Confusion Assessment Method for the ICU — скринінг делірію у ВІТ за чотирма ознаками: гострий/флюктуючий перебіг, неуважність, порушення свідомості, дезорганізоване мислення
CAM-ICU is built to catch delirium as a change from baseline, not simply an abnormal snapshot. Feature 1 is satisfied when the bedside nurse or examiner documents either (a) an acute change in mental status from the patient's pre-hospital baseline, or (b) a fluctuating course over the past 24 hours — a mental status that has been abnormal, then normal, then abnormal again across shifts. Because delirium is intrinsically a waxing-and-waning process, a single normal exam does not rule it out; the fluctuation itself is diagnostic information.
The examiner asks: is there evidence that the current mental status differs from the patient's pre-admission baseline? Sources of baseline information include:
• Family or caregiver interview: "Was your relative like this at home?" • Prior medical records: documented cognitive baseline, dementia diagnoses • Nursing admission assessment: mental status recorded on ICU arrival
A patient with pre-existing dementia can still screen CAM-ICU positive — the comparison is to THEIR baseline, not to population norms. A sudden new confusion superimposed on chronic cognitive impairment is exactly the acute change this feature is designed to detect.
If no clear baseline comparison is available (common in ICU patients transferred emergently), the examiner instead looks for fluctuation: did the patient's RASS score, level of arousal, or CAM-ICU status change unpredictably over the past 24 hours, as documented across nursing shifts?
Practical markers of fluctuation: • RASS oscillating between different values across consecutive checks (not simply trending toward baseline with sedation weaning) • Alternating periods of lucidity and confusion noted by different shift nurses • Inconsistent orientation or attention on repeated informal checks
Either sub-criterion alone — acute change OR fluctuation — is sufficient to mark Feature 1 positive; both do not need to be present simultaneously.
Feature 1 is a gate, not a nuance: because it is one of the two mandatory features (with Feature 2), a CAM-ICU exam can never be positive without it, no matter how abnormal Features 3 and 4 appear.
Inattention is the cognitive core of delirium and the second mandatory CAM-ICU feature. Rather than relying on subjective impression, the exam uses a standardized, language-independent attention task: the Attention Screening Examination (ASE). The most common version reads a 10-letter sequence aloud at one letter per second and asks the patient to squeeze the examiner's hand only on hearing the letter "A." Because it does not require verbal responses, ASE-letters can be performed even in intubated, non-verbal ICU patients — a critical design requirement for the ICU setting.
Instructions given to the patient: "I am going to read you a series of 10 letters. Squeeze my hand whenever you hear the letter 'A.'" The examiner then reads the sequence S-A-V-E-A-H-A-A-R-T clearly, one letter per second, while holding the patient's hand (or watching for a head nod/eye blink in patients unable to grasp).
Two error types are tallied: • Error of omission: patient fails to squeeze on an "A" • Error of commission: patient squeezes on a non-"A" letter
A total of more than 2 errors (out of 10 opportunities) is scored as a positive (abnormal) Feature 2. The test can be repeated with an alternate letter sequence if a repeat exam is needed on the same shift, to reduce practice effects.
Attention is the gateway cognitive process: without the ability to selectively attend and sustain focus, downstream processes like memory encoding, orientation, and organized thought collapse. This is why inattention — rather than memory loss or disorientation — was chosen as a mandatory CAM-ICU feature; it is present in essentially all delirium subtypes (hyperactive, hypoactive, and mixed), whereas features like agitation are present only in hyperactive delirium.
Inattention on ASE-letters correlates strongly with formal neuropsychological attention batteries and with EEG evidence of diffuse cortical dysfunction seen in delirium, giving it strong construct validity as a bedside proxy for the underlying acute encephalopathy.
Feature 3 asks a deceptively simple question: is the patient anything other than "alert and calm"? This is operationalized using the RASS, a validated 10-point scale from -5 (unarousable) to +4 (combative), with 0 defined as alert and calm. Any RASS score other than 0 satisfies Feature 3 — including agitation as well as sedation. Critically, RASS must be checked before proceeding with the rest of the CAM-ICU: patients at RASS -4 or -5 are too deeply sedated to be meaningfully assessed and are scored as "unable to assess" rather than screened further.
RASS levels (abbreviated):
+4 Combative — overtly violent, danger to staff +3 Very agitated — pulls at tubes/catheters, aggressive +2 Agitated — frequent non-purposeful movement +1 Restless — anxious, movements not aggressive 0 Alert and calm — the target, CAM-ICU-negative state −1 Drowsy — not fully alert, sustained awakening to voice (>10 s) −2 Light sedation — briefly awakens to voice (<10 s) −3 Moderate sedation — movement/eye opening to voice, no eye contact −4 Deep sedation — no response to voice, movement to physical stimulation −5 Unarousable — no response to voice or physical stimulation
RASS serves two roles in the CAM-ICU pathway: first as a gate (only patients at RASS ≥ −3 can proceed to the full exam), and second, if the exam proceeds, as Feature 3 itself — any non-zero RASS counts as positive.
