An acute, fluctuating disturbance of attention and awareness
"Delirium is acute brain failure — as much an organ dysfunction as acute kidney injury or the acute respiratory distress syndrome, and just as deserving of monitoring, prevention and a search for a reversible cause."
Summarised from Marino's The ICU Book; with the Washington Manual of Critical Care and Irwin & Rippe's Intensive Care Medicine.Delirium (DSM-5) is an acute (hours–days), fluctuating disturbance of attention and awareness, with an additional change in cognition (memory, orientation, language, perception), that is a direct physiological consequence of a medical condition, intoxication or withdrawal — and not better explained by a pre-existing dementia or a severely reduced level of arousal such as coma.
- Acute onset & fluctuating course
- Inattention (the cardinal feature)
- Disorganised thinking or altered level of consciousness
- An identifiable organic cause
Hyperactive, hypoactive and mixed
Motor subtype is defined by the level of arousal and psychomotor activity. The commonest form is the one most often missed.
| Subtype | Features | Frequency / prognosis |
|---|---|---|
| Hypoactive | Withdrawn, flat, drowsy, slow — mistaken for depression or fatigue | Commonest in the ICU; frequently missed; worse prognosis |
| Hyperactive | Agitated, restless, pulling at lines/tubes, hallucinating | Most easily recognised; least common alone |
| Mixed | Fluctuates between the two, often over hours | Very common |
Clinical point: because hypoactive delirium looks "settled", it is under-diagnosed unless you screen — quiet is not the same as well.
Delirium is not benign
- Independently associated with higher mortality, longer ventilation and longer ICU/hospital stay.
- Predicts long-term cognitive impairment (a dementia-like decline) months after discharge.
- Causes line/tube removal, falls, and the harms of the sedatives and restraints used to control it.
- Every day of delirium raises the risk of a worse functional outcome.
Predisposing vs precipitating
Delirium is almost always multifactorial: a vulnerable brain (predisposing) meets an acute insult (precipitating). The more vulnerable the patient, the smaller the insult needed.
| Predisposing (baseline) | Precipitating (acute — often modifiable) |
|---|---|
| Older age, pre-existing dementia/cognitive impairment | Benzodiazepines, deep sedation, anticholinergics, opioids |
| Frailty, functional dependence | Sepsis/systemic inflammation, hypoxia, shock |
| Visual/hearing impairment | Metabolic derangement (Na, glucose, uraemia, ammonia) |
| Alcohol/substance use, prior delirium | Sleep deprivation, immobility, restraints, pain |
| Multiple comorbidities | Sedation-related, infection, drug withdrawal |
Of all precipitants, benzodiazepine exposure and depth of sedation are the ones most under your direct control at the bedside.
You cannot assess thinking in a patient you cannot rouse
Screening is a two-step process done at least once per shift. First quantify arousal with the RASS; only if the patient is rousable (RASS ≥ −3) do you apply the CAM-ICU.
Step 1 — Richmond Agitation–Sedation Scale (RASS)
| Score | State |
|---|---|
| +4 to +1 | Combative → restless (agitation) |
| 0 | Alert and calm (target) |
| −1 to −3 | Drowsy → moderate sedation (responds to voice) |
| −4 / −5 | Responds to physical stimulus only / unrousable |
Target for most ventilated patients: RASS −2 to 0 (light sedation). At RASS −4/−5 the patient is too deeply sedated to test — that is over-sedation, not calm.
Step 2 — CAM-ICU algorithm (a flowchart)
Feature 1 — Acute change or fluctuating course of mental status
Is there an acute change from baseline, OR has mental status fluctuated in the past 24 h? No → not delirious. Yes → go on.
Feature 2 — Inattention
e.g. the "SAVEAHAART" letter test — squeeze on the letter A. >2 errors = inattention present. No error → not delirious. Present → go on.
Feature 3 — Altered level of consciousness
Current RASS anything other than 0? If yes → CAM-ICU POSITIVE (delirium). If RASS = 0, test Feature 4.
Feature 4 — Disorganised thinking
Simple yes/no questions + a command. >1 error → CAM-ICU POSITIVE. Otherwise negative.
CAM-ICU is positive when Feature 1 AND Feature 2 AND (Feature 3 OR Feature 4) are present. The ICDSC (8-item checklist over a shift) is an equally valid alternative.
Screen positive? Now hunt the cause
A positive screen is a prompt to find the reversible driver. A useful mnemonic is "DELIRIUM(S)" / "I WATCH DEATH" — work through them systematically.
| Category | Examples to actively exclude |
|---|---|
| Drugs / withdrawal | Benzodiazepines, opioids, anticholinergics, steroids; alcohol/benzo withdrawal |
| Infection | Sepsis, pneumonia, UTI, CNS infection |
| Metabolic | Hypo/hyperglycaemia, Na/Ca disturbance, uraemia, hepatic encephalopathy, hypoxia, hypercapnia |
| Neurological | Stroke, seizure/non-convulsive status, raised ICP, PRES |
| Deficiency | Thiamine (Wernicke) — give before glucose in the at-risk |
| Environment | Pain, urinary retention/constipation, sleep loss, immobility |
New agitation is hypoxia, hypoglycaemia, or urinary retention/pain until proven otherwise — check these before reaching for a sedative.
