Utility of the "surprise" question to identify dialysis patients with high mortality

Abstract

Background and objectives: Dialysis patients are increasingly characterized by older age, multiple comorbidities, and shortened life expectancy. This study investigated whether the "surprise" question, "Would I be surprised if this patient died in the next year?" identifies patients who are at high risk for early mortality.

Design, setting, participants, & measurements: This prospective cohort study of 147 patients in three hemodialysis dialysis units classified patients into "yes" and "no" groups on the basis of the "surprise" question response and tracked patient status (alive or dead) at 12 mo. Demographics, Charlson Comorbidity Index score, and Karnofsky Performance Status score were measured.

Results: Initially, 34 (23%) patients were classified in the "no" group. Compared with the 113 patients in the "yes" group, the patients in the "no" group were older (72.5 +/- 12.8 versus 64.5 +/- 14.9), had a higher comorbidity score (7.1 +/- 2.3 versus 5.8 +/- 2.1), and had a lower performance status score (69.7 +/- 17.1 versus 81.6 +/- 15.8). At 12 mo, 22 (15%) patients had died; the mortality rate for the "no" group was 29.4% and for the "yes" group was 10.6%. The odds of dying within 1 yr for the patients in the "no" group were 3.5 times higher than for patients in the "yes" group, (odds ratio 3.507, 95% CI 1.356 to 9.067, P = 0.01).

Conclusions: The "surprise" question is effective in identifying sicker dialysis patients who have a high risk for early mortality and should receive priority for palliative care interventions.

Figure 1.

Survival curves for “surprise” question response and comorbidity score in days alive at 12 mo. Data are means ± SE. (A) Curves of “yes” and “no” response groups to the “surprise” question, “Would I be surprised if this patient died in the next year?” (B) Curves of the lower (<8) and higher (≥8) Charlson Comorbidity Index (CCI) score groups.