Hepatic Impairment · Chapter 1
The Hepatic Decision Framework
How to think about psychotropic prescribing when the liver is the constraint — which score actually drives the dose, why most psychotropics are in play, and the two traps the kidney never had: drug-induced liver injury and encephalopathy.
Bottom Line Up Front
The 30-second version
- The default flips from the kidney. In renal disease, most psychotropics are hepatically cleared and escape adjustment. In hepatic disease, most psychotropics are the ones being metabolized — so the liver is the comorbidity that touches the largest share of your formulary. The posture changes from "scrutinize a short list" to "assume most agents are affected, and prove otherwise."
- Child-Pugh class — not MELD, not the LFTs — is the dosing tool. FDA hepatic dose recommendations are written against Child-Pugh A/B/C. MELD predicts transplant priority and mortality; transaminases report hepatocyte injury, not metabolic capacity.
- You are managing two separate risks, not one. Pharmacokinetic accumulation (dose it down) is one axis. Drug-induced liver injury — valproate, duloxetine, nefazodone — is a different axis entirely: an avoid-or-don't decision that no dose reduction fixes.
- The cirrhotic brain is more fragile than the cirrhotic liver is slow. Even when the pharmacokinetics barely move, sedatives, opioids, and CNS depressants can precipitate or mask hepatic encephalopathy. Pharmacodynamic caution often outranks the pharmacokinetic number.
Which Score to Trust (60-Second Version)
Three liver numbers sit on the chart, and only one of them is a dosing tool. The LFTs (AST, ALT, bilirubin, alkaline phosphatase) describe injury and cholestasis — they say nothing reliable about metabolic capacity, and a “burned-out” cirrhotic can have near-normal transaminases on top of almost no reserve. MELD (now MELD 3.0, which added sex and albumin and was adopted for US liver allocation in July 2023) is a transplant-allocation and 90-day-mortality score; it is in the chart because the patient is being risk-stratified for transplant, and no drug label uses it. Child-Pugh is the one built into prescribing: it bundles synthetic function (albumin, INR), excretory function (bilirubin), and clinical decompensation (ascites, encephalopathy) into a rough but usable proxy for metabolic reserve, and the FDA writes hepatic-impairment dose language against its A/B/C classes.
So the practical rule, the hepatic analogue of “match the number to the label”: dose off the Child-Pugh class; use MELD for prognosis and transplant referral; read the LFTs for the injury story, not the dose. The most common error at the point of care is reaching for whichever liver number is most abnormal — usually a transaminase — and titrating to it.
Why not the LFTs?
“Liver function tests” is a misnomer — AST/ALT measure hepatocyte leak, not function. They can be normal in advanced cirrhosis and wildly elevated in a viral hepatitis whose clearance is fine. The genuine markers of function are the synthetic ones (albumin, INR) and bilirubin — which is exactly what Child-Pugh captures, and why it, not the transaminases, drives the dose.
The Child-Pugh Score That Drives the Dose
Five components, each scored 1–3, summed to 5–15. Three are objective labs; two are clinical and observer-dependent (a known weakness of the score — grade ascites and encephalopathy after current treatment).
| Component | 1 point | 2 points | 3 points |
|---|---|---|---|
| Total bilirubin (mg/dL) | < 2 | 2–3 | > 3 |
| Serum albumin (g/dL) | > 3.5 | 2.8–3.5 | < 2.8 |
| INR | < 1.7 | 1.7–2.3 | > 2.3 |
| Ascites | None | Mild / controlled | Moderate–severe / refractory |
| Encephalopathy | None | Grade I–II | Grade III–IV |
The total maps to three classes that the FDA and drug labels use directly:
| Class | Score | What it means for dosing |
|---|---|---|
| A — mild | 5–6 | Compensated. Most agents tolerate standard or modestly reduced dosing — but still start low. |
| B — moderate | 7–9 | The pivot. Most hepatically-cleared psychotropics need real dose reduction; some agents move onto the avoid list. |
| C — severe | 10–15 | Decompensated. Many agents avoided; those used are substantially reduced, with specialist input. |
The single most useful break-point to memorize: Class B is where the dosing gets serious. Class A often behaves close to normal for many agents; Class C is specialist ground. Most FDA label language is written exactly in these three tiers — mild, moderate, severe.
