Background: Prognostication of survival in advanced cancer patients has been challenging and contributes to poor illness understanding. Disagreements among providers about prognosis occur and is a less studied phenomenon. We asked the Surprise Question (SQ), “Would I be surprised if this patient died within the next 1 year, 6 months, and 1 month?” to Palliative Care (PC) providers, Medical Oncology providers, Hospital Medicine providers, and bedside nurses for patients under their care on our inpatient Medical Oncology Service.
Methods: This was an observational survey study with chart review follow up at 1 year. We collected the Hospital Medicine provider, PC provider, Oncologist, and bedside nurses’ response to the SQ during multi-disciplinary rounds. If the answer to the 1-year surprise question was “No” then the providers were asked the 6-month surprise question (“Would you be surprised if this patient died in the next 6 months?”) and 1-month surprise question (“Would you be surprised if this patient died in the next 1 month?”). PC providers and Oncology providers only answered the SQ for patients they were seeing in the hospital.
Results: The SQ was asked for 324 admissions between March – May 2016 to a total of 20 Hospital Medicine, 6 PC, and 7 oncology providers. 51.9% of the sample was male and 48.3% female with an average age of 57.8 years (±14.4). Of the total sample, 44.8% of patients died within 1 year, 36.1% died within 6 months, and 13.1% of patients died within 1 month of admission. Average time to death was 131.9 days (±141.1).
Both Hospital Medicine and Oncologists were more accurate in their response to the 6 month SQ ( = 0.38 and 0.31 respectively) when compared to their 1 year ( = 0.27 and 0.25 respectively) and 1 month ( = 0.24 and 0.26 respectively) SQ responses. PC providers answered “No” to the SQ for the 1 year and 6-month time frame. For the 1 month time frame, the PC providers answered “Yes” for 11 admissions and of those 3 of them died within 1 month. Bedside nurses were the least accurate when compared to other provider groups for all time frames of the SQ ( = 0.18 (1 year) = 0.19 (6 month) and = 0.08 (1 month).
Conclusions: We aimed to better understand provider estimations of prognosis for cancer patients admitted to the hospital. Hospital Medicine and Oncology providers were equally poor in their estimations of prognosis based on kappa statistics though were most accurate for the 6-month SQ. Bedside nurses were least accurate in their estimations of prognosis. It is unclear why nursing estimations about prognosis are more optimistic than other providers. One consideration is that nurses spend more time involved in direct patient care activities and may form a more personal relationship that might influence their response. Further study is needed to better identify what factors providers consider when they answer the SQ and similar prognostic prompts.