533 READMISSIONS TO GERIATRIC MEDICINE IN A DISTRICT GENERAL HOSPITAL, A THEMATIC ANALYSIS

621 patients were readmitted within 30 days of discharge from our Geriatric Medicine department between October 2019 and March 2020. This project was undertaken to better understand whether some of these readmissions could be prevented. Methods Analysis showed a clear peak of 78 patients returning at day one post discharge, and a subsequent Pareto analysis revealed that nearly 80% of these were discharged from 2 locations, our frailty unit and MAU. This cohort became the focus of our review. With support from the Acute Frailty Network, a virtual structured case note review was undertaken by a multidisciplinary team including an independent Geriatrician. This considered the patient’s case, whether they were cared for by the right team in the right place, whether they were readmitted with a new diagnosis and whether any improvements in the discharge process or wider service could have prevented their readmission. Results The case note review process was halted early as clear recurring themes emerged. Firstly, feedback showed that all frail patients were identified early in the patient pathway, were cared for by the appropriate team and a CGA was commenced promptly. Secondly, whilst the majority of day one readmissions had a diagnosis of delirium, most were not diagnosed as such until their re-attendance, despite documented confusion. Failure to recognise this, treat their delirium, and plan for fluctuations in cognition upon discharge were found to contribute to a high readmission rate. Conclusion We found that delirium identification was a key factor in early readmissions to our service. We identified clear opportunities to strengthen and formalise delirium diagnoses, and better plan patient management and discharge plans as a result. A multidisciplinary quality improvement group looking at delirium has been initiated and has already achieved a statistically significant improvement in the use of 4AT, a delirium assessment tool.