Syndromic Surveillance: Is it Worth the Effort?

Unlike bombings, bioterrorism can be invisible, unapparent until people become ill, spreading silently as infected people interact with others. The flying bombs of September II, 200 I, the subsequent anthrax letters, and credible threats of future attacks have greatly heightened the need for a bioterrorism early warning system. The sooner public health officials know about a bioterrorist attack, the more effective their response can be. Early detection speeds quarantines, vaccinations, and treatments. Since the terrorist attacks on September II, a number of cities and states have worked to develop "syndromic surveillance" systems based on existing data. Syndromic surveillance involves the statistical analyses of data on individuals seeking care in emergency rooms or other health care settings for early symptoms of bioterrorist agents. Focusing on symptoms rather than confirmed diagnoses, syndromic surveillance aims to detect bioevents earlier than would be possible with traditional disease surveillance systems. Because potential bioterrorist agents such as anthrax, plague, brucellosis, tularemia, Q-fever, glanders, smallpox, and viral hemorrhagic fevers initially present a "flu-likeillness," data suggesting a sudden increase of individuals with fever, headache, muscle pain, and malaise could be the first indication of either a bioterrorist attack or a natural disease outbreak. Researchers and public health experts are developing a wide variety of syndromic surveillance systems. These systems involve gathering data, automating statistical analysis of the data, and communicating results among physicians and public health officials. Some systems collect and analyze more than symptom data. They include data such as over-the-counter (OTe) pharmaceutical sales and absenteeism. Syndromic surveillance systems are intended to raise an alarm, which then must be followed up by investigation and prevention. However,all alarm systems face trade-offs arising from imperfect knowledge about the occurrence of an attack. First, there is the common trade-off between true positives: sensitivity, sounding the alarm when an attack occurs, and false positives: one minus the specificity, sounding an alarm when there is no attack. Timeliness is a key factor in the tradeoff. Analyzing data for longer periods of time can help improve the accuracy of our alarm system, but too much time eliminates the value of the alarm. Usually, it is possible to make the false positive rate tolerably small, but lowering the false positive rate generally decreases the true positive rate, slows timeliness, or both. Moreover, it is likely that thousands of syndromic surveillance systems' will be running simultaneously in cities, counties, and states throughout the United States. Each of these jurisdictions might be looking at data in six to eight symptom categories, separately from each hospital in the area, and so on. As an example, suppose everycounty in the United States has an independent statistical