14.3 Encephalopathy, Delirium, and Acute Cognitive Change
Key Takeaways
The 2020 ten-society statement describes acute encephalopathy as a pathobiological process that can present as delirium or coma; delirium is not a clinically lesser psychiatric substitute.
Baseline, serial findings, causes, management and trajectory inform review. No universal score, laboratory threshold, treatment, or return-to-baseline requirement independently establishes every encephalopathy.
The FY 2027 Index distinguishes metabolic G93.41, unspecified toxic G92.9, and toxic metabolic G92.8; follow applicable drug-event instructions and actual provider documentation.
Hepatic encephalopathy NOS or without coma indexes to K76.82 with underlying-liver-disease and exclusion instructions. Do not infer coma or liver failure from confusion or treatment.
Final v44 lists F05 and G93.40 as CCs, and G93.41/G92.8/G92.9 as MCCs when secondary, subject to exclusions. Severity classification does not select the clinical diagnosis.
Acute cognitive change: establish the syndrome and its cause
A sudden change in attention, awareness, cognition, or responsiveness deserves an explanation in the clinical record. CDI must distinguish what observers recorded, what syndrome the clinician established, and what cause is documented. “Confused,” a Glasgow Coma Scale result, a positive delirium assessment, and a diagnosis of metabolic encephalopathy are different kinds of information. One should not be silently substituted for another.
Begin with the patient's baseline and the time course. Family reports, prior relevant records, nursing observations, examination, medication history, and serial assessments can help explain what changed. Preserve their sources and limitations. An existing dementia diagnosis does not automatically explain a new acute deterioration, but an acute illness in a patient with dementia does not automatically establish metabolic encephalopathy either.
Delirium and encephalopathy describe related concepts
The 2020 consensus endorsed by ten professional societies recommends acute encephalopathy for a rapidly developing pathobiological brain process, usually over hours or days and within a period shorter than four weeks. It can present clinically as subsyndromal delirium, delirium, or coma. The statement recommends using the appropriate clinical term for the observed presentation rather than using encephalopathy as a description of bedside findings alone.
Delirium concerns an acute and fluctuating disturbance in attention and awareness with additional cognitive change, evaluated in relation to baseline and the clinical context. It is not merely a psychiatric label for a clinically unimportant problem. The consensus warns that separating the terminology into unrelated disease hierarchies can impede understanding and care.
Coma concerns profoundly depressed responsiveness; it is not interchangeable with ordinary inattention or delirium. Dementia concerns an established neurocognitive disorder whose baseline course and functional effects must be distinguished from a superimposed acute syndrome. A quiet or hypoactive presentation can still warrant assessment; review actual findings instead of looking only for agitation.
These clinical distinctions do not independently authorize a particular ICD-10-CM code set. Do not treat a nomenclature statement as a coding instruction to report every related term separately, or to replace a documented delirium diagnosis with an encephalopathy code. Review the actual diagnoses, classification instructions, reportability, and any unresolved ambiguity.
Build an evidence timeline
| Review question | Useful evidence | Inference to avoid |
|---|---|---|
| What was the baseline? | Established cognition, communication, activities, and usual responsiveness | Assuming all current impairment is dementia because dementia is on the problem list |
| What changed and when? | Dated observations, family description, attention/awareness assessment, fluctuation | Calling any isolated confusion an established acute metabolic disorder |
| What explanations were considered? | Provider differential, examination, relevant testing, exposures | Selecting a cause from one abnormal laboratory value |
| What was actually done? | Evaluation, administered treatment, medication changes, monitoring, safety measures | Treating a medication order as proof of a particular encephalopathy |
| What was the subsequent course? | Improvement, persistence, fluctuation, new information, provider reassessment | Requiring return to baseline before the diagnosis can be valid |
GCS and delirium instruments help describe or assess particular clinical features; neither supplies a universal numerical threshold for every encephalopathy. A positive CAM assessment does not independently establish a metabolic or toxic etiology. A documented GCS may be available under its coding documentation exception, but that exception does not permit the coder to diagnose encephalopathy from the score.
