Yinong Young-Xu, ScD, MS¹, Ezekiel Emanuel, MD, PhD², James Flory, BA², Ipek Gurol, PhD³, Norman Levinsky, MD⁴, and Arlene Ash, PhD⁵. (1) General Internal Medicine, Boston University School of Medicine, 720 Harrison Avenue, DOB Suite 1108, Boston, MA 02118, 603 787 6460, yyoungxu@hsph.harvard.edu, (2) Department of Clinical Bioethics, Warren G. Magnuson Clinical Center, National Institutes of Health, 10 Center Drive, Building 10, Room 1C118, Bethesda, MD 20892, (3) Department of Economics, Bogazici University, Washburn Hall Rm. 206, DOB Suite 1108, Bebek Istanbul, 34342, Turkey, (4) Department of Medicine, Boston University School of Medicine, 715 Albany Street, Boston, MA 02118, (5) Health Care Research Unit, Boston University School of Medicine, 720 Harrison Ave., Suite 1108, Boston, MA 02118

Background: Geographic variations in the percent of deaths that occur in the hospital (i.e., the “hospital death rate”) have been linked to differences in bed availability. However, little is known about the dynamic effect of changes in hospital bed supply on hospital death rates. Methods: We used Center for Disease Control (CDC) death certificate data to identify place and cause of death, age, gender and race for 92% of all U.S. decedents from 1989 to 1997. We also recorded county-level changes in the supply of hospital beds and physicians from the Area Resource File for the same period. For confidentiality, the CDC codes an aggregated “county” within each state, pooling all decedents from counties with < 100,000 residents. Our data contain 392 identified and 50 aggregated counties. Results: Nationally, hospital-bed supply and hospital death rates decrease in tandem from 1989 to 1997. However, county-specific data varied substantially. The overall correlation between beds and hospital deaths existed in only 25% of counties. For 75% of the counties, little relationship, and sometimes even negative correlation, was found between change in hospital-bed supply and change in hospital death rate. We stratified into quartiles by change in hospital-bed supply over that period. Counties in the 1st quartile had no reduction in bed supply, while the 2nd through 4th quartiles saw mean declines of 0.2, 0.8, and 1.7 beds per 1,000 population. Percentage changes in hospital death rates in these quartiles were –6.6%, –7.2%, –8.1%, and –7.2%, respectively (test for trend, p = 0.45). Counties stratified into quartiles by initial bed availability averaged 1.8, 3.1, 4.3, and 6.7 beds per 1000 population in 1989. Percentage changes in hospital death rates in these quartiles were –7.1%, –8.1%, –7.3%, and –6.8% (test for trend, p=0.67). The longitudinal data in most counties showed little relationship between changes in the hospital-bed supply and the hospital death rate, and no consistent pattern emerged when simultaneously examining initial bed supply and changes in that supply as predictors of declines in the hospital death rate. Conclusion: Despite a parallel decline in hospital beds and hospital death rates nationally between 1989 and 1997, at the county level, neither the initial hospital bed supply, nor changes in that supply were associated with declines in the hospital death rate.

Learning Objectives:

• At the conclusion of the session, the participant (learner) in this session will be able to appreciate the complex relationship between hospital bed supply and hospital death rate.

Keywords: Death, Health Care Restructuring

Presenting author's disclosure statement:
I do not have any significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.