Abstract
BACKGROUND: The "surprise question" (SQ) ("Would you be surprised if this patient died in the next 12 months?") is the most frequently used screening tool in emergency departments (EDs) to identify patients with poor prognosis and potential unmet palliative needs. OBJECTIVE: To test and compare the accuracy of the SQ between emergency nurses (ENs) and emergency physicians (EPs) in predicting long-term mortality among older patients (OP) in the ED. DESIGN AND SETTING/SUBJECTS: A prospective cohort study of OPs (≥75 years) conducted in two Belgian EDs. EPs and ENs answered the SQ for the patients they cared for. Positive SQ (SQ+) was defined as a "no" answer. One-year mortality was assessed by phone call. RESULTS: EPs and ENs both answered the SQ for 291 OPs (mean age 83.2 ± 5.4, males 42.6%). The SQ was positive in 43% and 40.6%, respectively. Predictive values were similar in both groups: sensitivity, specificity, c-statistics, negative predictive value, and positive predictive value were 0.79 (0.66-0.88), 0.68 (0.62-0.76), 0.69 (0.63-0.75), 0.92 (0.86-0.96), and 0.4 (0.31-0.50), respectively, for EPs and 0.71 (0.57-0.82), 0.69 (0.62-0.75), 0.69 (0.63-0.75), 0.89 (0.83-0.93), and 0.41 (0.31-0.51), respectively, for ENs. SQ + was associated with a higher mortality risk in both group (EPs hazard ratio: 3.2 [1.6-6.7], p = 0.002; ENs hazard ratio: 2.5 [1.3-4.8], p = 0.006). The survival probability was lower when both EPs and ENs agreed on the SQ+ (p < 0.001). CONCLUSION: The SQ is a simple tool to identify older ED patients at high mortality risk. Concordant responses from EPs and ENs are more predictive than either alone.