Three pharmacokinetic parameters (clearance, volume of distribution, and half-life) govern how every drug behaves in the body. They are not independent: half-life is a derived parameter determined jointly by clearance and volume of distribution. Understanding this interdependence prevents the common clinical error of treating half-life as a fixed, intrinsic property of the drug divorced from the patient's physiology.
The elimination half-life (t1/2) is the time required for the plasma drug concentration to fall by exactly 50% during the elimination phase. For drugs obeying first-order kinetics, a constant fraction of the drug present is eliminated per unit time, and t1/2 is constant regardless of the initial concentration. The relationship between half-life, clearance (CL), and volume of distribution (Vd) is: t1/2 = (0.693 × Vd) / CL. This equation reveals that half-life is not a direct measure of either clearance or distribution alone. A drug may have a long half-life because its clearance is low, because its Vd is large, or both. Amiodarone and chloroquine have very long half-lives (weeks to months) primarily because of their enormous volumes of distribution (60 L/kg and 200–800 L/kg respectively), not because of impaired clearance. Conversely, lithium has a relatively modest Vd but a long half-life because its renal clearance is low. Interpreting a long half-life correctly requires knowing which parameter is responsible.1
First-Order Elimination Kinetics. Under first-order kinetics, the rate of drug elimination is proportional to the drug concentration: rate = ke × C, where ke is the first-order elimination rate constant related to half-life by ke = 0.693 / t1/2. First-order kinetics produce an exponential decline in plasma concentration over time, which appears linear on a semi-logarithmic plot. The practical clinical consequence of first-order kinetics is that it takes exactly five half-lives to eliminate approximately 97% of a drug from the body, regardless of the initial dose or concentration. This property determines washout times after discontinuation (five half-lives to functional elimination), the time required to reach steady state during repeated dosing (also five half-lives), and the duration of action after a single dose for drugs whose effects are concentration-dependent throughout the clinically relevant range.12
Total Body Clearance and Its Components. Total body clearance (CL) is the sum of all individual organ clearances: CL = CLrenal + CLhepatic + CLother. Clearance has units of volume per time (mL/min or L/hr) and represents the volume of plasma completely cleared of drug per unit time. It is determined by organ function and intrinsic metabolic capacity, not by drug concentration (in linear kinetics). When renal clearance is reduced by kidney disease, total body clearance falls by the fraction normally contributed by the kidney, increasing the area under the plasma concentration-time curve (AUC) and prolonging drug accumulation. When hepatic clearance is reduced by liver disease or enzyme inhibition, total clearance falls by the fraction contributed by hepatic metabolism, with effects on AUC and t1/2 governed by the hepatic extraction ratio model described in the preceding metabolism module. The interdependence of CL and Vd means that disease states which alter Vd (fluid overload, hypoalbuminemia) change t1/2 independently of any change in CL, a point frequently overlooked in clinical practice.12
t1/2 = (0.693 × Vd) / CL. Long t1/2 may reflect large Vd (amiodarone, chloroquine), low CL (lithium, renally impaired drugs), or both. Five half-lives to: (a) eliminate 97% after stopping, (b) reach steady state during repeated dosing. Doubling the dose does not change t1/2 or time to steady state under linear kinetics — it doubles the steady-state concentration. Halving the dose interval without changing total daily dose reduces peak-to-trough fluctuation without changing steady-state AUC.
Steady state is the condition in which the rate of drug input equals the rate of drug elimination over each dosing interval, producing reproducible plasma concentration-time profiles from cycle to cycle. Understanding the determinants of steady-state concentration and the time required to reach it is essential for rational dosing of all drugs administered chronically.