The Core Principle: There Is No eGFR for the Liver
This is the idea to carry out of the chapter. The kidney hands you a continuous, drug-agnostic filtration estimate — and for the dangerous renal drugs, a measurable blood level. The liver gives you neither. Child-Pugh is categorical (A/B/C), it can't tell you which metabolic pathway your patient has lost, and most psychotropics have no therapeutic level to titrate against. You are dosing semi-blind: on class, on the drug's own pharmacology, and on clinical response.
So the point-of-care question is not “Is the liver impaired?” — it's “By what route is this drug cleared, how much first-pass does it normally lose, and how sedating is it?” Three things complicate the simple picture:
- Pathway matters, and the score can't see it. CYP oxidation (Phase I) falls earlier and harder than glucuronidation (Phase II) — the basis for preferring conjugated agents — but Child-Pugh won't tell you which pathway is hit. You have to know the drug. (And glucuronidation is only relatively spared; it erodes in Class C.)
- First-pass collapse raises oral exposure. Portosystemic shunting plus reduced enzyme mass let high-extraction drugs reach the circulation at a much higher fraction — an oral dose can behave like a bigger dose than you wrote.
- Hypoalbuminemia unmasks free drug. For highly protein-bound agents, a “therapeutic” total level sits on top of a higher free fraction — the same trap as uremia in renal disease, here driven by failed synthesis.
The Adjustment Logic
Renal dosing had a clean rule: adjust only if renal route and narrow margin and threshold crossed. The liver is messier, because most psychotropics are hepatically cleared — so the default tilts toward caution rather than away from it. Three questions size the pharmacokinetic adjustment:
| Question | What you're asking | Why it matters |
|---|---|---|
| 1. Hepatic route & extraction | Is the drug hepatically metabolized, and is it high-extraction (large first-pass)? | High-extraction agents swing most with shunting — the biggest jump in oral exposure. Almost all psychotropics are hepatically cleared; the renally-cleared ones are the exceptions that escape. |
| 2. Severity (class) | Is the patient Child-Pugh B or C? | Class A often tolerates near-normal dosing; the meaningful reductions start at B; C is specialist territory. |
| 3. CNS sensitivity / margin | Is the agent sedating or narrow-margin? | Even modest accumulation can precipitate encephalopathy or toxicity — in cirrhosis the brain is often the limiting organ, not the plasma level. |
The fourth question sits outside the dosing logic
- Is the drug hepatotoxic? That is not a “reduce the dose” decision — it's a “don't use it, or use it with eyes open and monitoring” decision. Valproate, duloxetine (its label says it should ordinarily not be used in hepatic insufficiency, and should not be prescribed with substantial alcohol use or chronic liver disease), and nefazodone (black-box hepatotoxicity, effectively off the market) cause injury that no dose reduction mitigates. Treat hepatotoxicity and pharmacokinetics as two separate decisions on the same patient. The full watchlist is its own member chapter.
Which Psychotropics Need Caution (Orientation)
This chapter gives you the map, not the agent-by-agent dose tables — those live in the member chapters. The hepatic landscape sorts differently from the renal one: the “safe” agents are the ones that bypass the liver, and the benzodiazepine you choose matters more than almost any dose number.
| Tier | Agents (orientation only) |
|---|---|
| Avoid / highest concern hepatotoxic or HE-precipitating |
Valproate — the defining hepatic psychotropic: idiosyncratic hepatotoxicity and its own hyperammonemic encephalopathy. Nefazodone (black-box liver failure; essentially withdrawn). Duloxetine (label: should ordinarily not be used in hepatic insufficiency, and not prescribed with substantial alcohol use or chronic liver disease). Long-acting benzodiazepines — diazepam, chlordiazepoxide — whose active metabolites accumulate and precipitate encephalopathy (the alcohol-withdrawal tension is handled in that chapter). |
| Reduce / watch hepatically cleared, dose-dependent |
Most antipsychotics (quetiapine, olanzapine, aripiprazole and others — reduce in B/C; sedation feeds HE risk), most antidepressants (SSRIs, SNRIs, TCAs — reduce, favor simpler metabolism), carbamazepine (hepatically metabolized, autoinducing, with its own hepatotoxicity signal — sits near the avoid line), and lamotrigine (glucuronidated; reduce in moderate–severe impairment). |
| Relatively preferred conjugated or renally cleared |
The conjugated benzodiazepines — lorazepam, oxazepam, temazepam (the “LOT” rule) — skip oxidation, have no active metabolites, and rely on the relatively-spared Phase II route. The renally-cleared agents that bypass the liver: lithium (the inversion — the renal villain is the hepatic safe-harbor, as long as the kidney is sound), gabapentin / pregabalin, and in part paliperidone. “Preferred” is not “ignore”: still start low, and remember conjugation slows in Class C. |
Two moves carry most of the benefit: when you need a benzodiazepine, reach for a conjugated one (LOT); and when the liver is the problem, lean on agents the liver never has to touch. Both are direct inversions of the renal playbook.
