A Comparison of Methods for the Surveillance of Congenital Malformations

• Activity Number: 366
• Type: Contributed
• Date/Time: Wednesday, August 10, 2005 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Landon Sego*+ and William H. Woodall
• Companies: Virginia Polytechnic Institute and State University and Virginia Polytechnic Institute and State University
• Keywords: medical surveillance ; monitoring ; CUSUM ; Sets technique ; proportion

Abstract

The discussion of how best to monitor the incidence of congenital malformations began in earnest in the late 1960s. Since then, a number of methods have been proposed, including the Sets technique, the CUSCORE method, an extension of the Sets technique proposed by Sitter et al. (1990), and the CUSUM chart based on a Poisson variate. These methods have been compared to each other, with the Poisson CUSUM chart typically having the best performance. Often the underlying assumption of these methods is that the incidence of malformations follows a Poisson process. However, if one assumes all infants in a given hospital or region can be examined, diagnosed, and recorded sequentially, the assumption of a Bernoulli response is appropriate. Drawing on results from the statistical quality control literature, we compare the performance of the other methods to the Bernoulli CUSUM chart under a variety of circumstances. While these methods have been discussed in the context of the surveillance of congenital malformations, they are more broadly applicable to any situation where a proportion is monitored continuously.