|
Activity Number:
|
103
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, July 30, 2007 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #308983 |
|
Title:
|
CUSUMs Based on the Cox Model To Monitor Outcomes at Multicenter Studies
|
|
Author(s):
|
Pinaki Biswas*+ and John D. Kalbfleisch
|
|
Companies:
|
University of Michigan and University of Michigan
|
|
Address:
|
Dept of Biostatistics, , Ann Arbor, MI, 48109,
|
|
Keywords:
|
CUSUM ; Wald's SPRT ; Cox regression ; Poisson process ; Wald's Identity
|
|
Abstract:
|
In organ transplant it is important for a particular transplant center to assess the efficacy of transplants performed over time and provide a signal if there is evidence of poor performance. The CUSUM has been suggested as a sequential monitoring scheme in Page (1954) and has also been implemented in Axelrod et al. (2006) and Steiner et al. (2001). In the above articles, the unit of analysis is a discrete outcome associated with the transplant. In this paper, we utilize the CUSUM statistic based on a Cox regression model prospectively and update the event status continuously in time, also accounting for patient-mix covariates. Simulations demonstrate that this provides an earlier detection of a deteriorating situation than the traditional discrete time methods. Theoretical approximations to the average run length have also been derived for both discrete and continuous cases.
|
