Emergency laparotomy in an 81-year-old — delirium prediction, prevention bundle and cause-hunt

Candidate tasks and model management

Diagnose and show it was predictable. Name hypoactive postoperative delirium — acute fluctuating inattention at 36 hours, the PACU-dominant phenotype gestalt misses — confirmed by 4AT/CAM and repeated because fluctuation defeats snapshots.^[18][11] Show the prediction: Marcantonio high stratum (age over 70, poor cognition, abnormal potassium, major thoracic-equivalent emergency surgery — the 50% group) with cohort multipliers (prior risk, frailty, comorbidity burden).^[7][19] Hip-fracture numbers frame the stakes: 23% pooled delirium in the index operation.^[16]

Run the bundle and fix the physiology. Deploy all six HELP targets — orientation, sleep protection, mobilisation, fluids and feeding, vision/hearing aids, dehydration correction — quoting 15.0% to 9.9% (odds 0.60) and the 0.47 meta-analysis, executed through nurse-owned orders (51% to 33% in hips) with family at the bedside (44.4% to 29.5% with flexible visitation).^{[1][2][29][30]} Hunt causes in parallel: correct hypoxia and hypokalaemia, treat pain without deliriogenics, remove the catheter, rationalise every anticholinergic — the AGS goals are less severity, shorter course, kept safety.^[4]

Rationalise drugs and refuse the shortcut. Convert morphine PCA to multimodal analgesia with regional technique where feasible — postoperative opioids nearly double delirium odds and meperidine triples them — and review the whole chart for anticholinergic burden, which erases even haloperidol's hint of benefit.^[42][43] Refuse scheduled antipsychotics twice over: prevention equals placebo at high strength of evidence, and MIND-USA showed haloperidol and ziprasidone identical to placebo for shortening established delirium (p=.26) — exception only for dangerous agitation, lowest dose, shortest course, monitored.^[35][57]

Close with prognosis and follow-up. Tell the family the numbers: 30-day mortality odds 3.2 with lost independence, 1-year dementia odds 13.9 — then book cognitive follow-up before discharge and hand the GP a delirium flag, because this was a prognosis, not an episode.^[17][61]