New evidence for a strategy to identify patients nearing the end of life

The Surprise Question allows clinicians to identify seriously ill patients who would benefit from conversations about their goals and values. The aim is for physicians and patients to have better conversations, more often and earlier, so that care and services can be aligned with patient priorities and preferences. There is currently no generally accepted systematic approach for identifying such patients.

The Surprise Question ─ "Would you be surprised if this patient died within the next year?" ─ was developed in the 1990s. Until now, however, there has been limited evidence of its utility in the cancer care setting.

"In all areas of medicine, patients may become seriously ill. Some of these patients will recover completely. Unfortunately, some won't," said lead study author Judith Vick, a medical student at Johns Hopkins School of Medicine in Baltimore, Maryland. "Our results showed that oncology clinicians' own judgment using the Surprise Question identified about 60% of patients who died within a year."

The ongoing study, led by researchers at Ariadne Labs in Boston, enrolled 76 oncology clinicians, including oncologists, nurse practitioners, and physician assistants at the Dana-Farber Cancer Institute. The clinicians answered the Surprise Question regarding nearly 5,000 patients as the first step in Ariadne Labs' comprehensive, patient-centered Serious Illness Care Program.

When answering the question, the clinicians used their best clinical judgment. They answered, "Yes, I would be surprised," for roughly 85% of patients and, "No, I would not be surprised," for about 15% of the patients.

Ninety-five percent of patients for whom clinicians answered "Yes" to the Surprise Question were indeed alive at one year; 62% of patients for whom the Surprise Question was answered "No" were alive at one year. Of the patients who died in the next year, the Surprise Question failed to identify about 40%.

The Surprise Question can be used to identify many patients who will die in the next year and allow patients and their clinicians to plan for this possibility. More research is needed, however, to understand factors that contribute to non-recognition of so many patients and to develop better predictors of patient death.

The researchers plan to investigate clinician characteristics that might affect how accurately they answer the Surprise Question and to examine the characteristics of patients for whom the answer to the question was inaccurate, i.e., those who died or lived contrary to expectations.