|
Activity Number:
|
352
|
|
Type:
|
Invited
|
|
Date/Time:
|
Wednesday, August 6, 2003 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Health Policy Statistics
|
| Abstract - #302034 |
|
Title:
|
Evidence-Based Selective Referral for Infants with Very Low Birth Weights
|
|
Author(s):
|
Douglas Staiger*+ and Jeannette Rogowski and Jeffrey Horbar and Michael Kenny and Joseph H. Carpenter and Jeffrey Geppert
|
|
Companies:
|
Dartmouth College and RAND Corporation and Vermont Oxford Network and University of Vermont and Vermont Oxford Network and National Bureau of Economic Research
|
|
Address:
|
Department of Economics, Hanover, NH, 03755,
|
|
Keywords:
|
hierarchical models ; provider profiling ; quality of health care ; low birth weight infants ; hospital performance ; Newborn Intensive Care Units
|
|
Abstract:
|
Evidence-based selective referral strategies that rely on hospital characteristics such as volume or teaching status are being used by an increasing number of insurers. We investigate the extent to which such hospital characteristics can prospectively identify providers with significantly better or worse patient outcomes, and whether hierarchical estimates of past differences across hospitals in patient outcomes are more accurate for this purpose. We use data on 110,164 very low birth weight infants (500g-1500g) cared for by 340 neonatal intensive care units (NICUs) between 1995 and 2000. Both patient volume and NICU level are significant predictors of mortality, but together they explain only 16% of the systematic variation across NICUs and prospectively cannot identify hospitals with significantly higher or lower mortality rates. In contrast, using rankings based on hierarchical estimates of hospital mortality rates from 1995-1998, we find a more than two-fold difference in mortality rates between the top and bottom 20% of hospitals in the subsequent two years (1999-2000).
|