Compensated vs Decompensated: A Stable Liver vs a Moving Target
In compensated cirrhosis — Child-Pugh A, no ascites, encephalopathy, varices, or jaundice — metabolic reserve is reduced but stable, so the class is a usable dosing anchor: dose conservatively and monitor. Decompensated cirrhosis is different. Once ascites, overt encephalopathy, variceal bleeding, or jaundice appear, function can fall fast, and an acute decompensation or acute-on-chronic liver failure makes any single Child-Pugh snapshot a lagging estimate — the same problem AKI poses on the renal side, with the INR and encephalopathy components moving quickly.
And the traffic runs both ways. Psychotropics don't only accumulate in a failing liver — a sedating agent can precipitate decompensation, tipping a compensated cirrhotic into overt encephalopathy. A benzodiazepine or a sedating antipsychotic can be the event that converts a stable Child-Pugh A patient into a confused, ascitic admission. Dose to the trajectory, not the snapshot — and treat any new sedation in a cirrhotic as possible drug-precipitated encephalopathy until proven otherwise.
The Encephalopathy Trap (No Renal Analogue)
This is the hepatic mechanism with no equivalent in the kidney chapter, and it's why “just the dose” isn't enough. Psychotropics intersect hepatic encephalopathy two ways:
| Mode | What happens |
|---|---|
| Precipitates / deepens it sedatives, benzodiazepines, opioids, anticholinergics, sedating antipsychotics | The cirrhotic brain has reduced reserve; GABAergic and CNS-depressant drugs lower the threshold for overt encephalopathy and deepen it once present. Benzodiazepines are named precipitants in the AASLD/EASL guidance. |
| Masks it the diagnostic confound | Once you've sedated a cirrhotic, new confusion is ambiguous: is it encephalopathy (treat the liver — find the precipitant, lactulose, rifaximin) or oversedation (hold the drug)? They look identical at the bedside, and the management diverges completely. |
There is also a drug-specific variant with its own chapter: valproate-induced hyperammonemic encephalopathy, which can occur with normal LFTs and a therapeutic valproate level — so it's easy to miss and easy to misattribute to worsening liver disease. The practical posture across all of it: prefer the least-sedating effective agent, start low, and pre-commit to how you'll distinguish encephalopathy from oversedation. A serum ammonia is unreliable for diagnosing or grading encephalopathy; the clinical context, asterixis, and the response to encephalopathy therapy guide you. Treat a sedated cirrhotic as guilty until proven innocent.
The Action Ladder
Child-Pugh A (5–6), compensated
Most agents at standard or modestly reduced dose. Start low, and where two agents are otherwise equal, choose the less-sedating and less-hepatotoxic one. Avoid the hepatotoxic list regardless of class.
Child-Pugh B (7–9)
Reduce hepatically-cleared agents. Prefer conjugated benzodiazepines (LOT) and renally-cleared options; avoid the hepatotoxic list; watch actively for encephalopathy. This is where the dosing changes get real.
Child-Pugh C (10–15), decompensated
Specialist territory. Many agents avoided; those used are substantially reduced. Lean on renally-cleared agents, minimize sedatives, and involve hepatology/pharmacy. Even conjugation is slowed here.
Acute liver failure / acute-on-chronic
Function is a moving target — no single class anchors the dose. Dose to the trajectory, avoid hepatotoxic and sedating agents, and reassess frequently. Treat new confusion as encephalopathy until proven otherwise.
Clinical Pearls
Pearls
- Don't dose off the LFTs. Transaminases measure injury, not clearance; a burned-out cirrhotic can show near-normal enzymes on almost no reserve. Dose off the Child-Pugh class.
- Two axes, not one. Pharmacokinetic accumulation and drug-induced liver injury are separate decisions — reducing the dose does not make a hepatotoxin safe.
- When you need a benzo, reach for the conjugated ones. Lorazepam, oxazepam, temazepam (LOT) skip the oxidative step and carry no active metabolites — relatively spared in liver disease, though not in Class C.
- The renal villain is the hepatic safe-harbor. Lithium and the gabapentinoids are renally cleared, so they largely bypass the failing liver — watch the kidney instead. A clean inversion of the renal course.
