Is it possible to safely triage callers to EMS dispatch centers to alternative resources?

Objective. To develop guidelines allowing emergency medical services (EMS) dispatchers to safely match callers to an EMS response or, alternatively, to a nontraditional resource. Methods. This was a prospective cohort study of callers to an urban EMS dispatch center and an associated review of EMS patient care forms and emergency department (ED) patient care records. The following five "nature codes" (patient chief complaints) were included: back pain, fall, bleeding or laceration, sick, and trauma. Callers included in the study had been assigned the lowest severity level (Alpha), according to existing dispatch criteria. An a priori list of EMS and ED "important findings," indicating need for an EMS response, was used as the outcome variable. Classification and regression tree (CART) analysis was used to develop a decision rule to further identify a low-risk subgroup of patients who could potentially be served by alternative resources. Results. From November 1, 1998, to May 31, 1999, 656 subjects were entered into the study, including 263 males (40%) and 389 females (59%). The mean age was 51 years (range, 0-101 years). One hundred twenty-five (19%) callers had an important EMS finding, including the administration of comfort medications, morphine, benzodiazepines, and droperidol. Forty-six subjects (7%) had an important ED finding. When EMS and ED findings were combined, 158 subjects (24%) had an "important finding." Using CART analysis, having an age <12 years predicted a subset of patients who did not have an important finding suggesting the need for an EMS response. Using cross-validation, this decision rule had a 99% sensitivity, 13% specificity, and 98% negative predictive value. Conclusion. The authors were able to use a demographic variable (age) to predict a population of callers to a 911 dispatch center triaged to the lowest acuity category, who have a very low risk of having an EMS or ED important finding. The decision rule developed here is preliminary, requiring further validation.