Postoperative delirium — depth controversy, drug verdicts and the hip-fracture bundle

Examiner prompts and model answers

Is this delirium, and with what tool? Yes — acute fluctuating inattention on day 2 is hypoactive delirium, the phenotype that dominates PACU presentations (64.2%) and defeats gestalt, which can rule delirium in but never out. Confirm with 4AT (sensitivity 74.0%, specificity 95.4%, matching CAM-ICU off the ventilator) and repeat it — fluctuation defeats single snapshots.^[18][14][11]

His anaesthetic was BIS-guided — so depth is handled? Not proven. Prosecute Evered (BIS 50 versus 35: 19% versus 28%, odds 0.58, better 1-year cognition) and the p-EEG metas (RR 0.77–0.81) — then defend ENGAGES (guided 26.0% versus 23.0% despite halved suppression time) and BALANCED (1-year mortality 6.5% versus 7.2%, no difference). The honest synthesis: BIS-target-versus-target is not guided-versus-usual, populations and regions differ, frontal indices miss whole-brain vulnerability — the controversy itself is the answer, and depth is chosen by indication.^{[46][47][44][45][50]}

Which drugs prevent, which treat, which harm? Rank prophylaxis honestly — atypicals, haloperidol, dexmedetomidine, melatonergics in network order — then brake it: heterogeneity precludes firm conclusions and AHRQ finds haloperidol equals placebo at high strength of evidence. Dexmedetomidine cuts cardiac delirium (RR ~0.67–0.70, bradycardia cost, weak placebo signal) but intraoperative-only noncardiac infusion failed (12.2% versus 11.4%) — timing and population own the verdict. Nothing shortens established delirium: MIND-USA's three arms were indistinguishable (p=.26), so treat causes — stop or replace the morphine PCA (postoperative opioids nearly double odds; meperidine triples), remove the catheter, restore sleep — and keep antipsychotics for dangerous agitation only.^{[33][35][38][37][57][42]}

What bundle runs today, and what is the year ahead? All of HELP — orientation, mobilisation, hydration, feeding, sleep, senses, family — the 15.0-to-9.9% original with the 0.47 meta-analysis, nurse-delivered and visitation-backed. Then counsel with numbers: 30-day mortality odds 3.2, 1-year dementia odds 13.9 — and book cognitive follow-up, because prevention is survival medicine.^{[1][2][17][61]}