Pediatric ambulance utilization in a large American city: a systems analysis approach.
BACKGROUND: Research on utilization of ambulances by pediatric patients lacks an objective, reproducible tool for the evaluation of patterns of ambulance use by both the providers and the users of this resource. OBJECTIVES: 1) To develop an objective, diagnosis-based measure of appropriateness of ambulance utilization. 2) To use the measure to evaluate whether Municipal Ambulance Service dispatchers assign ambulances appropriately, and whether parents/caretakers request ambulances appropriately. STUDY DESIGN: 1) Development of the pediatric ambulance need evaluation (PANE) tool: The consensus of an expert panel was used to assign patients arriving by ambulance to three levels of prehospital transport need based upon their ultimate hospital discharge diagnoses, and were as follows: required advanced life support ambulance (ALS); required basic life support ambulance (BLS); required a less acute mode of transport (LAT). 2) Assessment of appropriateness of ambulance assignments by EMS call-receiving operators (CRO) and of ambulance requests by parents/caretakers: Comparison of actual type of ambulance assigned and of need for ambulance, using the PANE tool and hospital admission rates as gold standards. DATA COLLECTION: Level of prehospital transport provided (ALS vs BLS), ultimate ED diagnosis, and ED disposition (admission vs discharge) was collected for each patient from information abstracted from the prehospital and ED records. SETTING: Bellevue Hospital Center and Harlem Hospital Center, two level I trauma centers in New York City, both with Pediatric Emergency Departments staffed 24 hours a day by attending physicians and residents. PATIENT SELECTION: Consecutive sample of 2633 patients, birth to 18 years of age, who arrived to either hospital by ambulance as primary transports from the field over a one-year period. RESULTS: 1) Development of PANE tool: At Bellevue Hospital, 7% of ED visits arrived by ambulance; at Harlem Hospital, 5% arrived by ambulance. Using these ambulance arrivals, 215 diagnoses were identified for inclusion in the PANE tool. An expert panel categorized each diagnosis as requiring ALS, BLS, or LAT, with a high level of interobserver agreement (weighted kappa = 0.793). As a measure of external validity of the PANE, admission rates were highest in the ALS group, next highest in the BLS group, and lowest in the LAT group (chi2 for trend, P < 0.05). 2) Assessment of ambulance assignments and requests: According to the PANE tool, the sensitivity of dispatcher assignment of ALS ambulances was 72 %. Therefore, 28 % of patients who required an ALS ambulance received BLS care. 50% of patients assigned to an ALS ambulance did not require that level of care, and 1/3 of these were categorized by the PANE as not requiring an ambulance at all. CONCLUSIONS: The PANE tool compared favorably to admission rates as a measure of the severity of illness of patients arriving by ambulance. Applying the PANE tool, we conclude that the majority of requests for ambulances are appropriate, and that the majority of the time dispatchers were able to dispatch the appropriate level of care. However, there is room for significant improvement in utilization of ambulances, and tools like the PANE will be useful in achieving this goal.