Objectives: To implement a new five-level emergency department (ED) triage
algorithm, the Emergency Severity Index (ESI), into nursing practice, and v
alidate the instrument with a population-based cohort using hospitalization
and ED length of stay as outcome measures. Methods: The five-level ESI alg
orithm was introduced to triage nurses at two university hospital EDs, and
implemented into practice with reinforcement and change management strategi
es. Interrater reliability was assessed by a posttest and by a series of in
dependent paired patient triage assignments, and a staff survey was perform
ed. A cohort validation study of all adult patients registered during a one
-month period immediately following implementation was performed. Results:
Eight thousand two hundred fifty-one ED patients were studied. Weighted kap
pa for reproducibility of triage assignments was 0.80 for the posttest (n =
62 nurses), and 0.73 for patient triages (n = 219). Hospitalization was 28
% overall and was strongly associated with triage level, decreasing from 58
/63 (92%) of patients in triage category 1, to 12/739 (2%) in triage catego
ry 5. Median lengths of stay were two hours shorter at either triage extrem
e (high and low acuity) than in intermediate categories. Outcomes followed
a-priori predictions. Staff nurses rated the new program easier to use, and
more useful as a triage instrument than previous three-level triage. They
provided feedback, which resulted in significant revisions to the algorithm
and educational materials. Conclusions: Triage nurses at these two hospita
ls successfully implemented the ESI algorithm and provided useful feedback
for further refinement of the instrument. Emergency Severity Index triage r
eproducibly stratifies patients into five groups with distinct clinical out
comes.