Study objective: To determine whether implementation of a set of standardized pediatric telephone triage protocols results in consistent triage dispositions when applied by different operators.
Methods: A descriptive study with interrater comparisons was performed. Telephone interviews simulated the setting of a triage station in a university hospital-based pediatric emergency department. A mock parent presented 15 standardized respiratory cases in random order to 12 pediatric ED nurses. Nurses assigned patients' complaints to severity categories using 9 respiratory complaint protocols extracted from a commercially available pediatric telephone triage tool. Protocol selection and severity endpoints were recorded. Interobserver agreement among nurses was analyzed by the kappa statistic. Comparisons of operator characteristics and triage results were carried out by ANOVA.
Results: Interrater agreement in triage disposition among nurses was poor (kappa,.11; 95% confidence interval,.02 to.20). Protocol selection varied; the group used a mean of 3 different disposition-generating protocols per case. Disposition also varied, with up to 4 different severity endpoints per protocol in a given case. A post-hoc comparison of the mean disposition severity between nurses did not reach significance at an adjusted level (P =.04). Fifty-eight percent of the nurses felt confined by the protocols, and 42% admitted to at least 1 intentional deviation from them.
Conclusion: It may not simply be assumed that the use of protocols will standardize care. This is particularly important in the case of triage, with current trends toward medical decisionmaking by less skilled providers with diminishing patient contact. Although triage protocols may be useful to guide clinical thinking, their consistency must be validated before they may be safely disseminated for general use.