A number of syndromic surveillance systems include tools that quickly identify potentially large disease outbreak events. However, the high falsepositive rate continues to be a problem in all of these systems. Our earlier work has showed that multi-source information fusion can improve specificity of the syndromic surveillance systems. However, an anomalous health event that presents as only a few cases may remain undetected because the chief complaint data does not contain enough details. New linked data sources need to be used to enhance detection capabilities. The focus of this project examining the incorporation of laboratory, prescription medications and radiology data linked to the patient encounter within syndromic surveillance systems. These data source linkings may enhance the sensitivity of syndromic surveillance.
Hung Lang
The Veterans Health Administration (VHA) operates over 880 outpatient clinics across the nation. The Johns Hopkins Applied Physics Laboratory’s Electronic Surveillance System for the Early Notification of Community-Based Epidemics (ESSENCE) utilizes VHA ICD9 coded outpatient visit data for the detection of abnormal patterns of disease occurrence. The hemorrhagic illness (HI) syndrome category in ESSENCE is comprised of 25 different ICD9 codes, including 12 codes specific for viral hemorrhagic fever (VHF) (e.g., ebola, yellow fever, CrimeanCongo hemorrhagic fever, lassa, etc.) and 13 nonspecific conditions (e.g., purpura not otherwise specified (NOS), thrombocytopathy, and coagulation defect NOS).
Objective
We sought to evaluate the functionality of the diagnosis codes which fall into the syndrome category of hemorrhagic illness.
We evaluated several classifications of emergency department (ED) syndromic data to ascertain best syndrome classifications for ILI.
This paper describes a simple technique for utilizing linked health information in syndromic surveillance. Using knowledge of which patient encounters resulted in laboratory test requests and prescriptions may improve sensitivity and specificity of detection algorithms.