Consider causes without converting them into diagnoses
Metabolic illness, systemic infection, medication or toxin effects, and liver disease can be relevant clinical contexts. Multiple factors may coexist. Review the provider's interpretation of each factor and the alternative explanations rather than making a one-to-one rule from a laboratory threshold to a brain diagnosis.
For example, a creatinine rise with cognitive change warrants clinical assessment; it does not independently establish uremic or metabolic encephalopathy. Similarly, the mere presence of infection does not determine that all cognitive changes are septic encephalopathy. The clinician may identify a different cause, several causes, or unresolved uncertainty.
Medication review should distinguish exposure, prescribed use, administration, suspected adverse effect, overdose, and withdrawal. Sedation around a procedure and a drug-induced pathologic condition are not interchangeable. Document the implicated drug and relationship when established, and apply the relevant drug-event instructions rather than inferring poisoning from somnolence or a reversal-agent order.
The causal assessment can also change over the admission. An early provisional explanation may be ruled out or refined. Resolve meaningful conflicts rather than retaining every earlier possibility as a final diagnosis. Persistent impairment does not automatically invalidate an acute process, and improvement after treatment does not alone prove which process caused it.
Hepatic encephalopathy is a clinical assessment
Liver dysfunction and portosystemic shunting can contribute to hepatic encephalopathy. The AASLD's 2025 educational discussion emphasizes that ammonia results do not consistently track the severity of hepatic encephalopathy and that altered mental status in cirrhosis cannot always be attributed to it. Sample handling and processing can also affect ammonia measurements.
Thus, an elevated ammonia value is not a universal diagnostic or staging cutoff for HE. Preserve the examination, clinical course, precipitating factors, competing explanations, and specialist assessment. Treatment such as lactulose or rifaximin can support the clinical timeline but does not independently establish the diagnosis, coma, or liver failure.
The AASLD's currently posted dedicated chronic-liver-disease guideline is dated 2014; do not present it as a new 2026 definition or apply its chronic-liver population indiscriminately to acute liver failure. For this documentation task, the supported principle is that the clinician assesses the syndrome in its liver-disease context rather than CDI assigning it from ammonia or treatment alone.
Verify exact FY 2027 terms and instructions
| Documented term or distinction | FY 2027 classification point |
|---|---|
| Encephalopathy, unspecified | G93.40; do not invent a metabolic cause to avoid an unspecified diagnosis |
| Metabolic encephalopathy | G93.41, including septic encephalopathy |
| Unspecified toxic encephalopathy | G92.9 |
| Toxic metabolic encephalopathy | G92.8, other toxic encephalopathy; it is not interchangeable with G93.41 |
| Hepatic encephalopathy NOS or without coma | Index directs K76.82; review its code-also and exclusion instructions |
| Delirium due to a known physiological condition | F05, with a code-first instruction for the underlying physiological condition |
| Altered mental status, unspecified | R41.82, a symptom description rather than automatic proof of one of the above diagnoses |
G92.8 and G92.9 include instructions concerning poisoning and adverse effects when applicable: poisoning is coded first under the specified instructions, while an applicable adverse-effect code identifies the drug additionally. A medication history alone does not establish that a drug event occurred.
K76.82 directs coding also the underlying liver disease, such as specified hepatic failure or viral hepatitis when documented. Its exclusions address hepatic failure with coma. Do not infer coma from confusion, or hepatic failure merely from a diagnosis of hepatic encephalopathy. K72.90 means hepatic failure, unspecified, without coma; it does not mean chronic hepatic failure without coma, which is K72.10. Follow the full documented condition and applicable instructions rather than mapping every HE case to K72.-.
F05 includes delirium superimposed on dementia and has its own code-first instruction. Delirium NOS follows a different Index/Tabular path; not every use of the word “delirium” automatically establishes F05 or its underlying condition. Substance-related disorders also have specific classification pathways. Check the complete term and circumstances.