When a drug is administered at a constant rate (continuous IV infusion), plasma concentration rises exponentially toward a plateau determined by the infusion rate divided by total body clearance (CL): Css = Infusion Rate / CL. The time to reach 50% of steady state is one half-life, 75% of steady state is two half-lives, 87.5% is three half-lives, and 97% is five half-lives. This time course is determined solely by the half-life and is entirely independent of the infusion rate or dose. Of note, doubling the infusion rate doubles the ultimate steady-state concentration but does not change the time required to reach steady state. This principle has a direct clinical implication: for drugs with long half-lives (digoxin t1/2 approximately 36 hours, amiodarone t1/2 weeks), waiting for spontaneous accumulation to produce therapeutic concentrations would require days to weeks. A loading dose bypasses this accumulation phase by immediately placing sufficient drug in the body to achieve the target Vd-based concentration, after which maintenance doses sustain steady state.12
Steady State with Intermittent Dosing. When a drug is administered as repeated discrete doses at a fixed interval (τ), steady state is characterized by reproducible maximum (peak, Cmax,ss) and minimum (trough, Cmin,ss) concentrations within each dosing interval. The average steady-state concentration is Cavg,ss = F × Dose / (CL × τ), and the total area under the plasma concentration-time curve (AUC) per dosing interval at steady state equals the AUC after a single dose (for linear kinetics). The degree of peak-to-trough fluctuation is governed by the ratio of the dosing interval to the half-life: when τ equals one half-life, peak is approximately twice the trough; when τ is much shorter than t1/2 (as in continuous infusion or very frequent dosing), concentrations are nearly constant with minimal fluctuation; when τ greatly exceeds t1/2, concentrations rise steeply after each dose and fall nearly to zero before the next, producing large fluctuations. The clinical goal is to keep steady-state peaks below the toxic threshold and troughs above the minimum effective concentration (MEC) throughout the dosing interval.23
Practical Steady-State Calculations. The maintenance dose required to sustain a target average steady-state concentration can be calculated as: Maintenance Dose = Ctarget × CL × τ / F. For IV dosing (F = 1), this simplifies to: Maintenance Dose = Ctarget × CL × τ. As a clinical example, a target digoxin Cavg,ss of 1.0 ng/mL (1.0 mcg/L) with a clearance of 100 mL/min (0.1 L/min), and an oral daily dosing interval with F approximately 0.7: Maintenance Dose = 1.0 × (100 mL/min × 1440 min/day) / 0.7 = approximately 206 mcg/day, consistent with the clinical maintenance range of 125 to 250 mcg/day. When disease states alter clearance (renal failure reducing digoxin CL) or bioavailability (malabsorption reducing F), the maintenance dose must be adjusted proportionally to maintain the target steady-state concentration.23
1. Time to steady state depends only on half-life (approximately five t1/2), not on dose or dosing frequency. 2. Steady-state concentration is proportional to dose-to-clearance ratio (F × Dose/CL per interval). 3. Doubling the dose doubles Css; halving the dose halves Css. 4. Changing the dosing interval at fixed total daily dose alters fluctuation but not average Css. 5. Loading doses are needed when t1/2 is long relative to the urgency of achieving therapeutic effect; they are calculated from Vd, not clearance.
Most drugs follow linear (first-order) pharmacokinetics at therapeutic concentrations, meaning that clearance and half-life are constant and plasma concentration is proportional to dose. A clinically important minority of drugs obey nonlinear kinetics, in which clearance decreases as drug concentration rises because a saturable biological process (typically an enzyme or transporter) becomes limiting. The clinical consequence is that small dose increments produce disproportionately large increases in plasma concentration, with unpredictable toxicity risk.
