ABSTRACT
Introduction: Alzheimer's disease (AD) dementia is the sixth leading cause of death in the United States. The surprise question (SQ) "Would you be surprised if this patient were to die within the next 12 months?" was used to identify death-risk patients, who could benefit from palliative care. Objective: To examine the prognostic accuracy of the SQ by physicians and caregivers in outpatients with AD dementia. Methods: This is a longitudinal and prospective study involving 101 patients along 1 year, applying the SAS 9.2 software and adopting a .05 P-value to assess the variables that influenced answers to the accuracy of SQ using the chi-square test. Results: 27 patients (26.7%) died during the follow-up. When caregivers answered the SQ, it presented a 51.8% sensitivity (CI 31.9 - 71.3), a 66.7% negative predictive value (20.7 - 63.6), a 56.2% specificity (CI 29.8 - 80.2), and a 40.9% positive predictive value of (CI 43.0 - 85.4) with a 53.4% accuracy (CI 38.5 - 68.4). When physicians answered, the SQ had an 88.8% sensitivity (CI 70.8 - 97.6), a 40% negative predictive value (CI 5.2 - 85.3), a 12.5% specificity (CI 1.5 - 38.3), a 63.1% positive predictive value (CI 45.9 - 78.1) with a 60.4% accuracy (CI 45.8 - 75). Conclusion: SQ remains a good tool with high sensitivity for the identification of patients with advanced AD dementia when presented to the attending physician for planning palliative advanced care with accuracy of 60.4% and caregivers' accuracy of 53.4%.
