296 Risk-prediction in the Emergency Department: Comparing the Accuracy of Short Frailty Screens to Identify Adverse Healthcare Outcomes among Older Adults

Abstract

Abstract Background Multiple instruments are available to screen for frailty in the Emergency Department (ED). Despite this, few studies have compared their predictive validity among older adults attending ED. This study aimed to investigate the diagnostic accuracy of a variety of different short frailty and risk-prediction instruments to predict 30-day readmission, length of stay (LOS), one-year mortality and institutionalisation. Methods Consecutive patients aged ≥70 attending a university hospital ED were screened and assessed for frailty. Outcomes were obtained from hospital records. The following instruments were compared: the Clinical Frailty Scale (CFS), PRISMA-7, Identification of Seniors at Risk tool, FRAIL scale, Groningen Frailty Indicator (GFI) and Risk Instrument for Screening in the Community (RISC). Results In all, 193 patients were included, median age 79+/-9; 55% were female. Based upon a CGA, 60% (116/193) were classified as frail. Those identified as frail were significantly older (p=0.03) and reported lower quality of life scores (p<0.001). There was no significant difference in co-morbidity using the Charlson Index (p=0.15). The ED conversion rate was 77%, median LOS 8+/-9 days and 20% were re-admitted within 30 days. At one-year, 13.5% were accepted for long-term care and 17% had died. Comparing instruments, the combined RISC was had the highest accuracy based on the area under the ROC curve (AUC) scores for predicting mortality and nursing home admission at one year, AUC 0.77 (95% CI:0.68-0.87) and 0.73 (95% CI:0.64-0.82), respectively. The GFI, CFS and PRISMA-7 had statistically similar, albeit lower scores. No instrument was accurate in predicting 30-day readmission after discharge (AUC <0.70). Conclusion Short frailty screening instruments applied in ED have poor-modest predictive validity for important healthcare outcomes, particularly hospital re-admission. The RISC score had the highest diagnostic accuracy for institutionalisation and death but this was fair at best, suggesting that instrument selection should be pragmatic with the expectation of identifying frailty.