Prognosis for older people at presentation to emergency department based on frailty and aggregated vital signs.

BACKGROUND: Risk stratification for older people based on aggregated vital signs lack the accuracy to predict mortality at presentation to the Emergency Department (ED). We aimed to develop and internally validate the Frailty adjusted Prognosis in ED tool (FaP-ED) for 30-day mortality combining frailty and aggregated vital signs. METHODS: Single-center prospective cohort of undifferentiated ED patients aged 65 or older, consecutively sampled upon ED presentation from a tertiary Emergency Center. Vital signs were aggregated using the National Early Warning Score (NEWS) as a measure of illness or injury severity and frailty was assessed with the Clinical Frailty Scale (CFS). The FaP-ED was constructed by combining NEWS and CFS in multivariable logistic regression. The primary outcome was 30-day mortality. Measures of discrimination and calibration were assessed to evaluate predictive performance and internally validated using bootstrapping. RESULTS: 2250 patients were included, 67 (1.8%) were omitted from analyses due to missing CFS, loss to follow-up, or terminal illness. Thirty-day mortality rate was 5.4% (N = 122, 95% CI = 4.5%-6.4%). Median NEWS was 1 (Inter-Quartile Range (IQR): 0-3) and median CFS was 4 (IQR: 3-5). The Area Under Receiver Operating Characteristic (AUROC) for FaP-ED was 0.86 (95% CI = 0.83-0.90). This was significantly higher than NEWS (0.81, 95% CI = 0.77-0.85, DeLong: Z = 3.5, p < 0.001) or CFS alone (0.82, 95% CI = 0.78-0.86, DeLong: Z = 4.4, p < 0.001). Bootstrapped estimates of FaP-ED AUROC, calibration slope, and intercept were 0.86, 0.95, and -0.09, respectively, suggesting internal validity. A decision-threshold of CFS 5 and NEWS 3 was proposed based on qualitative comparison of positive Likelihood Ratio at all relevant FaP-ED cutoffs. CONCLUSION: Combining aggregated vital signs and frailty accurately predicted 30-day mortality at ED presentation and illustrated an important clinical interaction between frailty and illness severity. Pending external validation, the Fap-ED operationalizes the concept of such "geriatric urgency" for the ED setting.

Validation of the Clinical Frailty Scale for Prediction of Thirty-Day Mortality in the Emergency Department.

STUDY OBJECTIVE: We validate the Clinical Frailty Scale by examining its independent predictive validity for 30-day mortality, ICU admission, and hospitalization and by determining its reliability. We also determine frailty prevalence in our emergency department (ED) as measured with the Clinical Frailty Scale. METHODS: This was a prospective observational study including consecutive ED patients aged 65 years or older, from a single tertiary care center during a 9-week period. To examine predictive validity, association with mortality was investigated through a Cox proportional hazards regression; hospitalization and ICU transfer were investigated through multivariable logistic regression. We assessed reliability by calculating Cohen's weighted κ for agreement of experts who independently assigned Clinical Frailty Scale levels, compared with trained study assistants. Frailty was defined as a Clinical Frailty Scale score of 5 and higher. RESULTS: A total of 2,393 patients were analyzed in this study, of whom 128 died. Higher frailty levels were associated with higher hazards for death independent of age, sex, and condition (medical versus surgical). The area under the curve for 30-day mortality prediction was 0.81 (95% confidence interval [CI] 0.77 to 0.85), for hospitalization 0.72 (95% CI 0.70 to 0.74), and for ICU admission 0.69 (95% CI 0.66 to 0.73). Interrater reliability between the reference standard and the study team was good (weighted Cohen's κ was 0.74; 95% CI 0.64 to 0.85). Frailty prevalence was 36.8% (n=880). CONCLUSION: The Clinical Frailty Scale appears to be a valid and reliable instrument to identify frailty in the ED. It might provide ED clinicians with useful information for decisionmaking in regard to triage, disposition, and treatment.