The 'surprise' question in advanced cancer patients: A prospective study among general practitioners

Using the 'surprise' question 'Would you be surprised if this patient died in the next year?' may improve physicians' prognostic accuracy and identify people appropriate for palliative care.

Determine the prognostic accuracy of general practitioners asking the 'surprise' question about their patients with advanced (stage IV) cancer.

Prospective cohort study.

Between December 2011 and February 2012, 42 of 50 randomly selected general practitioners (Bologna area, Italy) prospectively classified 231 patients diagnosed with advanced cancer according to the 'surprise' question and supplied the status of each patient 1 year later.

Of the 231 patients, general practitioners responded 'No' to the 'surprise' question for 126 (54.5%) and 'Yes' for 105 (45.5%). After 12 months, 104 (45.0%) patients had died; 87 (83.7%) were in the 'No' group. The sensitivity of the 'surprise' question was 69.3%; the specificity was 83.6%. Positive predictive value was 83.8%; negative predictive value was 69.0%. The answer to the 'surprise' question was significantly correlated with survival at 1 year. Patients in the 'No' group had an odds ratio of 11.55 (95% confidence interval: 5.83-23.28) and a hazard ratio of 6.99 (95% confidence interval: 3.75-13.03) of being dead in the next year compared to patients in the 'Yes' group (p = 0.000 for both odds ratio and hazard ratio).

When general practitioners used the 'surprise' question for their patients with advanced cancer, the accuracy of survival prognosis was very high. This has clinical potential as a method to identify patients who might benefit from palliative care.