An 80-year-old woman weighing 50 kg has a serum creatinine of 1.0 mg/dL, a value most laboratories would not flag. Calculated by Cockcroft-Gault, her creatinine clearance is closer to 35 mL/min. For drugs such as enoxaparin, dabigatran or gabapentin, that gap can decide whether a standard dose is appropriate or excessive.

Organ dysfunction is one of the most common reasons a correct dose becomes an incorrect one. The kidneys and liver together eliminate most drugs. When either is impaired, drug exposure can rise well beyond the range studied in the trials that established the standard dose. Yet the two organs present very different clinical problems, and treating them as equivalent leads to predictable errors.

Renal impairment is comparatively tractable. Clinicians have imperfect but usable estimates of kidney function, and the clearance of many drugs tracks those estimates closely. Hepatic impairment offers no such anchor. Understanding why, and what can be done about it, clarifies one of the quieter but more consequential parts of prescribing.

The most useful predictor of renal dosing sensitivity is the fraction of a drug excreted unchanged in the urine, often written as fe. When fe is high, roughly above 50 percent, kidney function becomes the dominant determinant of total clearance. Aminoglycosides, vancomycin, lithium, dabigatran, gabapentin and many beta-lactams fall into this group. When fe is low, kidney function matters far less for the parent compound.

Renally dependent drugs share recognisable traits. They tend to be water-soluble, small enough to pass the glomerular barrier, and minimally protein-bound, because only unbound drug is filtered. Some are also actively secreted by tubular transporters such as OCT2 and the organic anion transporters. Competition at these sites explains why trimethoprim and cimetidine can raise serum creatinine without any real change in glomerular filtration.

When filtration falls, clearance falls roughly in proportion and elimination half-life lengthens. Because steady state takes about four to five half-lives to reach, accumulation is often slow and silent. A patient may tolerate the first several days of an unadjusted regimen and only later develop lithium toxicity, digoxin-related arrhythmia or aminoglycoside ototoxicity. By then, the link to dosing is easy to miss.

The parent drug is not the whole story. Some hepatically metabolised drugs produce active or toxic metabolites that depend on the kidney for removal. The glucuronide metabolites of morphine and normeperidine from pethidine accumulate in renal impairment, causing sedation, myoclonus or seizures. In these cases, fe for the parent drug underestimates risk, which is why some agents are avoided in advanced kidney disease rather than simply reduced.

Takeaway

Ask two questions of any drug in kidney disease: how much of it leaves unchanged in the urine, and what happens to the rest. The second question catches the problems the first one misses.

Renal dosing has a convenient anchor in creatinine. The liver has no equivalent. Bilirubin, albumin and INR reflect excretory and synthetic function, but none tracks the activity of the cytochrome P450 enzymes or conjugation pathways that actually clear drugs. The Child-Pugh score, widely used in drug labelling, was developed to predict outcomes in cirrhosis, not drug clearance. It classifies patients coarsely and explains only part of the pharmacokinetic variation.

Hepatic metabolism is also not a single process. In cirrhosis, phase I oxidative pathways tend to decline earlier and more steeply than phase II glucuronidation. Individual CYP isoenzymes are affected unevenly. This is the rationale for preferring lorazepam, oxazepam or temazepam, which undergo direct glucuronidation, over diazepam. Diazepam relies on oxidation and produces long-lived active metabolites that accumulate unpredictably.

Blood flow adds another layer. Drugs with a high hepatic extraction ratio, such as propranolol, morphine, verapamil and lidocaine, are cleared in proportion to hepatic blood flow. In cirrhosis, portosystemic shunting lets orally administered drug bypass the liver, so oral bioavailability can rise several-fold before intrinsic metabolic capacity changes much. Low-extraction drugs behave differently. Their clearance depends on enzyme capacity and on the unbound fraction, which rises as albumin falls.

The patient's response changes too, not only the concentration. People with cirrhosis show heightened sensitivity to sedatives and opioids, partly because they are vulnerable to hepatic encephalopathy. Their renal function is also frequently overestimated. Reduced muscle mass and impaired hepatic creatine synthesis lower serum creatinine, which can mask kidney impairment that compounds the hepatic picture. Liver disease therefore often becomes a two-organ dosing problem without announcing itself as one.

Takeaway

With the kidney, you can usually calculate; with the liver, you mostly infer. Hepatic dosing recommendations are best treated as a starting hypothesis, tested against the patient's response.

Adjustment begins with a reasonable estimate of function. Many drug labels derive their renal thresholds from studies that used Cockcroft-Gault creatinine clearance, while laboratories report eGFR normalised to 1.73 m² of body surface area. At extremes of age and body size the two can diverge meaningfully, and for narrow-therapeutic-index drugs the choice matters. In acute kidney injury, any creatinine-based estimate lags behind true function, because creatinine is not at steady state.

The loading dose depends mainly on volume of distribution, which organ dysfunction usually alters less than clearance. In most cases the loading dose is therefore unchanged, and reducing it only delays therapeutic concentrations. It is the maintenance regimen that must account for reduced clearance. Early under-dosing is a recognised hazard in sepsis. Beta-lactams are often given at full dose for the first 24 to 48 hours while renal function declares itself.

There are two levers: smaller doses at the usual interval, or the usual dose given less often. Both can achieve the same average concentration but produce different profiles. Extending the interval preserves high peaks with deeper troughs. This suits concentration-dependent agents like aminoglycosides, where peak concentration drives efficacy and low troughs limit toxicity. Reducing the dose keeps levels steadier, which suits beta-lactams, whose efficacy depends on time above the MIC.

Where possible, measurement replaces estimation. Therapeutic drug monitoring for vancomycin, aminoglycosides, lithium and digoxin allows individualised dosing whatever method estimated function. In haemodialysis, dialysable drugs are timed after sessions or supplemented. For hepatic impairment the tools are blunter. Clinicians select drugs cleared by preserved pathways, start low and titrate slowly, and reduce oral doses of high-extraction drugs more than intravenous ones.

Takeaway

Every adjustment trades the risk of accumulation against the risk of under-treatment. The right regimen matches how the drug works, not merely how it is cleared.

The underlying principle is shared. When an eliminating organ fails, clearance falls, half-life lengthens and exposure rises unless the regimen changes. What differs is how confidently that change can be predicted.

For renally cleared drugs, the fraction excreted unchanged and a well-chosen estimate of clearance support reasonably precise adjustment, refined by drug monitoring where available. For hepatic impairment, prediction gives way to judgement: choosing drugs with preserved pathways, accounting for shunting and protein binding, and titrating against clinical response.

In both cases the evidence points the same way. Dose adjustment is not simple subtraction. It requires knowing how the drug is eliminated, how it exerts its effect, and which error carries the greater cost for this patient.