Grant: Travel Awards - Institute Community Support

Introduction: Frailty is associated with increased mortality and morbidity among older adults, showing screening for frailty is essential in emergency departments (EDs). Our systematic review aimed to address this critical question: which frailty screening instrument has the highest prognostic accuracy for adverse outcomes in older adults visiting EDs? Methods: Studies included patients aged 60 and over admitted to the ED applying frailty instruments, reporting sensitivity, specificity, and area under the curve (AUC). Designs encompass observational, experimental, or quasi-experimental studies from databases MEDLINE, EMBASE, Cochrane Library, CINAHL, and CNKI up to January 2024. Two investigators independently reviewed abstracts and full-text. Meta-analysis was conducted for frailty tools with consistent cutoffs in at least four studies. Results: Forty-seven studies involving 24 frailty instruments with varied cutoffs (N= 87,198) included The Identification of Seniors at Risk, the most frequently used frailty tool, followed by the Clinical Frailty Scale and the Triage Risk Screening Tool. Most tools demonstrated high sensitivity but low specificity, regardless of the adverse outcomes measured. ISAR's pooled sensitivity, specificity, and AUC for predicting in-hospital mortality and hospitalization were 92% (confidence interval [CI], 88-95), 26% (95% CI, 21-33), 0.82, 81% (95% CI, 71-88), 42% (95% CI, 28-59), and 0.70, respectively. For ED revisits, ISAR's sensitivity, specificity, and AUC were 77% (95% CI, 73-82), 30% (95% CI, 24-38), and 0.64, while TRST showed 73% (95% CI,63-80), 39% (95% CI, 29-50), and 0.60. Conclusions: Most frailty instruments applied in ED demonstrated high sensitivity but low specificity, limiting their usefulness. Future studies should consider incorporating domains beyond geriatric conditions and known prognostic indicators into frailty instruments to enhance their specificity and test their effectiveness in large-scale, long-term studies.