Postoperative delirium in the PACU

Why it matters

Delirium after surgery is diagnosed late because it is looked for late, and it is looked for late because screening has been treated as a ward activity. The 2024 European Society of Anesthesiology and Intensive Care guideline is the document usually quoted to move that start line, and it needs reading carefully, because what it actually requires in the recovery room is not a screen but a treatment: if postoperative delirium is detected, patients should not be discharged from the recovery room to the ward without having started cause-based and symptom-based treatment, and the guideline adds that the longer the delirium lasts and the later the treatment starts, the more likely cognitive decline and worse clinical outcomes become aldecoa-2024.

The consequence of getting this wrong is the quiet patient. The form of delirium that is easiest to miss is the one that does not disturb anyone, and a screen applied only when someone is disturbed will not find it.

Mechanism

The mechanism of postoperative delirium is not settled, and the status chip on this page says so. Candidate contributions include neuroinflammation triggered by surgical injury, disruption of neurotransmission by anesthetic and sedative drugs, disturbed sleep and circadian rhythm, pain and its treatment, and the metabolic derangements of the perioperative period acting on a brain with reduced reserve. What is settled is the clinical shape: an acute disturbance of attention and awareness that fluctuates, which is why a single observation at one time of day can miss it and why the guideline specifies a validated tool applied repeatedly.

Evidence

The 2024 ESAIC guideline defines postoperative delirium per DSM-5. It contains 13 numbered recommendations, states that the certainty of evidence across them ranges from very low to moderate, and searched the literature from 1 April 2015 to 28 February 2022 aldecoa-2024. Its treatment recommendations are graded weak throughout: low-dose haloperidol if non-pharmacological measures fail, bolus-wise and at the lowest possible dose, with caution or avoidance in Parkinson disease and Lewy body dementia, at very low certainty; benzodiazepines not suggested for treatment except in alcohol withdrawal, very low certainty; and dexmedetomidine suggested for treatment only in cardiac surgery, very low certainty, because seven of its eight studies were cardiac aldecoa-2024. Its haloperidol dosing example is 0.125 to 0.25 mg as a single dose with a maximum below 3 mg per day, noting increased mortality at 6 mg per day or more aldecoa-2024. On prevention it does not suggest any drug as prophylaxis, at low certainty and weak strength, and where dexmedetomidine is used with that aim it recommends balancing the benefit against bradycardia and hypotension, at moderate certainty and strong strength, with bradycardia at an odds ratio of 1.60, 95% CI 1.30 to 1.96 aldecoa-2024.

It supersedes the 2017 ESAIC guideline, which is cited on this site only to show what the update changed aldecoa-2017.

What this does not show

The guideline rests on a large randomized literature for prevention. It does not rest on trials for the claim that emergence agitation and postoperative delirium are different entities with different time courses, or for the claim that the hypoactive form is both commoner and more often missed; those are descriptive epidemiology and screening-performance claims, and the guideline’s own certainty of evidence is very low to moderate aldecoa-2024. This day is therefore tagged as resting on a guideline, with the prevention material understood as the trial-based portion.

The rule the director’s content list places here, that hypoxemia, hypoglycemia and pain must be excluded before either the delirium or the emergence-agitation label is applied, was searched for in the full text of the guideline on 2026-09-06 and is not there. None of its 13 recommendations states it. The nearest text is its statement that alcohol-withdrawal delirium is a clinical diagnosis of exclusion after considering all other diagnoses aldecoa-2024. The rule is sound practice and it is not this guideline’s; it needs a different source or it should be carried as local teaching.

Pediatric emergence delirium after sevoflurane has its scale and its relative risks. The Pediatric Anesthesia Emergence Delirium scale has five items, eye contact with the caregiver, purposeful actions, awareness of surroundings, restlessness and inconsolability, each scored 1 to 4 and summed, with internal consistency 0.89 and reliability 0.84 (0.76 to 0.90); scores were higher after sevoflurane than halothane, P less than 0.008, and the abstract gives no diagnostic cut-off sikich-2004. A Cochrane review of 158 trials in 14,045 children found, against sevoflurane, a relative risk of 0.51 (0.41 to 0.63) with halothane and 0.35 (0.25 to 0.51) with propofol throughout, and as adjuncts a relative risk of 0.37 (0.29 to 0.47) with dexmedetomidine and 0.37 (0.27 to 0.50) with fentanyl, with the authors flagging pain as a confounder of the diagnosis; it gives relative risks, not an absolute incidence costi-2014.

