|
Activity Number:
|
388
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 9, 2006 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #306837 |
|
Title:
|
On Detecting a Rate Increase Using a Bernoulli-Based Scan Statistic
|
|
Author(s):
|
Michael Joner*+ and William H. Woodall and Marion Reynolds
|
|
Companies:
|
Virginia Polytechnic Institute and State University and Virginia Polytechnic Institute and State University and Virginia Polytechnic Institute and State University
|
|
Address:
|
406A Hutcheson Hall, Blacksburg, VA, 24061-0439,
|
|
Keywords:
|
scan statistic ; Bernoulli ; cusum ; surveillance ; prospective monitoring ; statistical quality control
|
|
Abstract:
|
Scan statistics are often used by public health specialists to detect disease clusters or increased incidence rates. Such events are signaled if there are an unusually large number of events close in time and/or space. Most of the work has been for the retrospective case, in which a single set of historical data are to be analyzed. In the temporal case it has been recommended that the scan statistic be used in a prospective manner to determine whether the underlying incidence rate has increased, preferably as soon as possible after such an increase. We evaluate the average number of times that the scan statistic must be calculated for the Bernoulli-process case in order to generate a signal that the incidence rate has increased. We show that the Bernoulli cumulative sum method can have much better performance than the scan-based methods.
|
