
Evaluating Statistical Methods for Syndromic Surveillance
Published in: Statistical methods in counterterrorism: game theory, modeling, syndromic surveillance, and biometric authentication / by Alyson G. Wilson, Gregory D. Wilson, and David H. Olwell (New York : Springer, 2006), Chapter 1, p. 1-32
Posted on RAND.org on January 01, 2006
The goals of this chapter are (1) to introduce the statistical issues in syndromic surveillance, (2) to describe and illustrate approaches to evaluating syndromic surveillance systems and characterizing their performance, and (3) to evaluate the performance of a couple of specific algorithms through both abstract simulations and simulations based on actual data. Section 1 of this chapter introduces and discusses the statistical concepts and issues in syndromic surveillance, illustrating them with data from an ER surveillance system from the District of Columbia. Section 2 presents methods from the statistical process control (SPC) literature, including variants on existing multivariate detection algorithms tailored to the syndromic surveillance problem, and compares and contrasts the performance of univariate and multivariate techniques via some abstract simulations. Section 3 then compares Evaluating Statistical Methods for Syndromic Surveillance 143 the new multivariate detection algorithms with commonly used approaches and illustrates the simulation approach to evaluation using simulations based on actual data from seven Washington, DC, hospital ERs. We conclude with a discussion about the implications for public health practice.
This report is part of the RAND Corporation External publication series. Many RAND studies are published in peer-reviewed scholarly journals, as chapters in commercial books, or as documents published by other organizations.
The RAND Corporation is a nonprofit institution that helps improve policy and decisionmaking through research and analysis. RAND's publications do not necessarily reflect the opinions of its research clients and sponsors.