A patient who is simply well-sedated for a procedure (RASS -2, calm, appropriately drowsy from a titrated propofol infusion) has an abnormal RASS but is not necessarily delirious — this is precisely why Feature 3 is only one of two "either/or" features (paired with Feature 4), and why Features 1 and 2 remain mandatory. Sedation alone shifts RASS without necessarily producing the acute fluctuating inattentive state that defines delirium.
Conversely, agitation (RASS +2/+3) is often mistakenly treated as "just agitation" and sedated away, when it may represent hyperactive delirium requiring evaluation of reversible causes (pain, hypoxia, withdrawal, infection) rather than escalating sedatives.
Deep sedation is deliberately excluded from CAM-ICU scoring (RASS −4/−5 = "unable to assess") because inattention and disorganized thinking cannot be reliably tested in an unarousable patient — the exam is only valid in a patient who can engage, even briefly.
Feature 4 probes higher-order reasoning: can the patient follow a simple logical chain and execute a basic instruction? The exam uses two components — four simple yes/no questions with a single correct answer each, and a two-step command. Because it is only tested when Features 1–3 leave the picture ambiguous (RASS 0 with positive 1 and 2, needing an OR-partner), Feature 4 is often the tie-breaker that determines whether an inattentive, fluctuating patient with a normal RASS is scored delirious.
A representative question set (Set A): 1. Will a stone float on water? 2. Are there fish in the sea? 3. Does one pound weigh more than two pounds? 4. Can you use a hammer to pound a nail?
Each has an unambiguous correct answer and requires only simple logical reasoning, not specialized knowledge, minimizing the influence of education or language proficiency. An alternating Set B is used if the exam must be repeated later in the same shift, to reduce the chance the patient is simply recalling the prior answers rather than reasoning through them again.
After the questions, the examiner says: "Hold up this many fingers" (examiner holds up two fingers), then "Now do the same with the other hand" (without repeating the number of fingers) — or an equivalent two-part instruction adapted for a patient with limited limb mobility.
This command tests working memory (retaining the instruction across the two steps) and motor planning, both of which are impaired in disorganized thinking. Errors across the four questions and the command are summed; a combined total of more than 1 error scores Feature 4 positive. Because this component depends on some residual motor and verbal capacity, it is adapted (or waived to rely more heavily on RASS) in patients with severe baseline physical limitations.
The full power of CAM-ICU comes from how its four features are combined, not merely how many are positive. Delirium is flagged only when Feature 1 AND Feature 2 are both positive, AND at least one of Feature 3 OR Feature 4 is also positive. This deliberately mirrors DSM criteria for delirium: a disturbance in attention with an acute/fluctuating course (Features 1+2, mandatory), plus an additional disturbance in either consciousness or cognition (Features 3 or 4, either sufficient). Three positive features that miss this exact pattern — for example, Features 1, 3, and 4 positive but Feature 2 negative — still score CAM-ICU negative.
Formally: CAM-ICU positive ⟺ (Feature 1 = positive) AND (Feature 2 = positive) AND [(Feature 3 = positive) OR (Feature 4 = positive)]
This is NOT a simple majority rule ("3 of 4 features positive = delirium"). Consider two hypothetical patients, both with 3 of 4 features positive:
Patient A: Feature 1 positive, Feature 2 positive, Feature 3 positive, Feature 4 negative → CAM-ICU POSITIVE (1 AND 2 satisfied, 3 satisfies the OR)
Patient B: Feature 1 positive, Feature 3 positive, Feature 4 positive, Feature 2 NEGATIVE → CAM-ICU NEGATIVE (Feature 2 is mandatory and missing, regardless of how many other features are positive)
This logic reflects the diagnostic reality that inattention with an acute/fluctuating course is the irreducible core of delirium — without it, an abnormal RASS or missed yes/no question more likely reflects sedation, baseline cognitive impairment, or a different process entirely.
Bedside teaching pearl: always check RASS and administer Features 1 and 2 first. If either is negative, the algorithm already resolves to CAM-ICU negative and Features 3/4 do not need to be tested for scoring purposes (though a full exam is still often documented).
A positive CAM-ICU screen should trigger a structured response rather than sedation alone:
1. Search for and treat reversible contributors: hypoxia, hypotension, infection/sepsis, metabolic derangement (glucose, sodium, ammonia), pain, alcohol/benzodiazepine withdrawal, urinary retention/constipation, deliriogenic medications (benzodiazepines, anticholinergics) 2. Apply the ABCDEF bundle: Assess/manage pain, Both spontaneous awakening and breathing trials, Choice of sedation/analgesia (favor non-benzodiazepine agents), Delirium monitoring/management, Early mobility, Family engagement 3. Optimize the environment: restore day/night cycle, minimize noise, provide glasses/hearing aids, encourage family presence and orientation 4. Reserve antipsychotics for severe agitation causing safety risk — they treat symptoms, not the underlying delirium, and have not been shown to shorten delirium duration
CAM-ICU should be repeated at least once per nursing shift, since the fluctuating nature of delirium (Feature 1) means a single negative screen does not exclude an evolving episode.