Accuracy of physician prognosis in heart failure and lung cancer: Comparison between physician estimates and model predicted survival

Abstract

Background: Anticipating adverse outcomes guides decisions but can be particularly challenging in heart failure. Aim: We sought to assess the accuracy and comfort of physicians in predicting prognosis in heart failure. Design: Cross-sectional survey Participants/setting: Faculty and trainees in internal medicine, cardiology, and oncology estimated survival for three standardized patients: (1) 59-year-old patient with stage IV lung cancer; (2) 79-year-old woman with New York Heart Association class 4 heart failure symptoms and preserved ejection fraction; and (3) 40-year-old man with New York Heart Association class 3 heart failure symptoms and reduced ejection fraction of 20%. Survival predictions were derived from surveillance, epidemiology, and end results-Medicare database and the Seattle Heart Failure Model. Accuracy was defined as <2-fold difference between the clinician and model estimate. Results: Totally, 79% (338/427) of participants responded. Physicians were more accurate in survival estimates for lung cancer than heart failure (74% vs 48%, respectively; p < 0.001). Cardiologists were more accurate in predicting survival in heart failure symptoms and reduced ejection fraction compared to generalists (67% vs 45%; p = 0.005) and oncologists (39%; p = 0.041) but no different at predicting heart failure symptoms and preserved ejection fraction. Cardiologists predicted longer survival in heart failure compared to others ( p < 0.05). Physicians felt more uncomfortable discussing palliative care with heart failure patients compared to lung cancer. Conclusions: Less than half of physicians accurately estimate survival in heart failure. Cardiologists were more accurate than other specialties for heart failure symptoms and reduced ejection fraction but no different for heart failure symptoms and preserved ejection fraction.