Clinical significance is not a severity-code hierarchy
Final MS-DRG v44 Appendix C lists F05 and G93.40 as CCs, while G93.41, G92.8, and G92.9 are MCCs when secondary, subject to exclusions and complete grouping. K76.82 is not listed as a CC/MCC in Appendix C part 1. F01.511, vascular dementia of unspecified severity with agitation, is a CC; an entire dementia code range must not be labeled non-CC. These facts cannot be generalized to all encephalopathies, all dementia codes, all DRGs, or other risk models.
An MCC designation does not make one diagnostic term clinically preferable, establish that it exists, or guarantee a different payment. A CC or symptom diagnosis can be accurate and clinically important. Do not request a metabolic or toxic label simply to replace delirium or an unspecified diagnosis with a higher severity category.
Ask the question the record actually leaves open
Suppose a patient with a stable documented dementia baseline develops a substantial new attention and responsiveness change during an acute illness. The final notes still say only “confusion.” A justified clarification can present the baseline, dated changes, relevant results, actual interventions, and the provider's existing assessments, then ask about the acute condition and cause, if known. Permit clinically supported alternatives and an inability to determine the answer. Do not announce that a sodium or BUN threshold has already substantiated an MCC.
If an involved consultant has already documented a compatible specific diagnosis, check whether a genuine unresolved question remains; do not demand duplicate attending wording solely as a routine rule. A true diagnostic conflict requires resolution through the appropriate process. Ensure the response meets permanent-record requirements and apply reportability and sequencing separately.
For a clinical-validation concern, ask for the provider's rationale or clarification using relevant supporting and contrary evidence. Baseline, neurological findings, possible causes, management, and trajectory are useful review dimensions. They are not four mandatory pillars guaranteeing a successful appeal, nor a rule requiring a specific drug, biomarker cutoff, or recovery before discharge.
Sources: 2020 ten-society nomenclature consensus; AASLD 2025 ammonia and encephalopathy discussion; AASLD dedicated guideline page and date; FY 2027 ICD-10-CM guidelines, I.A.19, I.B.14 and III; FY 2027 Index and Tabular List; final MS-DRG v44 Appendix C; 2026 query brief.
The provider’s established diagnosis is “toxic metabolic encephalopathy.” Which FY 2027 classification distinction should be applied?
Treat G93.41 and G92.8 as interchangeable and select whichever has the preferred grouping effect.
Infer poisoning automatically because the patient takes medications, before checking the actual drug-event circumstances.
The Index directs toxic metabolic encephalopathy to G92.8; verify applicable poisoning/adverse-effect instructions and do not replace the term with plain metabolic encephalopathy G93.41.
Use G92.9 for every documented toxic subtype because the classification has no distinction for toxic metabolic encephalopathy.
Which statement accurately reflects final MS-DRG v44 Appendix C without using severity to choose a diagnosis?
F05 is always non-CC because it appears in the mental-disorders chapter.
G93.40 is an MCC because every encephalopathy code has the same severity classification.
A patient with delirium should be relabeled metabolic encephalopathy whenever an abnormal laboratory value is present.
F05 and G93.40 are CCs, while G93.41 is an MCC when secondary, subject to exclusions and complete grouping; these labels do not establish the diagnosis.
A clinician documents hepatic encephalopathy without coma in a patient with cirrhosis. No hepatic-failure diagnosis is established. Which FY 2027 classification approach is appropriate?
Assign K72.90 because hepatic encephalopathy automatically establishes hepatic failure, regardless of documentation.
Follow the Index to K76.82, review its instruction to code also underlying liver disease and its exclusions, and do not infer coma or hepatic failure.
Assign G92.8 because lactulose administration establishes a toxic encephalopathy.
Replace the established diagnosis with R41.82 because encephalopathy cannot be coded without a particular ammonia threshold.
Sections you finish are checked off in the contents.