Michaelis-Menten (MM) kinetics describe the saturable relationship between drug concentration and elimination rate: rate of elimination = (Vmax × C) / (Km + C), where Vmax is the maximum elimination rate and Km is the Michaelis constant, the drug concentration at which the elimination rate is exactly half of Vmax. At low drug concentrations where C is much less than Km, the denominator approximates Km and the equation reduces to a first-order relationship: rate ≈ (Vmax / Km) × C. Under these conditions, the drug behaves as if it has constant clearance equal to Vmax/Km. At high concentrations where C greatly exceeds Km, the denominator approximates C and elimination proceeds at a constant rate (Vmax), independent of concentration. This is zero-order kinetics: a constant amount of drug is eliminated per unit time, not a constant fraction. Plasma concentration then declines linearly, not exponentially.34
Phenytoin: The Clinical Paradigm of Nonlinear Pharmacokinetics (PK). Phenytoin is the archetypal example of clinically important nonlinear pharmacokinetics. CYP2C9 (cytochrome P450 2C9)-mediated hydroxylation of phenytoin saturates within the therapeutic concentration range (10 to 20 mg/L), meaning that routine therapeutic doses push phenytoin into the region where Km and therapeutic concentrations are of the same order of magnitude, producing mixed first-order and zero-order elimination behavior. The clinical consequence is highly non-proportional dose-response: increasing phenytoin from 300 mg/day to 400 mg/day may increase the steady-state concentration from 15 mg/L to 25 mg/L (well above the toxic threshold), while decreasing from 300 to 200 mg/day may drop levels below the therapeutic range. This behavior is particularly dangerous because phenytoin has a narrow therapeutic index (NTI), its toxicity (nystagmus, ataxia, cognitive impairment) is dose-related and serious, and the degree of nonlinearity varies among patients depending on their individual Vmax and Km values, which in turn depend on CYP2C9 polymorphisms and body composition. Phenytoin dose adjustments must be made in small increments (25 to 50 mg/day), guided by repeated therapeutic drug monitoring (TDM) of free phenytoin concentrations, particularly in patients with hypoalbuminemia or renal failure where protein binding and thus the free fraction change.34
Other Drugs with Nonlinear Pharmacokinetics. Ethanol is eliminated by zero-order kinetics across most of its pharmacologically relevant concentration range because alcohol dehydrogenase (ADH) is saturated even at low blood ethanol concentrations. A standard drink raises blood alcohol concentration (BAC) by approximately 20 mg/dL, and elimination proceeds at approximately 15 to 20 mg/dL per hour regardless of BAC, making the duration of impairment predictable from initial BAC alone without reference to half-life. Aspirin exhibits dose-dependent nonlinear kinetics at high therapeutic or anti-inflammatory doses (above approximately 3 to 4 g/day) because hepatic glycine conjugation (the primary elimination pathway at high doses) saturates; at analgesic/antipyretic doses, aspirin follows approximate first-order kinetics. The antiviral drugs acyclovir and ganciclovir show saturable active tubular secretion at higher doses. High-dose methotrexate pharmacokinetics become nonlinear because renal tubular secretion approaches saturation, and the nonlinear elimination during high-dose methotrexate infusions is one reason why leucovorin rescue must be timed relative to measured methotrexate concentrations rather than simply elapsed time post-infusion.45
Drug behaves nonlinearly if: small dose increase produces disproportionately large concentration increase; half-life appears to lengthen at higher concentrations; plasma concentration declines linearly rather than exponentially after overdose. Affected drugs: phenytoin (most important; TDM mandatory for any dose change), ethanol (zero-order elimination always), high-dose aspirin, high-dose methotrexate. Rule: for phenytoin dose adjustments, never increase by more than 25–50 mg/day increments; re-check levels 2–3 weeks after each change before further adjustments.
Therapeutic drug monitoring (TDM) is the measurement of drug concentrations in biological fluids, used to individualize dosing when standard doses produce highly variable and clinically significant differences in drug exposure across patients. TDM is most valuable when the drug has a narrow therapeutic index, when the relationship between concentration and effect is better defined than the relationship between dose and effect, and when validated target concentration ranges have been established.
