548 FRAILTY ANALYSIS ACROSS POPULATIONS: THE USE OF THE HOSPITAL FRAILTY RISK SCORE IN SPECIALISED SERVICES USING NHS NATIONAL DATA

Frailty is increasingly used to risk stratify older people, but across specialised services there is no standardised approach. The aim of this study was to answer the question posed by NHS England to assess frailty could be measured and related to outcomes across Specialsed Services. The Hospital Frailty Risk Score (HFRS) was used and the data published in Age and Ageing in 2020. Method A retrospective cohort study was performed using the Secondary Uses Service (SUS) electronic database for people aged 75 or older admitted between April 2017 and March 2018. Based on HFRS, the populations were risk stratified into mild, moderate and severe frailty risk. The relationships with length of stay, readmission rate, mortality and some selected condition specific treatment complications were quantified using descriptive statistics. Results Frailty was differentially distributed across the specialties; around one-third had mild frailty; another third had moderate frailty and one-quarter severe frailty. Increasing frailty risk was associated with increased length of stay for the index admission, more days in hospital in the year following intervention and increased risk of dying in hospital. Severe frailty was a powerful discriminator of the risk of death; between 25 and 40% of those with severe frailty risk died at 30 months across all specialties. Conclusions This study demonstrates the first application of the HFRS to a national dataset to describe service outcomes and mortality for older people undergoing a range of specialised interventions. This information was used to identify those that might benefit from holistic assessment, aid prognostication, commissioning and service planning. It informed the promotion of quality improvement work in this area via the Specialised Clinical Frailty Network. By implementing the HFRS, changes in outcomes can be plotted across years to assess improvements, performance and benchmarking.