ISDS Conference

Modern surveillance systems use statistical process control (SPC) charts such as Cumulative Sum and Exponentially Weighted Moving Average charts for monitoring daily counts of such quantities as ICD-9 codes from ED visits, sales of medications, and doctors’ office visits. The working assumption is that such pre-clinical data contain an early signature of disease outbreaks, manifested as an increase in the count levels. However, the direct application of SPC charts to the raw counts leads to unreliable performance. A popular statistical solution is to precondition the data before applying the charts by modeling or removing explainable patterns from the data and then monitoring the residuals. Although the general idea is common practice, the specifics of how to identify the existing explainable components and how to account for them are domain-specific. Therefore, we seek to present a set of modeling and data-driven tools that are useful for syndromic data.

Objective

SPC charts are widely used in disease surveillance. The charts are very effective when monitored data meet the requirements of temporal independence, stationarity, and normality. However, when these assumptions are violated, the SPC charts will either fail to detect special cause variations or will alert frequently even in the absence of anomalies. Currently collected biosurveillance data contain predictable factors such as day-of-week effects, seasonal effects, holidays, autocorrelation, and global trends that cause the data to violate these assumptions. This work (1) describes a set of tools for identifying such explainable patterns and (2) examines several data preconditioning methods that account for these factors, yielding data better suited for monitoring by traditional SPC charts.

Modern biosurveillance relies on multiple sources of both prediagnostic and diagnostic data, updated daily, to discover disease outbreaks. Intrinsic to this effort are two assumptions: (1) the data being analyzed contain early indicators of a disease outbreak and (2) the outbreaks to be detected are not known a priori. However, in addition to outbreak indicators, syndromic data streams include such factors as day-of-week effects, seasonal effects, autocorrelation, and global trends. These explainable factors obscure unexplained outbreak events, and their presence in the data violates standard control-chart assumptions. Monitoring tools such as Shewhart, cumulative sum, and exponentially weighted moving average control charts will alert based largely on these explainable factors instead of on outbreaks. The goal of this paper is 2-fold: first, to describe a set of tools for identifying explainable patterns such as temporal dependence and, second, to survey and examine several data preconditioning methods that significantly reduce these explainable factors, yielding data better suited for monitoring using the popular control charts.

In October 2006, the Centers for Disease Control and Prevention funded four institutions, including Emory University, to conduct evaluations of the BioSense surveillance system. These evaluations include investigations of situations that represent actual or potential threats to public health in order to describe: 1) the pathways that health departments follow to assess and respond to such threats, 2) the role of various forms of surveillance, including BioSense and other syndromic surveillance systems, in enabling health departments to achieve critical milestones along these pathways, and 3) whether and how surveillance information informs healthcare practice during these events. We anticipate that these case studies will 1) identify approaches to improving BioSense and other syndromic surveillance systems, 2) describe the characteristics of events where syndromic surveillance is most apt to be useful, and 3) provide a baseline for assessing future impacts of advances in the development of BioSense and other forms of public health surveillance. This paper describes preliminary observations from initial case studies conducted by the Emory University team.

Objective

This paper describes preliminary observations from case study investigations of the uses of BioSense and other surveillance resources in public health practice.

Los Angeles County Department of Health Services is currently testing SaTScan’s space-time permutation model to assist in identifying aberrant illness activity in the community and determine it’s ability to detect outbreaks through analyzing real-time syndromic data. SaTScan could be useful especially due to its ability to provide geographic locations of outbreaks in the community.

Objective

To determine the usefulness of SaTScan as an outbreak and illness cluster detection tool in syndromic surveillance and to compare to a purely temporal CUSUM algorithm.

“The ultimate measure of whether a surveillance system has achieved the optimal balance of attributes lies in its usefulness.” No one is better qualified to comment on usefulness than the users. As system developers, we are well advised to consider the opinions of users when building, evaluating, and considering revisions to surveillance systems. 

Health Monitoring Systems, Inc. is a for-profit company that provides biosurveillance capabilities to public health agencies and hospitals using a software-as-a-service model.

