The Louisville Metro Health Department currently surveils community patterns of both acute morbidity and mortality in an effort to detect outbreaks of disease that might be associated with covert bioterror attacks or other emerging infectious or noninfectious public health threats.
Hospital Emergency Departments
Five sentinel hospitals
Ambulance dispatch records
Louisville Fire and Rescue
Jefferson County Coroner’s Office
Daily count of cases
These surveillance activities are intended to provide sufficient advanced warning of an event to allow for early intervention that could reduce consequent morbidity and mortality.
Increased concern over the possibility of large-scale bioterrorism since Anthrax attacks of October 2001.
In response, public health community seeks to develop and deploy systems to provide early warning of an event.
Given the absence of a widespread, effective system for the environmental detection of potential bioterrorism agents, the symptomatic presentation of affected people to healthcare providers may be the first detectable indication of a covert attack.
Although largely untested, syndromic surveillance has emerged as perhaps the most promising epidemiological method for detecting the intentional, covert release of a pathogen on a large scale.
Syndromic Surveillance Systems
Use health-related data that precede diagnosis but signal a sufficient probability of an outbreak to warrant further public health response.
These systems assess the number, not of specific diagnoses, but of cases occurring in predefined categories of symptomology referred to as “syndromes.”
The syndromes are intended to represent the majority of presentations associated with bioterrorism agents.
Syndromic Surveillance Systems
Seen as more likely to detect bioterror events than traditional, diagnosis-based disease-reporting systems for a number of reasons:
Practitioner awareness of potential bioterror agents and emerging or re-emerging infectious diseases is limited.
Second, many of these diseases have nonspecific prodromes similar to those of other, more common illnesses.
Consequently, affected people presenting to emergency departments or other healthcare facilities may not be recognized as victims of bioterrorism.
Rely on a diagnosis
Often requires laboratory confirmation.
Some lab tests or cultures require days or even weeks to become positive.
Data often incomplete.
Practitioners often do not report even traditionally reportable diseases.
In the past, the Health Department surveilled the aggregate daily volume of 16 predetermined dispatch categories of Louisville Fire & Rescue ambulance runs, deemed to be potentially indicative of infectious disease. The number of specified ambulance runs occurring on each day was compared to the year-to-date mean of daily run counts. Counts that were greater than two standard deviations above the mean were considered surveillance significant.
Statistical methods historically used to surveil ambulance runs would, by definition, recognize approximately 2.5% of all observations as “high.”
Nine false positives or “false alarms” per year
Also unable to detect gradual increases in the data because the “running average” of runs would be “dragged” upwards along with the slowly increasing volume of runs.
Unable to detect prolonged but sub-threshold increases in the number of runs
Because we are interested only in detecting increases in the mean disease frequency, not decreases, the cusum is not allowed to fall below zero.
If a negative value is obtained, the cusum is set to zero.
This method is designed to indicate false positives only once in a 500-day period.
Syndrome to None Ratio. The number of cases occurring in a given syndrome per 100 cases occurring in the “Other” syndrome. The SNR, rather than the raw number of cases, is used in used in the CUSUM calculations to control for the effect of daily fluctuations in total ED visit volume .