain drug concentrations within a defined therapeutic window, where efficacy is maximized and toxicity is minimized. It is most useful for drugs with narrow therapeutic indices, substantial interpatient pharmacokinetic variability, and a well-defined relationship between concentration and clinical effect.
Indications for TDM
TDM is indicated for drugs where the therapeutic index is narrow, meaning the difference between the minimum effective concentration and the minimum toxic concentration is small. Aminoglycoside antibiotics such as gentamicin and tobramycin require TDM because nephrotoxicity and ototoxicity are concentration-dependent. Vancomycin monitoring ensures trough concentrations are sufficient for efficacy while avoiding accumulation. Digoxin monitoring balances positive inotropic effect against the risk of serious arrhythmias.
Other drugs commonly monitored include lithium for bipolar disorder, phenytoin for seizure control, theophylline for asthma and chronic obstructive pulmonary disease, cyclosporine and tacrolimus for immunosuppression after organ transplantation, and methotrexate in high-dose cancer chemotherapy protocols. TDM is also valuable for drugs with unpredictable pharmacokinetics due to variable absorption, metabolism, or clearance, and for patients with organ dysfunction that alters drug elimination.
Sampling Times
The interpretation of TDM results depends critically on the timing of sample collection relative to drug administration. Trough concentrations are measured immediately before the next dose and reflect the lowest concentration during the dosing interval. Trough concentrations are used for drugs where the minimum effective concentration must be maintained throughout the interval, such as vancomycin for bactericidal activity. Peak concentrations are measured approximately one hour after intravenous administration or at the expected time of maximum absorption for oral drugs. Peaks are used for drugs with concentration-dependent efficacy, such as aminoglycosides, where achieving a high peak relative to the minimum inhibitory concentration correlates with clinical response.
Random sampling is appropriate for drugs with very long half-lives where the difference between peak and trough is negligible, such as lithium or phenytoin, provided the sample is drawn at steady state. The timing of the sample must be recorded precisely and interpreted in the context of the dosing history.
Target Concentration Ranges
Each drug monitored by TDM has an established therapeutic range, defined as the range of concentrations associated with maximal efficacy and minimal toxicity. These ranges are population-derived estimates and may require adjustment for individual patients based on clinical response. For vancomycin, the recommended trough concentration is typically 15 to 20 mg per L for serious infections, with higher targets for complicated infections. For lithium, the therapeutic range is 0.6 to 1.2 mmol per L for maintenance therapy and 0.8 to 1.5 mmol per L for acute mania. For digoxin, the therapeutic range is 0.5 to 2.0 ng per mL, with the lower end increasingly preferred for heart failure management.
The therapeutic range is not absolute, and some patients may respond well at concentrations below the range while others experience toxicity within the range. Clinical assessment remains essential, and TDM results should be interpreted as one component of a comprehensive evaluation.
Bayesian Feedback Dosing
A sophisticated approach to TDM is Bayesian feedback dosing, which combines population pharmacokinetic data with individual concentration measurements to estimate the patient’s own pharmacokinetic parameters. The Bayesian approach begins with a population model that describes the typical pharmacokinetic parameters and their variability. When individual concentration measurements are obtained, the Bayesian algorithm updates the parameter estimates, weighting the population data and the individual data according to their respective uncertainties.
As more individual concentration measurements become available, the Bayesian estimates become increasingly specific to the patient. This approach allows precise prediction of future concentrations under alternative dosing regimens, enabling dose adjustments tailored to the individual patient. Bayesian feedback dosing is implemented in clinical software platforms for vancomycin, aminoglycosides, and other drugs.
Examples in Clinical Practice
Vancomycin TDM illustrates the value of concentration-guided dosing. After initiation of vancomycin therapy, a trough concentration is measured at steady state, typically before the fourth dose. If the trough is below target, the dose or frequency is increased. If the trough is above target, the dose is reduced or the interval extended. In patients with unstable renal function, more frequent monitoring is required, and the dosing regimen may change substantially over the course of therapy.
Lithium TDM demonstrates the importance of adherence assessment. Poor adherence is common in psychiatric disorders, and a low lithium concentration may indicate missed doses rather than inadequate dosing. Conversely, a high concentration may result from toxicity, drug interactions, or dehydration. Interpretation requires integrating the concentration with the clinical presentation, the patient’s reported adherence, and concurrent medication changes.
The goal of TDM is not merely to generate a number but to use that number, in conjunction with clinical assessment, to make informed therapeutic decisions that improve patient outcomes while minimizing the risk of adverse effects.
