Over 25 million children annually receive emergency care in the United States. Many types of practitioners provide this care in a variety of settings; thus, there is the potential for wide variations in quality. However, there have been few studies of the quality of the care delivered in EDs or the determinants of ED quality of care. There is not even a widely accepted outcome for quality of care studies or methods for assessing the effects of the performance of the ED health care team on patient health outcomes. Because one of the major goals of ED care is the establishment of a working diagnosis and appropriate disposition to either inaptient or outpatient care settings, we propose that the correct patient disposition rate (hospital admission vs. ED discharge) adjusted for severity of illness and other case-mix variables is a global measure of ED performance. The primary aim of this proposal is the national validation of a pediatric ED severity of illness assessment method and its application to the investigation of ED care factors associated with quality. Institutional quality factors will include patient volume, presence or absence of pediatric emergency medicine specialists, and presence or absence of residents caring for patients. Hospital admission will be used as the primary outcome. Predictor variables will include physiologic, therapeutic, diagnostic, and other risk factors. This predictor model will be developed in a reference set of 16 hospitals with Pediatric Intensive Care Units (PICUs) to insure similar inpatient capabilities and a total sample size of 12,000 patients. Quality will be determined by the prediction error of number of admissions observed versus the number predicted. The construct validity of our method will be demonstrated by an analysis of resource use by a well-defined subset of admitted patients, those that demonstrate clear and unarguable need for hospital care. Secondary aims are establishing the relationship between the ED severity of illness method and the secondary outcomes of PICU admission, hospital mortality, and hospital length of stay. Quality of ED care will be compared by measuring the prediction error. Patient-level analyses will include investigation of socioeconomic status and payor status.
