Assessment of delirium

Assessment has two distinct steps: screening (a rapid bedside tool that flags possible delirium) and diagnosis (a full clinical evaluation against DSM-5 or ICD-10 criteria). A positive screen is not a diagnosis.

Why delirium is missed: the cognitive baseline is often unknown; there may be no reliable informant; symptoms wax and wane, so one assessment can look completely normal hours later; and hypoactive delirium is quiet and easily mistaken for tiredness, depression or "just being old".

Anchor the assessment on the core features — attention, awareness and level of consciousness — not on behaviour. This is the single most useful strategy when dementia coexists or when no informant is available.

DSM-5: (i) disturbance of attention and awareness; (ii) develops over a short period, a change from baseline, with fluctuation during the day; (iii) an additional disturbance in cognition; (iv) not better explained by a pre-existing, established or evolving neurocognitive disorder and not occurring in coma; (v) evidence of an underlying medical condition, substance intoxication or withdrawal, or multiple aetiologies.

ICD-10 criteria are similar but add detail on perceptual distortion (illusions, hallucinations), psychomotor disturbance, sleep–wake cycle disruption and emotional disturbance.

Kane's practical three questions for any confused older patient: was the onset acute (hours to days)? Are there physical factors (illness, drugs, sensory deprivation)? Are there psychological factors (depression, psychosis) contributing?

Confusion Assessment Method (CAM) — the most widely used. Four features: (i) acute onset and fluctuating course; (ii) inattention; (iii) disorganised thinking; (iv) altered level of consciousness. Delirium is screened positive if (i) AND (ii) are present, PLUS either (iii) OR (iv). Sensitivity 94%, specificity 89%, with high inter-rater reliability.

Testing inattention at the bedside: months of the year backwards, days of the week backwards, digit span 3 or 4 backwards; or simply observe that the patient is distracted or cannot follow the conversation.

Testing disorganised thinking: orientation to time and place, and observing an illogical flow of ideas.

3D-CAM — a structured 3-minute interview with scripted questions. In patients with dementia: sensitivity 96%, specificity 86%. In patients without dementia: sensitivity 93%, specificity 96%.

Ultra-brief screens: the single best item is "months of the year backwards" (sensitivity 83%, specificity 69%); the best two-item screen is "months of the year backwards" + "what is the day of the week?" (sensitivity 93%, specificity 64%).

CAM-ICU — for intubated or non-verbal patients. Inattention is tested with the letter task ("squeeze my hand every time you hear the letter A" → SAVEAHAART); altered consciousness is any RASS score other than 0; disorganised thinking uses yes/no questions ("will a stone float on water?").

Screening for delirium superimposed on dementia (DSD): best single item = "What day of the week is it?"; best two items = days of the week backwards + day of the week; best three items add "What type of place is this?" and "Does the patient appear sleepy?". Sensitivities are high (84%, 93%, 94%) but specificities low (41%, 30%, 42%) — so a negative screen helps rule out delirium, but a positive screen needs fuller evaluation.

Other validated tools named in Pathy: SPMSQ, Delirium Rating Scale and DRS-R-98, Delirium Symptom Interview, Delirium Assessment Scale, Cognitive Test for Delirium, NEECHAM Confusion Scale, Confusional State Evaluation, Memorial Delirium Assessment Scale, Delirium Index, Delirium Severity Scale, Delirium Detection Score, DDT-Pro, bCAM, 4AT, and the Stanford Proxy Test for Delirium. Prediction tools: E-PRE-DELIRIC and Stanford's Algorithm for Predicting Delirium (SAPD).

Screening frequency: hospitalised patients at least daily; post-acute care patients weekly; long-term care residents monthly — and at any time a clinician, carer or relative reports a change from baseline.

Never explain away a positive item. Do not attribute a failed test to being in hospital, being unwell, being old, having just had analgesia, or having slept badly — sleep disturbance is itself part of delirium.

When in doubt, assume delirium. If collateral history is unavailable or the patient is too drowsy to complete an interview, delirium should be presumed until proven otherwise because the adverse outcomes are so severe.

After a positive screen, proceed to a comprehensive medical evaluation — full history including collateral and drug history, examination, targeted investigations, and systematic review of every predisposing and precipitating factor in the DELIRIUM IS VERY PAINFUL table. Neuroimaging, LP and EEG only on specific indication.