Abstract Background Delirium is a common, serious, and often under-recognised condition affecting up to 31% of hospitalised older adults. It is associated with increased mortality, prolonged length of stay, and higher care needs post-discharge. Despite national guidelines recommending routine cognitive screening and non-pharmacological management, adherence remains variable. This closed-loop audit aimed to evaluate the impact of a targeted intervention on the early recognition and management of delirium in older medical inpatients. Methods We conducted a retrospective chart review of patients aged ≥65 admitted to two general medical wards in an Irish Model 3 hospital. The audit assessed whether a 4AT cognitive screen was performed within 24 hours, if it was abnormal, whether delirium was formally diagnosed, and whether risk factors were identified and addressed using the PINCH-ME framework. Documentation of conservative (non-pharmacological) management was also reviewed. Following baseline data collection (n=47), an intervention was introduced: a bedside “Delirium Tool” incorporating the 4AT and PINCH-ME prompts, accompanied by regular education sessions. A re-audit was conducted post-intervention (n=52). Results Following the intervention, 4AT screening within 24 hours improved from 34% to 44%. The proportion of patients formally diagnosed with delirium rose from 8.5% to 25%, aligning with expected prevalence in older inpatients. Identification of delirium risk factors increased from 8.5% to 21%, while documentation of risk factor management and non-pharmacological strategies both improved from 8.5% to 17%. Conclusion This closed-loop audit demonstrates that a structured bedside tool and staff education significantly improved the recognition and initial management of delirium. The observed post-intervention diagnosis rate reflects improved clinical awareness and documentation. Sustained use of structured tools may further enhance care quality for older adults at risk of delirium.
Devlin et al. (Mon,) studied this question.