Interrater agreement: a comparison between two emergency department triage scales

Abstract

Objective: The aim was to elucidate if, by strictly applying the Adaptive Process Triage (ADAPT) scale, the interrater agreement increased among the participating registered nurses (RNs) than when triaging according to the older scale, which allowed subjective interpretations of signs and symptoms.

Methods: Nineteen patient scenarios were triaged in 2006 by 45 RNs using the previous triage scale, and in 2008 by 30 RNs using ADAPT.

Results: There was no significant difference (P=0.65) between the two triage scales with regard to level of overall exact agreement (κ value 0.529 vs. 0.472). The same triage level was more often chosen when using the ADAPT system as compared to the earlier triage scale and dispersion across the triage levels was also reduced when using ADAPT. Eight (42%) of the patient scenarios were triaged as both unstable and stable by ADAPT, and 11 (58%) when the older scale was applied. Fourteen (74%) of the scenarios could not be allocated to a defined triage level by ADAPT. Five main reasons for such triage decisions were identified.

Conclusion: Both the triage scales showed moderate overall agreements, while dispersion of triage decisions across several triage levels declined when ADAPT was used. Although the algorithm for acuity allocation by ADAPT seemed well defined, many patient scenarios were triaged as both unstable and stable and thus allocated to various triage levels. If ADAPT is to function as a safe triage tool with low interrater variability, further revision of the triage algorithms is needed.