Closing the gap in postoperative delirium detection: addressing low screening rates in the surgical recovery room

Postoperative delirium (POD) is a common complication that increases the incidence of a variety of adverse patient outcomes. Routine screening of high-risk patients can mitigate these risks. This quality improvement (QI) project took place at Chelsea and Westminster Hospital using the Plan-Do-Study-Act (PDSA) cycle methodology. The project aimed to implement a routine POD screening programme as recommended by the National Institute for Health and Care Excellence (NICE).

The first cycle formalised a trust-wide guideline specific to perioperative delirium assessment. The second cycle made delirium tools easier to access on trust systems and embedded them into existing clinical workflows.

The primary outcome of baseline and post-intervention POD screening rates was compared using Fisher’s exact test. Screening rates for 246 high-risk POD patients (as defined by NICE guidelines) admitted to Chelsea and Westminster Hospital between January and May 2025 were included for analysis. The secondary outcomes were self-reported nursing measures (confidence in delirium screening and perceived guideline clarity). Pre- and post-intervention survey scores were compared using the Mann-Whitney U test. Post-intervention surveys were also used to assess the perceived long-term feasibility of interventions.

POD screening rates increased from a baseline of 2.3% to 71.64% (p<0.0001) in the post-intervention period. Of those screened, 17.5% (n=11) of patients screened positive for POD. Reported confidence and clarity of guidelines increased (p<0.0001) to a median (IQR) of 10/10 (8.75–10). Long-term feasibility scores achieved a median (IQR) score of 9/10 (8–10).

The formalisation of a delirium guideline and its integration with clinical workflows significantly increased screening rates and nursing confidence while detecting previously unrecognised cases of POD. This project’s interventions can be adopted by trusts facing similar challenges. Limitations included the likelihood of observer bias, the use of a surrogate marker as a primary outcome and a small sample size, which limits the robustness of statistical analysis.