Two studies now give an absolute incidence, which this day previously lacked, and they come with a warning about how to read it. In a prospective cohort of 137 children aged 4 to 12 having ambulatory circumcision under sevoflurane induction and maintenance, emergence delirium occurred in 75 of 137 (54.7%) by a PAED score of 10 or more, and emergence agitation in 73 of 137 (53.3%) by a RASS of 1 or more zhu-2023-emergence-delirium. In 155 children aged 0 to 3 having fundus examination under sevoflurane, the incidence was 51%, also by the PAED scale, though the cut-off used is not stated and must not be assumed zhao-2026-emergence-delirium. Quote either figure with its instrument and its threshold attached. In one cohort of 1222 children the reported incidence moved from 89.0% at a PAED of 10 or more to 19.3% at a PAED above 12, in the same patients. A number for pediatric emergence delirium without its cut-off is not a fact about children; it is a fact about where somebody drew a line.

Dexmedetomidine has no US indication for delirium, emergence agitation or shivering. Its label indications are sedation of intubated adults in intensive care by infusion not exceeding 24 h, and procedural sedation of non-intubated adults and children; emergence delirium and agitation appear only as pediatric adverse reactions after discontinuation; and its principal warnings are hypotension and bradycardia, including fatalities, and transient hypertension during loading fda-dexmedetomidine-label-2026. Any use for delirium or agitation is off-label, and this page recommends none.

At the bedside

Screen in the PACU, with the validated tool your institution has adopted, and record the result. Be clear about why: the guideline’s instruction to start screening in the recovery room and to continue at least twice daily to day 3 is addressed to future delirium studies, and the same wording elsewhere in the document is the rule the task force used to decide which studies counted as evidence aldecoa-2024. The graded clinical obligation that does bear on the PACU is the one at the other end of the encounter: a delirium found here has to have its treatment started here aldecoa-2024. Local practice differs. the screening tool used, who applies it, and how the result is recorded differ between institutions.

Before you label a patient delirious or agitated, look for what else could produce an inattentive or restless patient in the first hours after surgery, and treat what you find. The exclusion rule awaits verification against the guideline text, but the reasoning behind it is the reasoning of every other day on this site: explain the sign before you name it.

Hold the guideline’s recommendations with the firmness its own grading allows. Very low to moderate certainty is a reason to screen, because the cost of screening is low, and a reason to be modest about what any single preventive measure will achieve aldecoa-2024. Hold this page’s own framing loosely too. Until 2026-09-06 it said the guideline requires screening with a validated tool from the day of surgery onward. Reading the full text showed that sentence to be the guideline’s evidence-inclusion definition, with the recovery-room start recommended for future studies rather than for practice. The page has been rewritten around what the document actually grades, and the rotation director has to agree with that reading before this day leaves draft.

Local practice

Two things on this day are decided by your institution rather than by the guideline, and the page says so rather than asserting one answer.

Local practice differs. which validated delirium screening tool a unit uses differs between institutions. The ones in common adult use are the confusion assessment method for the intensive care unit, the nursing delirium screening scale, and the 4AT. None of the three is named in the body of the 2024 ESAIC guideline; its tool list sits in a supplementary table that has not been retrieved, so treat the choice as institutional. Using none is not a choice.

Local practice differs. the non-pharmacological bundle applied before any drug differs between institutions. It commonly includes reorientation, protected sleep, restoring glasses and hearing aids, early mobilization, hydration and pain control. The guideline’s graded position on what follows if that bundle fails is cited in the evidence slot above.

Check yourself

Three items. Every option carries an explanation. Progress is not saved.