Non-pharmacological prevention is the cornerstone
No drug reliably prevents delirium. The evidence base is the ABCDEF ("A2F") bundle — bundle adherence reduces delirium, ventilator days and mortality.
| Letter | Element |
|---|---|
| A | Assess, prevent and manage pain |
| B | Both spontaneous awakening trials (SAT) and spontaneous breathing trials (SBT) |
| C | Choice of analgesia & sedation (light sedation; avoid benzodiazepines) |
| D | Delirium — assess, prevent and manage (screen every shift) |
| E | Early mobility and exercise |
| F | Family engagement and empowerment |
Plus the basics every night: restore the day–night cycle (light by day, dark/quiet at night, cluster care to protect sleep), give back glasses and hearing aids, reorient, and remove lines/catheters/restraints as soon as possible.
Treat the cause; drugs are for danger, not for a quiet ward
Treat the precipitant
Correct hypoxia, sepsis, metabolic derangement, pain, retention; stop/deprescribe deliriogenic drugs (especially benzodiazepines and anticholinergics).
Optimise sedation
Target light sedation (RASS −2 to 0). If a ventilated patient needs sedation and is agitated/delirious, dexmedetomidine is preferred over benzodiazepines (less delirium, shorter time to extubation).
Non-pharmacological first
Reorientation, family presence, sleep hygiene, mobilisation, sensory aids — for every patient, every shift.
Antipsychotics — only for distressing agitation
Haloperidol or an atypical (e.g. quetiapine) may be used short-term to control agitation that threatens safety — they do not shorten delirium duration or improve survival, so they are not for prevention or routine treatment. Watch QTc and extrapyramidal effects.
Trials (e.g. MIND-USA) show antipsychotics do not reduce delirium duration or mortality. Reserve them for the agitated, endangering patient — and stop them at discharge.
Where teams go wrong
- Calling a withdrawn, quiet patient "comfortable" — missing hypoactive delirium by not screening.
- Reaching for a benzodiazepine to settle agitation — the very drug that worsens delirium.
- Treating the CAM-ICU result instead of hunting the cause (hypoxia, sepsis, retention, pain).
- Using antipsychotics as prophylaxis or expecting them to shorten delirium.
- Over-sedation (RASS −4/−5) mistaken for good control; deep sedation prolongs ventilation and delirium.
- Continuing ICU antipsychotics at ward transfer/discharge — a common medication-safety error.
High-yield for NEET-SS / DNB
- CAM-ICU = Feature 1 (acute/fluctuating) AND 2 (inattention) AND (3 altered LOC OR 4 disorganised thinking).
- Inattention is the cardinal feature; hypoactive is the commonest and most-missed subtype.
- Assess RASS before CAM-ICU; target RASS −2 to 0.
- PADIS: light sedation, avoid benzodiazepines, prefer propofol/dexmedetomidine.
- Prevention = ABCDEF bundle; no drug prevents delirium.
- Antipsychotics do not shorten delirium (MIND-USA) — for dangerous agitation only.
- ICDSC is the valid checklist alternative to CAM-ICU.
Side-by-side comparisons
The classic exam traps: delirium vs dementia, and the two sedatives that pull in opposite directions on delirium risk.
Delirium vs Dementia
| Feature | Delirium | Dementia |
|---|---|---|
| Onset | Acute (hours–days) | Insidious (months–years) |
| Course | Fluctuating, worse at night | Slowly progressive, stable day-to-day |
| Attention | Impaired (cardinal) | Usually preserved until late |
| Consciousness | Altered/clouded | Clear until late |
| Reversibility | Usually reversible (treat cause) | Generally irreversible |
Dexmedetomidine vs Benzodiazepines (for sedation)
| Feature | Dexmedetomidine | Benzodiazepine (e.g. midazolam) |
|---|---|---|
| Mechanism | Central α₂-agonist | GABA-A potentiation |
| Delirium risk | Lower | Higher (independent risk factor) |
| Arousability | "Cooperative" sedation, easily roused | Deeper obtundation |
| Respiratory drive | Largely preserved | Depressed |
| Notable effects | Bradycardia, hypotension | Accumulation, tolerance, withdrawal |
| PADIS stance | Preferred | Avoid where possible |
References
- Marino PL. Marino's The ICU Book. 5th ed. Wolters Kluwer; 2025.
- Kollef MH, Isakow W, Burks AC, Despotovic VN (eds). The Washington Manual of Critical Care. 4th ed. Wolters Kluwer; 2024.
- Irwin RS, Lilly CM, Mayo PH, Rippe JM (eds). Irwin & Rippe's Intensive Care Medicine. 9th ed. Wolters Kluwer; 2023.
- Devlin JW, Skrobik Y, Gélinas C, et al. Clinical Practice Guidelines for the Prevention and Management of Pain, Agitation/Sedation, Delirium, Immobility, and Sleep Disruption in Adult Patients in the ICU (PADIS). Crit Care Med. 2018;46(9):e825–e873.
- Girard TD, Exline MC, Carson SS, et al. (MIND-USA). Haloperidol and Ziprasidone for Treatment of Delirium in Critical Illness. N Engl J Med. 2018;379:2506–2516.
- Ely EW. The ABCDEF Bundle: Science and Philosophy of How ICU Liberation Serves Patients and Families. Crit Care Med. 2017;45(2):321–330.