Background
Measuring respiratory rate (RR) is one of the most basic clinical observations performed when accessing acutely ill patients. RR is included in most triage systems and risk stratifications tools, but unlike the other vital signs, RR is typically obtained by a manual count by the nursing staff. Considering how often RR is used in clinical practice and contained in triage systems, it is remarkable how few studies on inter-observer/rater agreement have actually been performed. Furthermore, the existing studies are all made with few observers and many patients, and none of them have been performed in an actual acute setting of an emergency department (ED). We therefore aimed to determine the interobserver variability of RR counts, using a larger number of observers on few same patients in the setting of an ED.