Criteria for TDM Utility. Not all drugs benefit from TDM. The criteria for TDM to be clinically useful are: the drug must have a narrow therapeutic index such that the difference between effective and toxic concentrations is small; plasma concentration must correlate better with clinical effect and toxicity than dose alone; analytical methods must be available, reproducible, and clinically accessible; and target concentration ranges must have been validated by clinical outcome data. Drugs meeting all these criteria include aminoglycosides, vancomycin, phenytoin, carbamazepine, valproic acid, lithium, tacrolimus, cyclosporine, digoxin, and methotrexate (at high doses). For drugs with wide therapeutic indices (amoxicillin, most statins, most antihistamines), TDM adds no clinical value because the therapeutic margin is so wide that concentration-response relationships are not clinically limiting within usual dosing ranges. For drugs where the pharmacodynamic endpoint is directly measurable (international normalized ratio (INR) for warfarin, blood pressure for antihypertensives), the measured effect is a superior guide to dosing than plasma drug concentration.5
TDM Sampling Strategy: When to Draw Levels. The timing of blood sampling relative to the dose is the most common source of TDM interpretation errors in clinical practice. Levels drawn at the wrong time relative to the dose are meaningless or actively misleading. The fundamental requirement is that samples be drawn at steady state (after approximately five half-lives of the maintenance regimen) and at a defined time point in the dosing cycle. Trough levels (drawn immediately before the next dose) are the most commonly used TDM metric for most drugs because they represent the minimum concentration in the dosing cycle and are most reproducible. For aminoglycosides on once-daily dosing, both peak (1 hour post-infusion) and trough (immediately pre-dose) are measured: peaks confirm adequate bactericidal exposure and troughs confirm adequate drug-free periods to minimize nephrotoxicity. For vancomycin, the preferred metric has shifted from trough-only to area under the plasma concentration-time curve (AUC)-guided monitoring (target AUC/minimum inhibitory concentration (MIC) ratio of 400 to 600 mg·h/L), which better predicts both efficacy and nephrotoxicity than trough alone. Digoxin samples must be drawn at least 6 hours post-dose to allow complete distribution into cardiac tissue; earlier samples represent the distribution phase and will falsely appear toxic.5
Free Drug Monitoring and Protein Binding Corrections. Standard TDM assays measure total drug concentration (free plus protein-bound). For drugs where protein binding is variable or reduced, total concentration may not accurately reflect the free (pharmacologically active) drug concentration, and TDM results must be interpreted with this in mind. Phenytoin illustrates this problem: the standard therapeutic range of total phenytoin 10 to 20 mg/L assumes normal albumin (approximately 4.0 g/dL) and normal protein binding (approximately 90%). In hypoalbuminemia (albumin below 3.5 g/dL) or uremia (where organic anion accumulation reduces albumin binding), the free fraction rises substantially. The Winter-Tozer equation corrects total phenytoin for reduced albumin: adjusted phenytoin = measured total phenytoin / ((0.2 × albumin / 4.4) + 0.1), allowing interpretation of the measured total level in the context of actual free drug exposure. Alternatively, direct free phenytoin measurement (reference range 1 to 2 mg/L, approximately 10% of total) is preferred when protein binding abnormalities are suspected.5
Vancomycin: AUC/MIC 400–600 mg·h/L (preferred over trough-only); traditional trough 10–15 mg/L (15–20 for severe infections). Aminoglycosides (gentamicin/tobramycin): peak 5–10 mg/L (once-daily); trough <1 mg/L. Phenytoin: total 10–20 mg/L; free 1–2 mg/L; draw at steady state, any time in dosing interval (not time-sensitive if at steady state). Carbamazepine: 4–12 mg/L trough. Valproic acid: 50–100 mg/L trough. Lithium: 0.6–1.2 mmol/L (12 hr post-dose trough). Digoxin: 0.5–0.9 ng/mL (chronic HF); draw ≥6 hr post-dose. Tacrolimus: 5–15 ng/mL whole blood trough (varies by transplant type and time post-transplant). Cyclosporine: 100–400 ng/mL trough (transplant-specific).
Standard pharmacokinetic parameters derived from studies in healthy young adults do not apply uniformly to patients at the extremes of age, during pregnancy, or with extreme body weight. Each of these populations exhibits predictable, physiology-driven deviations in drug absorption, distribution, metabolism, and elimination that require systematic anticipation rather than simple dose extrapolation.
