- Design
- Pragmatic randomised controlled trial embedded in routine care
- Population
- 7294 surgical patients with preoperative cognitive impairment at one US hospital
- Primary outcome
- Postoperative delirium by 4AT within seven days
- Effect
- 35.1% vs 32.7% (P = 0.37) among the 17.2% assessed; no change in best-practice adherence
A pragmatic randomised trial at a New York hospital used natural language processing to flag 7294 surgical patients with cognitive impairment, then randomised them to a non-interruptive decision support display in the electronic record, prompting 12 delirium-prevention practices, or to routine care.
Adherence to none of the 12 practices changed. Only 17.2% of patients had a documented delirium assessment, and among them delirium occurred in 35.1% with the prompt and 32.7% without (P = 0.37). The low assessment rate means the trial cannot show whether the tool affects delirium itself.
The lesson is about implementation: a passive display that clinicians can ignore is ignored. Identifying at-risk patients is the easy part.
- Do not rely on passive electronic prompts to change perioperative delirium practice.
- Screen older surgical patients for delirium after surgery with a validated tool such as the 4AT, and record it.
- Assign named responsibility for delirium prevention steps rather than leaving them to an alert.
- Audit delirium assessment rates; you cannot improve what is not measured.
Why it matters
It shows that identifying at-risk patients does not by itself change care.
Don't overread it
Fewer than one in five patients had delirium assessed, so the trial cannot conclude the tool has no effect on delirium.
The statistics, in plain English
Delirium rates of 35.1% and 32.7% with P = 0.37 show no detectable difference, but they come from only the 17% of patients assessed, who may not represent the rest.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for anaesthesiology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free