|
Activity Number:
|
242
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 5, 2008 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Health Policy Statistics
|
| Abstract - #301480 |
|
Title:
|
Statistics in Evaluating Trauma Center (TC) Performance
|
|
Author(s):
|
Wei Xiong*+ and Khumar Huseynova and Sandra Goble and Avery Nathens
|
|
Companies:
|
University of Toronto and University of Toronto and NTDB, American College of Surgeons and University of Toronto
|
|
Address:
|
Queen Wing, 3N-082, Toronto, ON, M5B 1W8, Canada
|
|
Keywords:
|
Statistics (Z, W, O/E) ; Confidence Interval ; Zero Observed Deaths ; Performance Evaluation ; Kappa
|
|
Abstract:
|
For the purposes of external benchmarking of TC care, it is important to identify high/low performing TCs by comparing the ratio of observed versus expected deaths (O/E). A TC with CI (O/E) above/below the value of one is a low/high performing center. The most frequently used methodology in literature uses a Z (TRISS) or W (excess number of survivors /100 patients) statistic. This study compares an O/E statistic with CI based on Chi-square and Poisson distribution. Unlike the usual CIs for O/E in the literature, the proposed method correctly produces reasonable CI for zero observed events. To validate this methodology, we evaluated the concordance of this O/E statistic in a study of 97 TCs across the US. The concordance between O/E and Z (or W) is excellent with Kappa coefficient from 0.81 to 1.00. This study provides a methodology for benchmarking TC with zero observed events.
|