Pediatric Pharmacokinetics. Children are not small adults. Drug disposition in pediatric patients differs from adults across all four pharmacokinetic processes, with the magnitude and direction of differences varying markedly by age. Neonates (0 to 28 days) have markedly reduced renal function, with glomerular filtration rate (GFR) approximately 40% of adult values at term birth and even lower in preterm infants, requiring substantial dose reduction for renally eliminated drugs. Neonatal hepatic cytochrome P450 (CYP) enzyme activity is also substantially reduced at birth: the cytochrome P450 3A4 (CYP3A4) isoform activity rises from approximately 30% of adult at birth to near-adult levels by 6 to 12 months; the cytochrome P450 1A2 (CYP1A2) isoform is nearly absent at birth and matures slowly over the first year; phase II glucuronosyltransferase (UGT) conjugation enzymes are similarly immature, a limitation that contributes to the gray baby syndrome observed when chloramphenicol is administered to neonates (inability to glucuronidate chloramphenicol leads to toxic accumulation).7
Gastric acid secretion is reduced in the first month of life, affecting the absorption of acid-dependent drugs. Total body water as a fraction of body weight is higher in neonates (approximately 80%) than adults (approximately 60%), increasing the Vd of hydrophilic drugs on a per-kilogram basis, which generally requires higher weight-adjusted doses of hydrophilic drugs. Plasma albumin and alpha-1-acid glycoprotein (AAG) concentrations are lower in neonates, increasing the free fraction of protein-bound drugs.7
Geriatric Pharmacokinetics. Aging produces a constellation of physiological changes that predictably alter drug disposition. Renal function declines with age independent of disease: GFR falls by approximately 1 mL/min/1.73 m2 per year after age 40, reaching approximately 50 to 70% of young adult values by age 70 in the absence of overt renal disease. The serum creatinine may remain within the normal range despite substantially reduced GFR in elderly individuals because of simultaneously reduced muscle mass and creatinine production, making Cockcroft-Gault weight and age correction essential. Hepatic blood flow and liver mass both decrease with age (by approximately 30 to 40%), reducing the clearance of high-extraction drugs. CYP enzyme activity declines modestly with age, primarily affecting phase I oxidative metabolism and reducing the clearance of low-extraction drugs metabolized by CYP3A4 and CYP1A2.8
Body composition changes substantially with age: lean body mass decreases and adipose tissue increases as a percentage of body weight, increasing the Vd of lipophilic drugs and prolonging their half-lives. Plasma albumin concentrations are generally preserved in healthy elderly individuals but fall in malnourished or acutely ill elderly patients. The net pharmacokinetic consequence of aging is broadly reduced drug clearance and prolonged half-lives for most drugs, supporting the general principle of starting at lower doses and titrating more slowly in elderly patients.68
Pharmacokinetics in Pregnancy. Pregnancy produces progressive physiological changes throughout gestation that alter every pharmacokinetic parameter. Gastric motility is reduced and gastric emptying delayed, altering absorption kinetics. Plasma volume expands by approximately 40 to 50% and total body water by 8 liters, substantially increasing the Vd of hydrophilic drugs. Plasma albumin concentration falls by approximately 20% due to dilutional effects, increasing the free fraction of highly albumin-bound drugs. Renal blood flow increases by approximately 50% and GFR rises by approximately 50% above pre-pregnancy values by the third trimester, substantially increasing renal clearance of renally eliminated drugs such as lamotrigine, lithium, and beta-lactam antibiotics; doses of these drugs often require upward adjustment during pregnancy to maintain therapeutic concentrations.8
Hepatic CYP3A4 and CYP2D6 (cytochrome P450 2D6) isoform activity increases during pregnancy (estrogen and progesterone induction), increasing the metabolism of their substrates; CYP1A2 activity is reduced during pregnancy, potentially increasing exposure to drugs such as theophylline and caffeine. These changes create a dynamic pharmacokinetic environment that may require dose adjustments at different trimesters, and therapeutic drug monitoring of narrow-index drugs is particularly important during pregnancy and postpartum as physiology normalizes.8
Neonates: reduced CYP enzymes (especially CYP1A2, CYP2C19), reduced UGT (avoid chloramphenicol), reduced GFR (40% of adult), large body water (increased hydrophilic drug Vd). Geriatrics: GFR falls ~1 mL/min/year after 40; use Cockcroft-Gault with actual weight; reduced hepatic blood flow and CYP activity; increased fat mass prolongs lipophilic drug t1/2; start low, go slow. Pregnancy: increased Vd (hemodilution), increased GFR by 50% in T3 (dose-adjust lamotrigine, lithium, beta-lactams upward), increased CYP3A4/2D6, decreased CYP1A2; TDM essential for NTI drugs throughout gestation and postpartum.
