A Method for Detecting Current Temporal Clusters of Toxic Events Through Data Monitoring by Poison Control Centers

Abstract

Background: Poison control centers have become a widely recognized source of data for chemical and other environmental exposures. Concurrently, increased emphasis on early identification of temporal clustering of toxic exposures has stimulated development of sensitive statistical methods to detect clusters at the time of occurrence. Method: This paper discusses the scan test which has been applied to retrospective data to detect carbon monoxide poisoning clusters. We propose a more sensitive method, based on the binomial distribution to detect current clusters of only 1 or more days duration during the ongoing data collection and monitoring process. Results: Applied to daily carbon monoxide poisoning incidence data on which the scan test has been applied, the new test for current clustering evidently has much more power. For retrospective identification of previous clusters lasting more than 1 day, the scan test is recommended. However, for previous daily clusters, a third method is recommended. Conclusion: Certain toxic events such as carbon monoxide poisoning occasionally occur in daily clusters, or in clusters lasting a few days. Timely detection of clusters requires application of early intervention strategies fostered by sensitive statistical methods of detection, as presented here.