- Sedation in a cirrhotic is encephalopathy until proven otherwise. New confusion on a sedating agent is indistinguishable from hepatic encephalopathy at the bedside, and the two treatments diverge.
- Low albumin unmasks free drug. For highly bound agents, read the level against the albumin — the same free-fraction trap as uremia, here driven by failed synthesis.
- Class B is the pivot. Class A often tolerates near-normal dosing; the meaningful reductions and the avoid-lists kick in at B, and C is specialist ground.
- Valproate can cause encephalopathy with normal LFTs and a normal level. Hyperammonemic encephalopathy is the trap that doesn't show on the liver panel — check an ammonia if a valproate patient turns confused. (Full treatment in the valproate chapters.)
Red Flags — Stop and Reassess
Hold the agent and look closer
- New or deepening confusion, drowsiness, or asterixis in a cirrhotic on any sedating psychotropic — assume drug-precipitated or unmasked hepatic encephalopathy; hold and reassess.
- Rising bilirubin or INR, or new ascites, after starting a hepatically-cleared agent — suspect decompensation and reassess the whole regimen.
- Any transaminase rise after starting a known hepatotoxin (valproate, duloxetine, carbamazepine) — a DILI signal; don't wait for jaundice.
- A long-acting benzodiazepine (diazepam, chlordiazepoxide) running in a cirrhotic outside a structured withdrawal protocol — active metabolites accumulate and precipitate encephalopathy.
- New confusion on valproate with normal LFTs — check an ammonia; hyperammonemic encephalopathy does not show on the liver panel.
Patient Counseling Script
Plain-language script
“This medication is processed by your liver, and because your liver isn't working at full capacity, we've lowered the dose and we'll go slowly. Two things to watch for. First, if you or your family notice you becoming more confused, very drowsy, hard to wake, or unsteady, call us or seek care — that can be a sign the medication or your liver needs attention, and we'd rather catch it early. Second, don't add any new medication, supplement, or alcohol without checking with us first, because they can stack up in a way they wouldn't in someone with a healthy liver. And please don't stop this on your own — let's adjust it together.”
EMR / Documentation Template
References
- U.S. Food & Drug Administration. Pharmacokinetics in Patients with Impaired Hepatic Function — Study Design, Data Analysis, and Impact on Dosing and Labeling: Guidance for Industry. May 2003. (Child-Pugh class as the basis for hepatic-impairment dose labeling.)
- Pugh RNH, Murray-Lyon IM, Dawson JL, Pietroni MC, Williams R. Transection of the oesophagus for bleeding oesophageal varices. Br J Surg. 1973;60(8):646–649. (Pugh's modification of the Child-Turcotte classification.)
- Kamath PS, Wiesner RH, Malinchoc M, et al. A model to predict survival in patients with end-stage liver disease (MELD). Hepatology. 2001;33(2):464–470.
- Kim WR, Mannalithara A, Heimbach JK, et al. MELD 3.0: The Model for End-Stage Liver Disease Updated for the Modern Era. Gastroenterology. 2021;161(6):1887–1895. (Adds sex and albumin; transplant-allocation / mortality tool, not a dosing instrument.)
- Verbeeck RK. Pharmacokinetics and dosage adjustment in patients with hepatic dysfunction. Eur J Clin Pharmacol. 2008;64(12):1147–1161. (First-pass collapse with portosystemic shunting; Phase I oxidation reduced more than Phase II glucuronidation, which is only relatively spared; protein-binding effects.)
- Vilstrup H, Amodio P, Bajaj J, et al. Hepatic Encephalopathy in Chronic Liver Disease: 2014 Practice Guideline by the AASLD and the EASL. Hepatology. 2014;60(2):715–735. (Benzodiazepines and sedatives as precipitants of overt encephalopathy.)
- Lewis JH, Stine JG. Review article: prescribing medications in patients with cirrhosis — a practical guide. Aliment Pharmacol Ther. 2013;37(12):1132–1156.
Last reviewed June 2026. Part of the Psychiatry Education Forum Academy; for clinician education — it supports, and does not replace, individual clinical judgment and current local protocols.
You have the framework. Now get the drug-by-drug answers.
This chapter taught you how to decide whether a psychotropic needs a hepatic adjustment at all — and how to keep the dosing question separate from the hepatotoxicity question. The member chapters give you the agent-by-agent calls, starting with the drug that sits at the center of both: valproate.
Educational use only. Refer to the sources cited above and current prescribing information for clinical decisions. Psychiatry Education Forum and authors assume no liability for use of this material.
Responses