246Use of The ‘Surprise Question’ in Predicting Adverse Outcomes Among Frail Older Patients after Hospital Admission

Abstract

Background: The 'surprise question' (SQ) (Would you be surprised if this patient died in the next 12 months?) is considered useful in predicting mortality and the need for palliative care, though recent meta-analysis suggests it has only modest accuracy in predicting death. It is unknown if it has predictive validity for adverse outcomes including prolonged length of stay (LOS) and 30-day readmission rates or correlates with frailty on admission. Methods: Consecutive patients aged >70 attending an Irish university hospital Emergency Department (ED) had the SQ scored by a trained clinician after completing a detailed comprehensive geriatric assessment. Measures of frailty including the Clinical Frailty Scale (CFS), Risk Instrument for Screening in the Community (RISC) mortality sub-score, caregiver burden score (CBS) and Euroqol EQ-5D quality of life (QOL) measure were also recorded. Results: The SQ was available for 191 patients, median age 79 +/-10, 55% female. In all, 56/191 (29%) screened SQ positive (i.e. not surprised). There was no statistically significant difference in age (80 versus 79 years, p = 0.06) or gender (55% versus 46%, p = 0.07) between SQ positive and negative patients. SQ positive patients were frailer, median CFS 6/9 versus 4/9 (p < 0.001) and had lower EQ-5D scores (p < 0.001). Correlations between the SQ and CBS (r = -0.35) and RISC (r = -0.68) were moderate-strong but poor for measures of frailty and the EQ-5D (all < 0.1). While SQ positive patients had significantly reduced survival times (LogRank X2=14.5, p < 0.001), accuracy in predicting one-year mortality, admission, prolonged LOS and readmission were also poor (area under curve <0.65). Conclusion: A large proportion aged >70 attending ED screened SQ positive. These were significantly frailer with lower QOL. The SQ correlated with caregiver burden but not with measures of frailty or QOL. As with mortality, the predictive accuracy of the SQ for healthcare system-important outcomes appears poor.