Prognostic value of the surprise question in HeartMate 3 patients
I Santos, S Azevedo, A M Garcia, M Presume, A R Bello, C Sampaio, S Maltes, B Rocha, C Brizido, C Strong, M Marques, C AguiarAbstract
Background
Prognostication in patients receiving left ventricular assist devices (LVAD) is challenging. The Surprise Question (SQ) – "Would you be surprised if this patient died in the next 12 months?" – has been proposed as a simple, low-burden tool to help identify patients at increased mortality risk, including those with advanced heart failure (AHF). However, its performance in the LVAD population remains unclear.
Purpose
To evaluate the ability of the SQ to predict 12-month mortality in patients supported with HeartMate 3™ (HM3).
Methods
This retrospective observational cohort study included all HM3 patients under follow-up at a single center in 2024. In November 2024, the assistant AHF physician answered the SQ for each patient. All-cause mortality was assessed after 12 months. Prognostic performance of the SQ was assessed using a 2×2 contingency table comparing physician SQ responses ("Not surprised" vs "Surprised") with observed outcomes. Accuracy measures for the SQ were calculated and correlations with mortality were analyzed.
Results
Eleven patients were included (mean age of 60.7 ± 12.0 years); most with AHF due to ischemic heart disease (63.6%) and on a bridge-to-transplantation strategy (54.5%). The SQ was answered as "Yes, I would be surprised" in ten cases (90.9%), and as "No, I would not be surprised" in one case (9.1%). At 12-month follow-up, ten patients were alive. All cause mortality was 9.1%, with one patient dying at 11 months due to intestinal ischemia, corresponding to the only case considered not surprising to occur. The SQ correctly identified all outcomes: sensitivity 100%, specificity 100%, positive predictive value 100%, negative predictive value 100%, and overall accuracy 100%. Chi-Square test showed a strong association between the SQ prognosis prediction and 12-month mortality (Pearson χ² = 11.0, p < 0.001); however, due to the small sample size, Fisher’s exact test did not reach statistical significance (p = 0.091).
Conclusion
In this cohort, the SQ demonstrated perfect concordance with observed outcomes, allowing correctly identifying the only patient who died within this 12-month period. Although limited by the small sample size, these results suggest that the SQ may be an accurate and practical prognostic tool, also in the HM3-recipient population, and may help trigger timely advance care planning. Larger studies are needed to validate its accuracy and clinical role in this context.