Pharmacokinetics describes what the body does to the drug; pharmacodynamics describes what the drug does to the body. The pharmacokinetic-pharmacodynamic (PK-PD) relationship links these two domains and provides the scientific rationale for dosing regimens, dosing intervals, and target concentration strategies that maximize efficacy while minimizing toxicity. For anti-infective agents in particular, PK-PD modeling has transformed empirical dose selection into a quantitative, mechanism-driven science.
Concentration-Effect Relationships. The relationship between drug concentration at the effect site (Ce) and the pharmacological response is typically described by the Hill equation (Emax model): Effect = (Emax × Cen) / (EC50n + Cen), where Emax is the maximum achievable effect, EC50 is the concentration producing 50% of Emax, and n (the Hill coefficient) determines the steepness of the concentration-response curve. For drugs where n equals 1, the sigmoid curve is relatively shallow (a 10-fold increase in concentration from EC50 produces a large but submaximal increase in effect). For drugs where n is large, the concentration-response curve is steeper, meaning that small concentration changes near EC50 produce large changes in effect, and concentrations either far below or far above EC50 produce near-maximal or near-zero effects respectively. The shape and steepness of the concentration-response relationship, together with the therapeutic index (the ratio of EC50 for toxicity to EC50 for efficacy), determines how precisely drug exposure must be controlled to achieve efficacy without toxicity.8
Time-Dependent vs. Concentration-Dependent Killing. The most clinically consequential application of PK-PD principles is in the optimization of antibiotic dosing. Antibacterial drugs are classified by their primary PK-PD driver of efficacy. Time-dependent (time-above-MIC) antibiotics achieve maximal killing when drug concentrations remain above the minimum inhibitory concentration (MIC) for a sufficient proportion of the dosing interval; the pharmacodynamic target is the time above MIC (T>MIC) as a fraction of the dosing interval (τ). Beta-lactam antibiotics (penicillins, cephalosporins, carbapenems) are the canonical time-dependent agents: bactericidal killing is maximized when T>MIC exceeds 40 to 70% of the dosing interval depending on the agent. Increasing the beta-lactam dose above that required to achieve T>MIC adds no additional bactericidal benefit; the preferred strategies for maximizing T>MIC are extending the infusion duration (extended or continuous infusion) or shortening the dosing interval, not simply increasing the dose.9
Concentration-dependent (AUC/MIC (area under the curve/minimum inhibitory concentration) or Cmax/MIC driven) antibiotics achieve maximal killing at high peak concentrations relative to MIC; increasing the dose produces greater bactericidal activity even when T>MIC is already maximized. Aminoglycosides and fluoroquinolones are the principal concentration-dependent agents: the target PK-PD indices are AUC/MIC (for fluoroquinolones, target above 125 for gram-negative organisms) and Cmax/MIC (for aminoglycosides, target 8 to 10). This pharmacodynamic rationale justifies once-daily aminoglycoside dosing, which delivers a high peak (high Cmax/MIC) followed by a drug-free trough period that reduces adaptive resistance and nephrotoxicity.910
