Use of CUSUM Control Charts When Population Size Changes Over Time

• Activity Number: 475
• Type: Contributed
• Date/Time: Wednesday, August 4, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Section on Quality and Productivity
• Author(s): J. Brooke Marshall*+ and Dan J. Spitzner
• Companies: Merck Research Laboratories and University of Virginia
• Keywords: public health surveillance ; prospective monitoring ; statistical process control ; control charts ; CUSUM ; population adjustment

Abstract

Cumulative sum (CUSUM) control charts are advantageous in many monitoring applications because they signal more quickly than Shewhart-type control charts for small parameter shifts. In public health surveillance applications, the implementation of control charts is challenging because of changes in population size over time. When using a Shewhart-type chart, the control limits can be adjusted in each time interval to account for changes in population because estimates plotted for the time intervals are assumed to be independent. Since CUSUM charts incorporate information from past time intervals, changing the control limit for each interval is not appropriate. We will discuss the implementation of a population-adjusted CUSUM chart, where the chart is reconstructed each time a new point is plotted so that the control limit and points are on the same scale with respect to population size.