The following category queries were performed during March–December 2008: 1) Severe or time-sensitive diseases/conditions: anthrax, botulism, smallpox, and meningitis; 2) outbreaks not detected by regular alerts in ESSENCE: diarrhea, vomiting, and food poisoning with spatial-temporal clustering; and 3) other reportable diseases/conditions, consisting of varicella (chickenpox), carbon monoxide poisoning, ciguatera, cryptosporidiosis, cyclosporiasis, dengue fever, encephalitis, hepatitis, malaria, measles, mercury poisoning, mumps, pertussis, salmonellosis, shigellosis, and rabies.

When descriptions of the diseases/conditions were found in the chief complaint field, staff contacted the hospital’s Infection Control Practitioner. If the disease was confirmed, further investigation was performed. Potential outbreaks were detected from clustering by age, gender, race/ethnicity, resident ZIP code, hospital, time of visit, and chief complaint.

A total of 740,320 emergency department visits (mean 2,419 visits/day) were monitored in ESSENCE during the study period; 1,813 (0.25%) of those had information leading to 1 of the queried reportable diseases in the chief complaint (mean 5.9 visits/day). After further investigation, we found 58.0% (1,052/1,813) of these additional queries were relevant after excluding unrelated terms, such as “malabsorption” instead of “malaria” or “chicken bone in the throat” instead of “chicken pox.”

On August 31, 2008, the newly designed query for severe or time-sensitive diseases or conditions detected a group of 5 women who had arrived at the same hospital within a 2-hour period with a chief report of either meningococcemia or exposure to meningococcemia (Table 1). This cluster of potentially epidemiologically linked patients did not initiate an automated alert in a syndrome in ESSENCE, and the hospital did not report it by phone immediately. Therefore, without the query, the health department would not have been aware of these patients in a timely manner. The findings from the query enabled health department staff to give postexposure prophylaxis to 36 persons identified as close contacts.

Gastrointestinal outbreaks are an example of outbreaks not detected by regular alerts in ESSENCE that were often detected through the use of the specialized query. One outbreak identified was among persons staying in the same homeless shelter (Table 2), another was among residents at an assisted living facility, and a third was among a group of persons who visited a restaurant. After the outbreaks were confirmed with the Infection Control Practitioner, recommendations for infection control and prevention were made to each facility.

The most common terms found under other reportable diseases or conditions were meningitis, hepatitis, chicken pox, and postexposure prophylaxis for rabies (Table 2). These 4 conditions accounted for 68.2% (717/1,052) of the queried chief complaints. When we contacted providers with regard to query findings, it was apparent that some providers were not familiar with reporting requirements for chicken pox and animal bites requiring postexposure prophylaxis for potential rabies exposure.

Results suggest ESSENCE can enhance healthcare practitioners’ compliance with reportable disease requirements for individual diseases and potential outbreaks. Results demonstrated how expanded queries can detect potential outbreaks or diseases not found in automated syndrome categories. Without the specialized queries, we would have missed the opportunity to implement proper disease control measures required for these events.

Indiana State Health Department investigators found that certain keywords such as “exposure” and “meningitis” may uncover trends previously undetected and they continue to explore similar data-mining techniques (5). Syndromic surveillance systems use statistical algorithms to alert users when the number of reports for a syndrome exceeds the norm (6). Current spatial–temporal algorithms are used to detect large-scale outbreaks over a certain extended period (7,8). However, this method has not been successful for detecting many small clusters of patients with similar characteristics visiting the emergency department from the same home ZIP code, hospital, or within a short period, such as a few minutes. By contacting the Infection Control Practitioner when reportable disease names are found in the chief complaint field, the health department has developed a stronger relationship with hospitals.

One of the limitations of the study was that even when queries were performed with parsers, there were often misspellings, typographical errors, and abbreviations that can lead to a failure to capture all possible events (9). Because the level of investigation can vary from making a phone call to a participating hospital to dispatching a team to interview patients, the cost of time spent on each disease may need to be weighed before initiating action (10). Replication of this study depends on a health department’s capabilities to contact the hospital for follow-up. Future research will examine the information gathered from this new project, and we expect that better disease reporting compliance will result from this innovative use of syndromic surveillance.