Emergency triage to intensive care: can we use prognosis and patient preferences?

To identify predictors of 6-month mortality known before emergent admission to intensive care (IC) and to describe obstacles to the use of patient preferences in emergency triage decisions.

Historical cohort.

A 600-bed university hospital.

263 consecutive patients triaged in the emergency room to receive intensive care.

Medical records were abstracted for age, performance status, and chronic disease severity as predictors of 6-month survival. Acute Physiology Score (APS) in the emergency room was used as a measure of acute illness severity. Deaths during the 6 months following IC admission were determined from record review and death certificate data. Obstacles to communication of patient treatment preferences at the time of triage were described. Six-month mortality was 19 percent, and increased with increasing APS, age > or = 80 (43%), poor performance status (56%), and severe chronic disease (33%) (P < or = 0.01). In multivariate analysis, APS, age > or = 80 and performance status were independent predictors of 6-month mortality. Only APS predicted mortality in hospital. The most common obstacles to use of patient preferences in triage decisions were absence of documented advance directives (95%) and the brief duration of acute illness (72%). Mental status changes were very common in the emergency room for nonsurvivors (61%), but chronic cognitive impairment was rare (3%).

Patients with poor performance status or very advanced age have increased mortality within 6 months of emergent triage to IC. Mental status changes, absence of advance directives, and time constraints are common barriers to communication of patient preferences at the time of triage. Primary care physicians need to elicit and record patients' preferences before the time of emergent decisions about IC.