The Emergency Department as a Gateway for Hospitalization: Evidence from the National Medical Expenditure Survey

288 Brief communication THE EMERGENCY DEPARTMENT AS A GATEWAY FOR HOSPITALIZATION: EVIDENCE FROM THE NATIONAL MEDICAL EXPENDITURE SURVEY African Americans are known to use emergency department (ED) services with greater frequency than whites.1 "3 Much of that difference has been attributed to the greater lack of non-hospital-based usual providers of care for the African American population, this being attributable to poorer health insurance coverage, residence in medically underserved areas, and lower income among African Americans. The factors influencing the increased use of emergency rooms by minorities were examined by Cunningham, Clancy, Cohen, and Wilets1 and by WhiteMeans and colleagues.2,3 The research by Cunningham et al. based on the National Medical Expenditure Survey (NMES) found that African Americans were significantly more likely to use the ED for non-urgent care (odds ratio 1.68), even after controlling for health status, insurance coverage, location, employment, and other demographics. A multivariate analysis of a national sample by White-Means and colleagues2,3 found that ethnicity and age were important predictors of emergency room use. Specific health conditions also determined whether the emergency room was chosen as a health care option. They found that both blacks and whites go to hospital emergency rooms for nonemergency health problems . Age, marital status, and health conditions were significant sociodemographic determinants for blacks' visits on these occasions, while those for whites included sex, age, education, insurance, employment status, region of residence, and health conditions. In later work, White-Means and Thornton4 found that African Americans had greater expenses from nonurgent ED visits than whites or Hispanics. Several studies have shown discrepancies in a variety of health services provided for African Americans compared with whites.5"9 One study showed that in a large city with nine managed care organizations, African Americans were more likely to be denied access to the emergency room for care than whites regardless of the managed care organization.10 The purpose of this paper is to analyze health care services utilization in the ED for African Americans and whites and to note any differences. In this paper, we will analyze ED services for African Americans and whites in terms of Journal of Health Care for the Poor and Underserved · Vol. 13, No. 3 · 2002 Hogan, Bouknight 289 1. the utilization of EDs, 2. the likelihood of being hospitalized following an ED visit, 3. the temporal lag in admission to hospital following an ED visit, and 4. the expense and length of stay for the post-ED hospitalization. Method Data for this study were taken from the NMES.11 Collected during 1987 by the Agency for Health Care Policy and Research, NMES provides information on health insurance coverage, health status, health care utilization and treatment costs, prescription drug use, work, sources of income, disability, and other sociodemographic characteristics of a nationally representative sample of the noninstitutionalized civilian population of the United States. This analysis used data from Public Use Files 9 (Health Status Questionnaire), 13 (Population Characteristics and Insurance Coverage), 14.4 and 14.5.3 (Health Services Utilization: Inpatient Hospital and Emergency Department), and 16 (Summary Data). The sample population to be analyzed was made up of African American and white adults responding to the NMES. We first identified all African American or white adults with ED utilization in File 13. We then extracted all service records for these individuals from the NMES ED services utilization file (14.5.3). We searched the NMES hospital file (14.4) for hospital records for the identified cases with admission within seven days of the ED visit. It was then determined whether the ICD-9 diagnosis codes (up to four) of the hospital admission matched any of the diagnosis codes on the ED record. If there was a match, then a matching indicator variable was created to document this comparison. We also determined the number of days between the ED visit and the hospitalization for all cases with a diagnostic match. To measure the extent of patients' burden of comorbid conditions, we employed the Charlson Index12 as it had been adapted by the Deyo, Cherkin, and Ciol13 for administrative databases. The Charlson Index is based on the presence or absence of 17...