Signal Validation

In light of recent communicable disease outbreaks, the ability of Florida Department of HealthÕs (FDOH) syndromic surveillance system, ESSENCE-FL, to identify emergent disease outbreaks using reportable disease data and algorithms originally designed for emergency department chief complaint data was examined. Preliminary work on this analysis presented last year was recently updated and expanded to include additional diseases, further levels of locale, and detector algorithm comparisons. Cases are entered into Merlin, the Bureau of EpidemiologyÕs secure web-based reporting and epidemiologic analysis system, by all 67 county health departments and the de-identified case data are sent hourly to ESSENCE-FL. These data are then available for ad hoc queries, allowing users to observe unusual changes in disease activity and assist in timely identification of infectious disease outbreaks. Based on system algorithms, weekly case tallies are assigned an increasing intensity awareness status from normal to alert and are monitored by county and state epidemiologists to guide timely disease control efforts, but may not by themselves be definitive actionable information.

Objective

To determine if there is an association between known outbreak activity and ESSENCE generated alerts. 

Calls to NHS Direct (a national UK telephone health advice line) which may be indicative of infection show marked seasonal variation, often peaking during winter or early spring. This variation may be related to the seasonality of common viruses. There is currently no routine microbiological confirmation of the cause of illness in NHS Direct callers. Modelling trends in NHS Direct syndromic call data against laboratory data may help by attributing the likely cause of these calls the and surveillance ‘signals’ generated by syndromic surveillance.

Multiple linear regression has been used previously to estimate the contribution of rotavirus and RSV to hospital admission for infectious intestinal disease and lower respiratory tract infections respectively. We applied a similar regression model to NHS Direct syndromic surveillance data and laboratory reports.

Objective

To provide weekly estimates of the proportions of NHS Direct respiratory calls attributable to common infectious disease pathogens.

Use of robust and broadly applicable statistical alerting methods is essential for a public health Biosurveillance system. We compared several algorithms related to the Early Aberration Reporting System C2 (adaptive control chart) method for practical detection sensitivity and timeliness using a realistic but stochastic signal inject strategy with a variety of data streams. The comparison allowed detail examination of strategies for adjusting daily syndromic counts for day-of-week effects and the total daily volume of facility visits. Adjustment for the total visit volume allows monitoring of surrogate rates instead of just counts, and the use of real data with both syndromic and total visit counts enables this adjustment.

Objective

We compared several aberration detection algorithms using a set of syndromic data streams from a large number of treatment facilities in the CDC Biosense 1.0 system. A realistic signal injection strategy was devised to compare different ways of adjusting for total facility visits and background day-of-week effects.

San Diego County Public Health has been conducting syndromic surveillance for the past few years. Currently, the system has become largely automated and processes and analyzes data from a variety of disparate sources including hospital emergency departments, 911 call centers, prehospital transports, and over-the-counter drug sales. What has remained constant since the system’s initial conceptualization is the local opinion that the data should be analyzed and interpreted in a variety of ways, in anticipation for the variety of contexts in which events that are of public health interest may unfold. Relatively small increases in volume that are sustained over time will likely be detected by methods designed to detect “small process shifts”, and include the CUSUM and EWMA methods. Larger increases in volume that are not sustained over time will likely be detected by other employed methods (P-Chart in the event of a non-proportional increase in volume, U-Chart in the event of a proportional increase in volume). A retrospective analysis was conducted on historical data from various data sources to determine the frequency of signals and detected events as well as the context within which the alert occurred (i.e., the “shape” of the data). Findings regarding several actual public health events will also be discussed.

Objective

This paper describes the frequency, various “shapes” and magnitudes of data anomalies, and varying ways actual public health events may present themselves in syndromic data.

While early event detection systems aim to detect disease outbreaks before traditional means, following up on the many alerts generated by these systems can be time-consuming and a drain on limited resources.

Authorized users at local, regional and state levels in North Carolina rely on the North Carolina Disease Event Tracking and Epidemiologic Collection Tool's (NC DETECT) Java-based Web application to monitor and follow-up on signals based on the CDC’s EARS CUSUM algorithms. The application provides users with access to aggregate syndrome-based reports as well as to patient-specific line listing reports for three data sources: emergency departments, ambulance runs and the statewide poison control center. All NC DETECT Web functionality is developed in a user-centered, iterative process with user feedback guiding enhancements and new development. This feedback, along with the need for improved situational awareness and the desire to improve communication among users drove the development of the Annotation Reports and the Custom Event Report.

Objective

We describe the addition of two reports to NC DETECT designed to improve NC public health situational awareness capability.

One limitation of syndromic surveillance systems based on emergency department (ED) data is the time and expense to investigate peak signals, especially when that involves phone calls or visits to the hospital. Many EDs use electronic medical records (EMRs) which are available remotely in real time. This may facilitate the investigation of peak signals.

Since 2004, the French syndromic surveillance system SurSaUD® coordinated by the French Public Health Agency (Sante publique France) daily collects morbidity data from two data sources: the emergency departments (ED) network Oscour® and the emergency general practitioners associations SOS Medecins. Almost 92% of the French ED attendances are recorded by the system. SOS Medecins network is a group of 62 associations of general practitioners, dispatched all over the territory. Sante publique France received data from 61 out of 62 associations. Both data sources collect medical diagnosis, using ICD10 codes in the ED network and specific medical thesaurus in SOS Medecins associations. These data are routinely analyzed to detect and follow-up various expected or unusual public health events all over the territory. The system is also used for reassurance of decision makers. In that framework, in March 2017, the French Ministry of Health requested Sante publique France to validate a potential scarlet fever outbreak in France.

Objective:

Describe a case study of validation of a scarlet fever outbreak using syndromic surveillance data sources.