Effect-Site Concentration and Hysteresis. For many drugs, there is a delay between peak plasma concentration and peak pharmacological effect because the drug must distribute from plasma to its effect site. This delay is modeled by an effect compartment (linked to the central compartment by an equilibration rate constant ke0), and when plasma concentration is plotted against effect over time, a counterclockwise hysteresis loop is observed: for any given drug effect level, the concentration during the ascending phase of the plasma level curve is higher than during the descending phase, because the effect site has not yet fully equilibrated with plasma. Clinically, this explains why the pharmacological peak effect of many IV anesthetics, opioids, and antiepileptics occurs some minutes after the plasma concentration peak. The half-life of effect-site equilibration (t1/2,ke0) determines how rapidly the drug concentration at the effect site tracks plasma concentration. For propofol and ketamine, t1/2,ke0 is 1 to 3 minutes, producing rapid clinical onset and offset; for morphine, t1/2,ke0 is approximately 15 to 20 minutes, explaining the delayed respiratory depression sometimes observed after IV morphine boluses when plasma concentration has already begun to fall.8
Absorption (PK-01): bioavailability, first-pass extraction, formulation effects. Distribution (PK-02): Vd, protein binding, tissue barriers, compartment models. Metabolism (PK-03): CYP450, phase II, enzyme interactions, pharmacogenomics. Elimination (PK-04): renal filtration, tubular transport, biliary excretion, enterohepatic recirculation. PK Principles (PK-05): t1/2–CL–Vd triangle, steady state, nonlinear kinetics (phenytoin), TDM principles and targets, special population dose adjustments, PK-PD drivers for antibiotic optimization. The PK-PD target: keep concentrations above the MEC for efficacy while below the toxic threshold throughout the dosing interval at steady state.
Rowland M, Tozer TN. Clinical Pharmacokinetics and Pharmacodynamics: Concepts and Applications. 4th ed. Philadelphia: Lippincott Williams & Wilkins; 2011. ISBN 9780781750462.
Shargel L, Wu-Pong S, Yu ABC. Applied Biopharmaceutics and Pharmacokinetics. 7th ed. New York: McGraw-Hill; 2016. ISBN 9780071830935.
Winter ME. Basic Clinical Pharmacokinetics. 5th ed. Philadelphia: Lippincott Williams & Wilkins; 2009. ISBN 9780781750868.
Ludden TM. Nonlinear pharmacokinetics: clinical implications. Clin Pharmacokinet. 1991;20(6):429-446.
doi:10.2165/00003088-199120060-00001Burton ME, Shaw LM, Schentag JJ, Evans WE, eds. Applied Pharmacokinetics and Pharmacodynamics: Principles of Therapeutic Drug Monitoring. 4th ed. Philadelphia: Lippincott Williams & Wilkins; 2006. ISBN 9780781748483.
Pea F, Viale P, Furlanut M. Antimicrobial therapy in severely ill patients: a review of pathophysiological conditions responsible for altered disposition and pharmacokinetic variability. Clin Pharmacokinet. 2005;44(10):1009-1034.
doi:10.2165/00003088-200544100-00002Kearns GL, Abdel-Rahman SM, Alander SW, Blowey DL, Leeder JS, Kauffman RE. Developmental pharmacology — drug disposition, action, and therapy in infants and children. N Engl J Med. 2003;349(12):1157-1167.
doi:10.1056/NEJMra030126Brunton LL, Hilal-Dandan R, Knollmann BC, eds. Goodman & Gilman's: The Pharmacological Basis of Therapeutics. 13th ed. New York: McGraw-Hill; 2018. ISBN 9781259584732.
Craig WA. Pharmacokinetic/pharmacodynamic parameters: rationale for antibacterial dosing of mice and men. Clin Infect Dis. 1998;26(1):1-10.
doi:10.1086/516284Drusano GL. Antimicrobial pharmacodynamics: key interactions of ‘bug and drug’. Nat Rev Microbiol. 2004;2(4):289-300.
doi:10.1038/nrmicro862