Objective

This paper describes results from a survey of public health department users of biosurveillance. The survey solicited input regarding sophistication of analytic methods, perceived value of potential data sources, and utilization resulting in timelier public health interventions.

Aerial transmission and direct contact are important factors for flu. Consequently, close contact with large groups of people, such as during mass transit, present opportunities for transmission. One protective method that decreases the probabilities of becoming ill is vaccination. The potential health impact of  erminating subway service during a flu epidemic depends on both the potential for transmission and vaccination rates among riders. Mass transit, a major method of transit in NYC, exhibits a non-random distribution of riders based on demographics and socio-economic status. There is also a trend in vaccination rates by demography and socio-economic status. This analysis uses individual-based data on vaccination and ridership to separately predict vaccination and ridership for inclusion in agent-based models that can be used to assess impact of public health interventions.

Objective

Agent-based models (ABMs) have been developed to simulate epidemics including smallpox and pandemic flu. The ABM approach is an effective method to assess the potential impact of interventions on disease spread. Integrating the ABM approach with syndromic surveillance data provides potential benefits such  as permitting a realistic specification of some critical model contact parameters, and permitting synthetic outbreaks to be generated with extremely fine resolution (e.g., age, gender, and address). This would provide the ability to test various clustering detection algorithms – a key component of syndromic surveillance methods. RTI International (the Models of Infectious Disease Agent Study (MIDAS) informatics group) and NYC DOHMH (a premier syndromic surveillance research center) collaborated to create a NYC-ABM of flu transmission. This poster describes implementation of several features required for accurate model specification, including assigning immunization rates and subway ridership. Incorporating subway ridership is of great interest, because a large subway system, like the NYC system, has never been investigated as a contributor of disease spread.

The North Carolina Disease Event Tracking and Epidemiologic Collection Tool (NC DETECT) is the early event detection system that serves public health users across North Carolina. One important data source for this system is North Carolina emergency department visits. ED data from hospitals across the state are downloaded, standardized, aggregated, and updated twice daily.

After hurricane Katrina devastated the Gulf Coast on August 29, 2005, federal officials evacuated two large groups of evacuees into Wake and Mecklenburg counties in North Carolina. In order to identify and monitor the hospital-based public health needs of these and other “unofficial” evacuees, NC state officials used both NC DETECT and hospital-based Public Health Epidemiologist reporting methods, along with other public health surveillance initiatives.

Objective

To compare two different methods of monitoring hurricane Katrina evacuees’ hospital visits in North Carolina.

Shenzhen is a special economic region in southern China, adjacent to Hong Kong, with a population of approximately 14 million. The pioneering efforts of Shenzhen in the development of electronic disease surveillance started as early as in 1995. The setup of syndromic surveillance was started after the SARS outbreak in 2003, including surveillance in Fever Clinics, GI clinics, selected schools, and sentinel surveillance for the workers in selected chicken farms and bird markets. In 2007, a regional plan was developed for systematically integrating the surveillance for environmental health, food safety, lab information systems, infectious disease notification, and outbreak management.

Objective

This paper introduces the challenges and lessons learned from the planning and development of a regional integrated disease surveillance system, presenting a new method to quantitatively measure IT support capabilities in disease surveillance and control, as well as a collaboration model integrating the information from multiple sources.

School absenteeism data could be used as an early indicator for disease outbreaks. The increase in absences, however, may be driven by non-sickness related factors. Reason for absence combined with syndrome-specific information might make absenteeism data more useful for early outbreak detection.

Objective

This is a pilot evaluation to determine the usefulness of syndrome-specific school absenteeism data for public health surveillance systems.

The 2003 heat wave in France (15,000 extra deaths in 10 days) led the French institute for public health surveillance to modify its public health surveillance system. One of the major objectives of this program was a real time surveillance based on emergency departments (EDs). Trials experiments started in 2004 with a daily automatic data collection from 20 hospitals in the Paris area. The objectives of this new system were: 1) to detect early all threats for public health; and 2) to measure the impact of an identified phenomena.

In 2006 France was concerned by a new heat wave. It was the opportunity for recording health data during a hot period through this real time system.

Objective

This paper describes the performances of a syndromic surveillance system based on EDs during a heat wave.