The 131st Annual Meeting (November 15-19, 2003) of APHA

Abstract #53079

Sergio Recuenco, MD MPH¹, Millicent Eidson, MA, DVM², Kathryn J. Schmit², Dennis J. White, MS, PhD², Laura D. Kramer, PhD³, and Barbara J. Wallace, MD². (1) School of Public Health, State University of New York, Albany, One University Place, Rensselaer, Albany, NY 12144, 518 427 0595, sxr03@health.state.ny.us, (2) Bureau of Communicable Disease Control, New York State Department of Health, Corning Tower, Empire State Plaza, Albany, NY 12237, (3) Wadsworth Center, Arbovirus Laboratories, New York State Department of Health, 5668 State Farm Road, Slingerlands, NY 12159

Since the WNV epidemic in NYS started in 1999, research has been conducted on the development of indicators to forecast human risk, some of which are based on bird deaths and infections. This study explored the use of SaTScan software and the spatial scan statistic as tools for detecting clusters of dead crow sightings at a local level (smaller than county).

Reports provided by the surveillance system of dead crow sightings in Long Island were geocoded and prepared for analysis weekly from CDC Week 23 (6/3/01-6/9/01) to Week 38 (9/16/01-9/22/01). The data were analyzed in a case-control design, where the cases were the dead crow sightings for the week under study and the controls were the dead crow sightings during prior periods without evidence of WNV circulation (before the first WNV-positive bird was identified). As output SaTScan identified clusters and the census tracts for each, along with log-likelihood, relative risk, and p values. Statistical significance was set at p=0.10. Maps showing the clusters, crow sighting points, WNV-positive dead birds, and human cases were produced weekly.

Significant clusters were found near the residence of the single human case in Suffolk County for eight weeks before symptom onset. In Nassau County clusters were found intermittently in the general area of the four human cases. Differences in county-level surveillance approaches and difficulties in determining the ideal control period may have influenced the success of this approach in forecasting human cases. Additional investigation is recommended for real-time implementation in future periods.

Learning Objectives:

• 1.Recognize the use of SaTScan as a useful tool for the surveillance of West Nile Virus. 2.Identify risk areas for Human WNV based on dead crows clusters.
• 3.Discuss possible use of the scan statistic method for the surveillance of other infectious diseases.

Keywords: Surveillance, Geocoding

Presenting author's disclosure statement:
I do not have any significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.