Premium
An improved cumulative sum‐based procedure for prospective disease surveillance for count data in multiple regions
Author(s) -
Dassanayake Sesha,
French Joshua P.
Publication year - 2016
Publication title -
statistics in medicine
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.996
H-Index - 183
eISSN - 1097-0258
pISSN - 0277-6715
DOI - 10.1002/sim.6887
Subject(s) - count data , computer science , scan statistic , context (archaeology) , statistic , statistics , poisson distribution , outbreak , data mining , mathematics , medicine , geography , pathology , archaeology
We present an improved procedure for detecting outbreaks in multiple spatial regions using count data. We combine well‐known methods for disease surveillance with recent developments from other areas to provide a more powerful procedure that is still relatively simple and fast to implement. Disease counts from neighboring regions are aggregated to compute a Poisson cumulative sum statistic for each region of interest. Instead of controlling the average run length criterion in the monitoring process, we instead utilize the FDR, which is more appropriate in a public health context. Additionally, p ‐values are used to make decisions instead of traditional critical values. The use of the FDR and p ‐values in testing allows us to utilize recently developed multiple testing methodologies, greatly increasing the power of this procedure. This is verified using a simulation experiment. The simplicity and rapid detection ability of this procedure make it useful in disease surveillance settings. The procedure is successfully applied in detecting the 2011 Salmonella Newport outbreak in 16 German federal states. Copyright © 2016 John Wiley & Sons, Ltd.